diff options
Diffstat (limited to 'servers')
151 files changed, 20878 insertions, 13870 deletions
diff --git a/servers/audio/audio_driver_dummy.cpp b/servers/audio/audio_driver_dummy.cpp index 12cadb9301..a28dcb1015 100644 --- a/servers/audio/audio_driver_dummy.cpp +++ b/servers/audio/audio_driver_dummy.cpp @@ -39,16 +39,16 @@ Error AudioDriverDummy::init() { exit_thread = false; samples_in = nullptr; - mix_rate = GLOBAL_GET("audio/mix_rate"); + mix_rate = GLOBAL_GET("audio/driver/mix_rate"); speaker_mode = SPEAKER_MODE_STEREO; channels = 2; - int latency = GLOBAL_GET("audio/output_latency"); + int latency = GLOBAL_GET("audio/driver/output_latency"); buffer_frames = closest_power_of_2(latency * mix_rate / 1000); samples_in = memnew_arr(int32_t, buffer_frames * channels); - thread = Thread::create(AudioDriverDummy::thread_func, this); + thread.start(AudioDriverDummy::thread_func, this); return OK; }; @@ -86,31 +86,18 @@ AudioDriver::SpeakerMode AudioDriverDummy::get_speaker_mode() const { }; void AudioDriverDummy::lock() { - if (!thread) { - return; - } mutex.lock(); }; void AudioDriverDummy::unlock() { - if (!thread) { - return; - } mutex.unlock(); }; void AudioDriverDummy::finish() { - if (!thread) { - return; - } - exit_thread = true; - Thread::wait_to_finish(thread); + thread.wait_to_finish(); if (samples_in) { memdelete_arr(samples_in); }; - - memdelete(thread); - thread = nullptr; }; diff --git a/servers/audio/audio_driver_dummy.h b/servers/audio/audio_driver_dummy.h index 617ffb2c79..7d84e7ffc8 100644 --- a/servers/audio/audio_driver_dummy.h +++ b/servers/audio/audio_driver_dummy.h @@ -37,7 +37,7 @@ #include "core/os/thread.h" class AudioDriverDummy : public AudioDriver { - Thread *thread = nullptr; + Thread thread; Mutex mutex; int32_t *samples_in; diff --git a/servers/audio/audio_filter_sw.cpp b/servers/audio/audio_filter_sw.cpp index 580e061496..bcfa4c4c37 100644 --- a/servers/audio/audio_filter_sw.cpp +++ b/servers/audio/audio_filter_sw.cpp @@ -58,7 +58,7 @@ void AudioFilterSW::prepare_coefficients(Coeffs *p_coeffs) { final_cutoff = 1; //don't allow less than this } - double omega = 2.0 * Math_PI * final_cutoff / sampling_rate; + double omega = Math_TAU * final_cutoff / sampling_rate; double sin_v = Math::sin(omega); double cos_v = Math::cos(omega); @@ -132,7 +132,7 @@ void AudioFilterSW::prepare_coefficients(Coeffs *p_coeffs) { double hicutoff = resonance; double centercutoff = (cutoff + resonance) / 2.0; double bandwidth = (Math::log(centercutoff) - Math::log(hicutoff)) / Math::log((double)2); - omega = 2.0 * Math_PI * centercutoff / sampling_rate; + omega = Math_TAU * centercutoff / sampling_rate; alpha = Math::sin(omega) * Math::sinh(Math::log((double)2) / 2 * bandwidth * omega / Math::sin(omega)); a0 = 1 + alpha; @@ -197,7 +197,7 @@ void AudioFilterSW::set_stages(int p_stages) { //adjust for multiple stages /* Fouriertransform kernel to obtain response */ float AudioFilterSW::get_response(float p_freq, Coeffs *p_coeffs) { - float freq = p_freq / sampling_rate * Math_PI * 2.0f; + float freq = p_freq / sampling_rate * Math_TAU; float cx = p_coeffs->b0, cy = 0.0; diff --git a/servers/audio/audio_rb_resampler.cpp b/servers/audio/audio_rb_resampler.cpp index efdcb916ed..3c8a1469cd 100644 --- a/servers/audio/audio_rb_resampler.cpp +++ b/servers/audio/audio_rb_resampler.cpp @@ -131,7 +131,7 @@ bool AudioRBResampler::mix(AudioFrame *p_dest, int p_frames) { src_read = read_space; } - rb_read_pos = (rb_read_pos + src_read) & rb_mask; + rb_read_pos.set((rb_read_pos.get() + src_read) & rb_mask); // Create fadeout effect for the end of stream (note that it can be because of slow writer) if (p_frames - target_todo > 0) { @@ -183,8 +183,8 @@ Error AudioRBResampler::setup(int p_channels, int p_src_mix_rate, int p_target_m src_mix_rate = p_src_mix_rate; target_mix_rate = p_target_mix_rate; offset = 0; - rb_read_pos = 0; - rb_write_pos = 0; + rb_read_pos.set(0); + rb_write_pos.set(0); //avoid maybe strange noises upon load for (unsigned int i = 0; i < (rb_len * channels); i++) { @@ -205,8 +205,8 @@ void AudioRBResampler::clear() { memdelete_arr(read_buf); rb = nullptr; offset = 0; - rb_read_pos = 0; - rb_write_pos = 0; + rb_read_pos.set(0); + rb_write_pos.set(0); read_buf = nullptr; } @@ -214,8 +214,8 @@ AudioRBResampler::AudioRBResampler() { rb = nullptr; offset = 0; read_buf = nullptr; - rb_read_pos = 0; - rb_write_pos = 0; + rb_read_pos.set(0); + rb_write_pos.set(0); rb_bits = 0; rb_len = 0; diff --git a/servers/audio/audio_rb_resampler.h b/servers/audio/audio_rb_resampler.h index 7b74e3a2a1..c0f981704b 100644 --- a/servers/audio/audio_rb_resampler.h +++ b/servers/audio/audio_rb_resampler.h @@ -32,6 +32,7 @@ #define AUDIO_RB_RESAMPLER_H #include "core/os/memory.h" +#include "core/templates/safe_refcount.h" #include "core/typedefs.h" #include "servers/audio_server.h" @@ -44,8 +45,8 @@ struct AudioRBResampler { uint32_t src_mix_rate; uint32_t target_mix_rate; - volatile int rb_read_pos; - volatile int rb_write_pos; + SafeNumeric<int> rb_read_pos; + SafeNumeric<int> rb_write_pos; int32_t offset; //contains the fractional remainder of the resampler enum { @@ -62,8 +63,8 @@ struct AudioRBResampler { public: _FORCE_INLINE_ void flush() { - rb_read_pos = 0; - rb_write_pos = 0; + rb_read_pos.set(0); + rb_write_pos.set(0); offset = 0; } @@ -78,8 +79,8 @@ public: _FORCE_INLINE_ int get_writer_space() const { int space, r, w; - r = rb_read_pos; - w = rb_write_pos; + r = rb_read_pos.get(); + w = rb_write_pos.get(); if (r == w) { space = rb_len - 1; @@ -95,8 +96,8 @@ public: _FORCE_INLINE_ int get_reader_space() const { int space, r, w; - r = rb_read_pos; - w = rb_write_pos; + r = rb_read_pos.get(); + w = rb_write_pos.get(); if (r == w) { space = 0; @@ -110,48 +111,52 @@ public: } _FORCE_INLINE_ bool has_data() const { - return rb && rb_read_pos != rb_write_pos; + return rb && rb_read_pos.get() != rb_write_pos.get(); } _FORCE_INLINE_ float *get_write_buffer() { return read_buf; } _FORCE_INLINE_ void write(uint32_t p_frames) { ERR_FAIL_COND(p_frames >= rb_len); + int wp = rb_write_pos.get(); + switch (channels) { case 1: { for (uint32_t i = 0; i < p_frames; i++) { - rb[rb_write_pos] = read_buf[i]; - rb_write_pos = (rb_write_pos + 1) & rb_mask; + rb[wp] = read_buf[i]; + wp = (wp + 1) & rb_mask; } } break; case 2: { for (uint32_t i = 0; i < p_frames; i++) { - rb[(rb_write_pos << 1) + 0] = read_buf[(i << 1) + 0]; - rb[(rb_write_pos << 1) + 1] = read_buf[(i << 1) + 1]; - rb_write_pos = (rb_write_pos + 1) & rb_mask; + rb[(wp << 1) + 0] = read_buf[(i << 1) + 0]; + rb[(wp << 1) + 1] = read_buf[(i << 1) + 1]; + wp = (wp + 1) & rb_mask; } } break; case 4: { for (uint32_t i = 0; i < p_frames; i++) { - rb[(rb_write_pos << 2) + 0] = read_buf[(i << 2) + 0]; - rb[(rb_write_pos << 2) + 1] = read_buf[(i << 2) + 1]; - rb[(rb_write_pos << 2) + 2] = read_buf[(i << 2) + 2]; - rb[(rb_write_pos << 2) + 3] = read_buf[(i << 2) + 3]; - rb_write_pos = (rb_write_pos + 1) & rb_mask; + rb[(wp << 2) + 0] = read_buf[(i << 2) + 0]; + rb[(wp << 2) + 1] = read_buf[(i << 2) + 1]; + rb[(wp << 2) + 2] = read_buf[(i << 2) + 2]; + rb[(wp << 2) + 3] = read_buf[(i << 2) + 3]; + wp = (wp + 1) & rb_mask; } } break; case 6: { for (uint32_t i = 0; i < p_frames; i++) { - rb[(rb_write_pos * 6) + 0] = read_buf[(i * 6) + 0]; - rb[(rb_write_pos * 6) + 1] = read_buf[(i * 6) + 1]; - rb[(rb_write_pos * 6) + 2] = read_buf[(i * 6) + 2]; - rb[(rb_write_pos * 6) + 3] = read_buf[(i * 6) + 3]; - rb[(rb_write_pos * 6) + 4] = read_buf[(i * 6) + 4]; - rb[(rb_write_pos * 6) + 5] = read_buf[(i * 6) + 5]; - rb_write_pos = (rb_write_pos + 1) & rb_mask; + rb[(wp * 6) + 0] = read_buf[(i * 6) + 0]; + rb[(wp * 6) + 1] = read_buf[(i * 6) + 1]; + rb[(wp * 6) + 2] = read_buf[(i * 6) + 2]; + rb[(wp * 6) + 3] = read_buf[(i * 6) + 3]; + rb[(wp * 6) + 4] = read_buf[(i * 6) + 4]; + rb[(wp * 6) + 5] = read_buf[(i * 6) + 5]; + wp = (wp + 1) & rb_mask; } } break; } + + rb_write_pos.set(wp); } int get_channel_count() const; diff --git a/servers/audio/audio_stream.cpp b/servers/audio/audio_stream.cpp index 91fce5d34e..ae07f999ed 100644 --- a/servers/audio/audio_stream.cpp +++ b/servers/audio/audio_stream.cpp @@ -54,21 +54,21 @@ void AudioStreamPlaybackResampled::mix(AudioFrame *p_buffer, float p_rate_scale, for (int i = 0; i < p_frames; i++) { uint32_t idx = CUBIC_INTERP_HISTORY + uint32_t(mix_offset >> FP_BITS); - //standard cubic interpolation (great quality/performance ratio) - //this used to be moved to a LUT for greater performance, but nowadays CPU speed is generally faster than memory. + // 4 point, 4th order optimal resampling algorithm from: http://yehar.com/blog/wp-content/uploads/2009/08/deip.pdf float mu = (mix_offset & FP_MASK) / float(FP_LEN); AudioFrame y0 = internal_buffer[idx - 3]; AudioFrame y1 = internal_buffer[idx - 2]; AudioFrame y2 = internal_buffer[idx - 1]; AudioFrame y3 = internal_buffer[idx - 0]; - float mu2 = mu * mu; - AudioFrame a0 = y3 - y2 - y0 + y1; - AudioFrame a1 = y0 - y1 - a0; - AudioFrame a2 = y2 - y0; - AudioFrame a3 = y1; - - p_buffer[i] = (a0 * mu * mu2 + a1 * mu2 + a2 * mu + a3); + AudioFrame even1 = y2 + y1, odd1 = y2 - y1; + AudioFrame even2 = y3 + y0, odd2 = y3 - y0; + AudioFrame c0 = even1 * 0.46835497211269561 + even2 * 0.03164502784253309; + AudioFrame c1 = odd1 * 0.56001293337091440 + odd2 * 0.14666238593949288; + AudioFrame c2 = even1 * -0.250038759826233691 + even2 * 0.25003876124297131; + AudioFrame c3 = odd1 * -0.49949850957839148 + odd2 * 0.16649935475113800; + AudioFrame c4 = even1 * 0.00016095224137360 + even2 * -0.00016095810460478; + p_buffer[i] = (((c4 * mu + c3) * mu + c2) * mu + c1) * mu + c0; mix_offset += mix_increment; @@ -184,7 +184,7 @@ void AudioStreamPlaybackMicrophone::start(float p_from_pos) { return; } - if (!GLOBAL_GET("audio/enable_audio_input")) { + if (!GLOBAL_GET("audio/driver/enable_input")) { WARN_PRINT("Need to enable Project settings > Audio > Enable Audio Input option to use capturing."); return; } diff --git a/servers/audio/effects/audio_effect_capture.cpp b/servers/audio/effects/audio_effect_capture.cpp new file mode 100644 index 0000000000..78837c7531 --- /dev/null +++ b/servers/audio/effects/audio_effect_capture.cpp @@ -0,0 +1,138 @@ +/*************************************************************************/ +/* audio_effect_capture.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "audio_effect_capture.h" + +bool AudioEffectCapture::can_get_buffer(int p_frames) const { + return buffer.data_left() >= p_frames; +} + +PackedVector2Array AudioEffectCapture::get_buffer(int p_frames) { + ERR_FAIL_COND_V(!buffer_initialized, PackedVector2Array()); + ERR_FAIL_INDEX_V(p_frames, buffer.size(), PackedVector2Array()); + int data_left = buffer.data_left(); + if (data_left < p_frames || p_frames == 0) { + return PackedVector2Array(); + } + + PackedVector2Array ret; + ret.resize(p_frames); + + Vector<AudioFrame> streaming_data; + streaming_data.resize(p_frames); + buffer.read(streaming_data.ptrw(), p_frames); + for (int32_t i = 0; i < p_frames; i++) { + ret.write[i] = Vector2(streaming_data[i].l, streaming_data[i].r); + } + return ret; +} + +void AudioEffectCapture::clear_buffer() { + const int32_t data_left = buffer.data_left(); + buffer.advance_read(data_left); +} + +void AudioEffectCapture::_bind_methods() { + ClassDB::bind_method(D_METHOD("can_get_buffer", "frames"), &AudioEffectCapture::can_get_buffer); + ClassDB::bind_method(D_METHOD("get_buffer", "frames"), &AudioEffectCapture::get_buffer); + ClassDB::bind_method(D_METHOD("clear_buffer"), &AudioEffectCapture::clear_buffer); + ClassDB::bind_method(D_METHOD("set_buffer_length", "buffer_length_seconds"), &AudioEffectCapture::set_buffer_length); + ClassDB::bind_method(D_METHOD("get_buffer_length"), &AudioEffectCapture::get_buffer_length); + ClassDB::bind_method(D_METHOD("get_frames_available"), &AudioEffectCapture::get_frames_available); + ClassDB::bind_method(D_METHOD("get_discarded_frames"), &AudioEffectCapture::get_discarded_frames); + ClassDB::bind_method(D_METHOD("get_buffer_length_frames"), &AudioEffectCapture::get_buffer_length_frames); + ClassDB::bind_method(D_METHOD("get_pushed_frames"), &AudioEffectCapture::get_pushed_frames); + + ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "buffer_length", PROPERTY_HINT_RANGE, "0.01,10,0.01"), "set_buffer_length", "get_buffer_length"); +} + +Ref<AudioEffectInstance> AudioEffectCapture::instance() { + if (!buffer_initialized) { + float target_buffer_size = AudioServer::get_singleton()->get_mix_rate() * buffer_length_seconds; + ERR_FAIL_COND_V(target_buffer_size <= 0 || target_buffer_size >= (1 << 27), Ref<AudioEffectInstance>()); + buffer.resize(nearest_shift((int)target_buffer_size)); + buffer_initialized = true; + } + + clear_buffer(); + + Ref<AudioEffectCaptureInstance> ins; + ins.instance(); + ins->base = Ref<AudioEffectCapture>(this); + + return ins; +} + +void AudioEffectCapture::set_buffer_length(float p_buffer_length_seconds) { + buffer_length_seconds = p_buffer_length_seconds; +} + +float AudioEffectCapture::get_buffer_length() { + return buffer_length_seconds; +} + +int AudioEffectCapture::get_frames_available() const { + ERR_FAIL_COND_V(!buffer_initialized, 0); + return buffer.data_left(); +} + +int64_t AudioEffectCapture::get_discarded_frames() const { + return discarded_frames.get(); +} + +int AudioEffectCapture::get_buffer_length_frames() const { + ERR_FAIL_COND_V(!buffer_initialized, 0); + return buffer.size(); +} + +int64_t AudioEffectCapture::get_pushed_frames() const { + return pushed_frames.get(); +} + +void AudioEffectCaptureInstance::process(const AudioFrame *p_src_frames, AudioFrame *p_dst_frames, int p_frame_count) { + RingBuffer<AudioFrame> &buffer = base->buffer; + + for (int i = 0; i < p_frame_count; i++) { + p_dst_frames[i] = p_src_frames[i]; + } + + if (buffer.space_left() >= p_frame_count) { + // Add incoming audio frames to the IO ring buffer + int32_t ret = buffer.write(p_src_frames, p_frame_count); + ERR_FAIL_COND_MSG(ret != p_frame_count, "Failed to add data to effect capture ring buffer despite sufficient space."); + base->pushed_frames.add(p_frame_count); + } else { + base->discarded_frames.add(p_frame_count); + } +} + +bool AudioEffectCaptureInstance::process_silence() const { + return true; +} diff --git a/servers/audio/effects/audio_effect_capture.h b/servers/audio/effects/audio_effect_capture.h new file mode 100644 index 0000000000..81d4ed6b0f --- /dev/null +++ b/servers/audio/effects/audio_effect_capture.h @@ -0,0 +1,82 @@ +/*************************************************************************/ +/* audio_effect_capture.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef AUDIO_EFFECT_CAPTURE_H +#define AUDIO_EFFECT_CAPTURE_H + +#include "core/config/engine.h" +#include "core/math/audio_frame.h" +#include "core/object/reference.h" +#include "core/templates/vector.h" +#include "servers/audio/audio_effect.h" +#include "servers/audio_server.h" + +class AudioEffectCapture; + +class AudioEffectCaptureInstance : public AudioEffectInstance { + GDCLASS(AudioEffectCaptureInstance, AudioEffectInstance); + friend class AudioEffectCapture; + Ref<AudioEffectCapture> base; + +public: + virtual void process(const AudioFrame *p_src_frames, AudioFrame *p_dst_frames, int p_frame_count) override; + virtual bool process_silence() const override; +}; + +class AudioEffectCapture : public AudioEffect { + GDCLASS(AudioEffectCapture, AudioEffect) + friend class AudioEffectCaptureInstance; + + RingBuffer<AudioFrame> buffer; + SafeNumeric<uint64_t> discarded_frames; + SafeNumeric<uint64_t> pushed_frames; + float buffer_length_seconds = 0.1f; + bool buffer_initialized = false; + +protected: + static void _bind_methods(); + +public: + virtual Ref<AudioEffectInstance> instance() override; + + void set_buffer_length(float p_buffer_length_seconds); + float get_buffer_length(); + + bool can_get_buffer(int p_frames) const; + PackedVector2Array get_buffer(int p_len); + void clear_buffer(); + + int get_frames_available() const; + int64_t get_discarded_frames() const; + int get_buffer_length_frames() const; + int64_t get_pushed_frames() const; +}; + +#endif // AUDIO_EFFECT_CAPTURE_H diff --git a/servers/audio/effects/audio_effect_chorus.cpp b/servers/audio/effects/audio_effect_chorus.cpp index 1542273a24..eb2268aa2e 100644 --- a/servers/audio/effects/audio_effect_chorus.cpp +++ b/servers/audio/effects/audio_effect_chorus.cpp @@ -84,7 +84,7 @@ void AudioEffectChorusInstance::_process_chunk(const AudioFrame *p_src_frames, A if (v.cutoff == 0) { continue; } - float auxlp = expf(-2.0 * Math_PI * v.cutoff / mix_rate); + float auxlp = expf(-Math_TAU * v.cutoff / mix_rate); float c1 = 1.0 - auxlp; float c2 = auxlp; AudioFrame h = filter_h[vc]; @@ -104,7 +104,7 @@ void AudioEffectChorusInstance::_process_chunk(const AudioFrame *p_src_frames, A float phase = (float)(local_cycles & AudioEffectChorus::CYCLES_MASK) / (float)(1 << AudioEffectChorus::CYCLES_FRAC); - float wave_delay = sinf(phase * 2.0 * Math_PI) * max_depth_frames; + float wave_delay = sinf(phase * Math_TAU) * max_depth_frames; int wave_delay_frames = lrint(floor(wave_delay)); float wave_delay_frac = wave_delay - (float)wave_delay_frames; @@ -309,7 +309,7 @@ void AudioEffectChorus::_bind_methods() { ClassDB::bind_method(D_METHOD("set_dry", "amount"), &AudioEffectChorus::set_dry); ClassDB::bind_method(D_METHOD("get_dry"), &AudioEffectChorus::get_dry); - ADD_PROPERTY(PropertyInfo(Variant::INT, "voice_count", PROPERTY_HINT_RANGE, "1,4,1"), "set_voice_count", "get_voice_count"); + ADD_PROPERTY(PropertyInfo(Variant::INT, "voice_count", PROPERTY_HINT_RANGE, "1,4,1", PROPERTY_USAGE_DEFAULT | PROPERTY_USAGE_UPDATE_ALL_IF_MODIFIED), "set_voice_count", "get_voice_count"); ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "dry", PROPERTY_HINT_RANGE, "0,1,0.01"), "set_dry", "get_dry"); ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "wet", PROPERTY_HINT_RANGE, "0,1,0.01"), "set_wet", "get_wet"); diff --git a/servers/audio/effects/audio_effect_delay.cpp b/servers/audio/effects/audio_effect_delay.cpp index f04ab45ec1..ba50eeb0a3 100644 --- a/servers/audio/effects/audio_effect_delay.cpp +++ b/servers/audio/effects/audio_effect_delay.cpp @@ -75,7 +75,7 @@ void AudioEffectDelayInstance::_process_chunk(const AudioFrame *p_src_frames, Au tap2_vol.r *= CLAMP(1.0 + base->tap_2_pan, 0, 1); // feedback lowpass here - float lpf_c = expf(-2.0 * Math_PI * base->feedback_lowpass / mix_rate); // 0 .. 10khz + float lpf_c = expf(-Math_TAU * base->feedback_lowpass / mix_rate); // 0 .. 10khz float lpf_ic = 1.0 - lpf_c; const AudioFrame *src = p_src_frames; diff --git a/servers/audio/effects/audio_effect_distortion.cpp b/servers/audio/effects/audio_effect_distortion.cpp index 8f713ace22..06d51776a3 100644 --- a/servers/audio/effects/audio_effect_distortion.cpp +++ b/servers/audio/effects/audio_effect_distortion.cpp @@ -36,8 +36,8 @@ void AudioEffectDistortionInstance::process(const AudioFrame *p_src_frames, Audi const float *src = (const float *)p_src_frames; float *dst = (float *)p_dst_frames; - //float lpf_c=expf(-2.0*Math_PI*keep_hf_hz.get()/(mix_rate*(float)OVERSAMPLE)); - float lpf_c = expf(-2.0 * Math_PI * base->keep_hf_hz / (AudioServer::get_singleton()->get_mix_rate())); + //float lpf_c=expf(-Math_TAU*keep_hf_hz.get()/(mix_rate*(float)OVERSAMPLE)); + float lpf_c = expf(-Math_TAU * base->keep_hf_hz / (AudioServer::get_singleton()->get_mix_rate())); float lpf_ic = 1.0 - lpf_c; float drive_f = base->drive; @@ -58,7 +58,8 @@ void AudioEffectDistortionInstance::process(const AudioFrame *p_src_frames, Audi switch (base->mode) { case AudioEffectDistortion::MODE_CLIP: { - a = powf(a, 1.0001 - drive_f); + float a_sign = a < 0 ? -1.0f : 1.0f; + a = powf(abs(a), 1.0001 - drive_f) * a_sign; if (a > 1.0) { a = 1.0; } else if (a < (-1.0)) { diff --git a/servers/audio/effects/audio_effect_phaser.cpp b/servers/audio/effects/audio_effect_phaser.cpp index 5e4e183ccf..9b70f03a19 100644 --- a/servers/audio/effects/audio_effect_phaser.cpp +++ b/servers/audio/effects/audio_effect_phaser.cpp @@ -38,13 +38,13 @@ void AudioEffectPhaserInstance::process(const AudioFrame *p_src_frames, AudioFra float dmin = base->range_min / (sampling_rate / 2.0); float dmax = base->range_max / (sampling_rate / 2.0); - float increment = 2.f * Math_PI * (base->rate / sampling_rate); + float increment = Math_TAU * (base->rate / sampling_rate); for (int i = 0; i < p_frame_count; i++) { phase += increment; - while (phase >= Math_PI * 2.f) { - phase -= Math_PI * 2.f; + while (phase >= Math_TAU) { + phase -= Math_TAU; } float d = dmin + (dmax - dmin) * ((sin(phase) + 1.f) / 2.f); diff --git a/servers/audio/effects/audio_effect_record.cpp b/servers/audio/effects/audio_effect_record.cpp index e8832c92a3..2015ede81f 100644 --- a/servers/audio/effects/audio_effect_record.cpp +++ b/servers/audio/effects/audio_effect_record.cpp @@ -118,7 +118,7 @@ void AudioEffectRecordInstance::init() { #ifdef NO_THREADS AudioServer::get_singleton()->add_update_callback(&AudioEffectRecordInstance::_update, this); #else - io_thread = Thread::create(_thread_callback, this); + io_thread.start(_thread_callback, this); #endif } @@ -126,9 +126,7 @@ void AudioEffectRecordInstance::finish() { #ifdef NO_THREADS AudioServer::get_singleton()->remove_update_callback(&AudioEffectRecordInstance::_update, this); #else - if (thread_active) { - Thread::wait_to_finish(io_thread); - } + io_thread.wait_to_finish(); #endif } diff --git a/servers/audio/effects/audio_effect_record.h b/servers/audio/effects/audio_effect_record.h index 14e646e29d..b97ec43946 100644 --- a/servers/audio/effects/audio_effect_record.h +++ b/servers/audio/effects/audio_effect_record.h @@ -48,7 +48,7 @@ class AudioEffectRecordInstance : public AudioEffectInstance { Ref<AudioEffectRecord> base; bool is_recording; - Thread *io_thread; + Thread io_thread; bool thread_active = false; Vector<AudioFrame> ring_buffer; diff --git a/servers/audio/effects/audio_effect_spectrum_analyzer.cpp b/servers/audio/effects/audio_effect_spectrum_analyzer.cpp index 7f73f2e880..3f7ab74a74 100644 --- a/servers/audio/effects/audio_effect_spectrum_analyzer.cpp +++ b/servers/audio/effects/audio_effect_spectrum_analyzer.cpp @@ -110,10 +110,11 @@ void AudioEffectSpectrumAnalyzerInstance::process(const AudioFrame *p_src_frames while (p_frame_count) { int to_fill = fft_size * 2 - temporal_fft_pos; to_fill = MIN(to_fill, p_frame_count); + const double to_fill_step = Math_TAU / (double)fft_size; float *fftw = temporal_fft.ptrw(); for (int i = 0; i < to_fill; i++) { //left and right buffers - float window = -0.5 * Math::cos(2.0 * Math_PI * (double)temporal_fft_pos / (double)fft_size) + 0.5; + float window = -0.5 * Math::cos(to_fill_step * (double)temporal_fft_pos) + 0.5; fftw[temporal_fft_pos * 2] = window * p_src_frames->l; fftw[temporal_fft_pos * 2 + 1] = 0; fftw[(temporal_fft_pos + fft_size * 2) * 2] = window * p_src_frames->r; diff --git a/servers/audio/effects/eq.cpp b/servers/audio/effects/eq.cpp index 2181411b9e..e0c3eb6d3a 100644 --- a/servers/audio/effects/eq.cpp +++ b/servers/audio/effects/eq.cpp @@ -89,8 +89,8 @@ void EQ::recalculate_band_coefficients() { double frq_l = round(frq / pow(2.0, octave_size / 2.0)); double side_gain2 = POW2(Math_SQRT12); - double th = 2.0 * Math_PI * frq / mix_rate; - double th_l = 2.0 * Math_PI * frq_l / mix_rate; + double th = Math_TAU * frq / mix_rate; + double th_l = Math_TAU * frq_l / mix_rate; double c2a = side_gain2 * POW2(cos(th)) - 2.0 * side_gain2 * cos(th_l) * cos(th) + side_gain2 - POW2(sin(th_l)); diff --git a/servers/audio/effects/reverb.cpp b/servers/audio/effects/reverb.cpp index eb96e21659..7df2f99f67 100644 --- a/servers/audio/effects/reverb.cpp +++ b/servers/audio/effects/reverb.cpp @@ -91,7 +91,7 @@ void Reverb::process(float *p_src, float *p_dst, int p_frames) { } if (params.hpf > 0) { - float hpaux = expf(-2.0 * Math_PI * params.hpf * 6000 / params.mix_rate); + float hpaux = expf(-Math_TAU * params.hpf * 6000 / params.mix_rate); float hp_a1 = (1.0 + hpaux) / 2.0; float hp_a2 = -(1.0 + hpaux) / 2.0; float hp_b1 = hpaux; @@ -293,7 +293,7 @@ void Reverb::update_parameters() { float auxdmp = params.damp / 2.0 + 0.5; //only half the range (0.5 .. 1.0 is enough) auxdmp *= auxdmp; - c.damp = expf(-2.0 * Math_PI * auxdmp * 10000 / params.mix_rate); // 0 .. 10khz + c.damp = expf(-Math_TAU * auxdmp * 10000 / params.mix_rate); // 0 .. 10khz } } diff --git a/servers/audio_server.cpp b/servers/audio_server.cpp index 295059a1ea..138cb6e1f8 100644 --- a/servers/audio_server.cpp +++ b/servers/audio_server.cpp @@ -71,13 +71,20 @@ void AudioDriver::update_mix_time(int p_frames) { } } -double AudioDriver::get_time_since_last_mix() const { - return (OS::get_singleton()->get_ticks_usec() - _last_mix_time) / 1000000.0; +double AudioDriver::get_time_since_last_mix() { + lock(); + uint64_t last_mix_time = _last_mix_time; + unlock(); + return (OS::get_singleton()->get_ticks_usec() - last_mix_time) / 1000000.0; } -double AudioDriver::get_time_to_next_mix() const { - double total = (OS::get_singleton()->get_ticks_usec() - _last_mix_time) / 1000000.0; - double mix_buffer = _last_mix_frames / (double)get_mix_rate(); +double AudioDriver::get_time_to_next_mix() { + lock(); + uint64_t last_mix_time = _last_mix_time; + uint64_t last_mix_frames = _last_mix_frames; + unlock(); + double total = (OS::get_singleton()->get_ticks_usec() - last_mix_time) / 1000000.0; + double mix_buffer = last_mix_frames / (double)get_mix_rate(); return mix_buffer - total; } @@ -181,10 +188,10 @@ int AudioDriverManager::get_driver_count() { } void AudioDriverManager::initialize(int p_driver) { - GLOBAL_DEF_RST("audio/enable_audio_input", false); - GLOBAL_DEF_RST("audio/mix_rate", DEFAULT_MIX_RATE); - GLOBAL_DEF_RST("audio/output_latency", DEFAULT_OUTPUT_LATENCY); - GLOBAL_DEF_RST("audio/output_latency.web", 50); // Safer default output_latency for web. + GLOBAL_DEF_RST("audio/driver/enable_input", false); + GLOBAL_DEF_RST("audio/driver/mix_rate", DEFAULT_MIX_RATE); + GLOBAL_DEF_RST("audio/driver/output_latency", DEFAULT_OUTPUT_LATENCY); + GLOBAL_DEF_RST("audio/driver/output_latency.web", 50); // Safer default output_latency for web. int failed_driver = -1; @@ -394,6 +401,7 @@ void AudioServer::_mix_step() { for (int k = 0; k < bus->channels.size(); k++) { if (!bus->channels[k].active) { + bus->channels.write[k].peak_volume = AudioFrame(AUDIO_MIN_PEAK_DB, AUDIO_MIN_PEAK_DB); continue; } @@ -427,7 +435,7 @@ void AudioServer::_mix_step() { } } - bus->channels.write[k].peak_volume = AudioFrame(Math::linear2db(peak.l + 0.0000000001), Math::linear2db(peak.r + 0.0000000001)); + bus->channels.write[k].peak_volume = AudioFrame(Math::linear2db(peak.l + AUDIO_PEAK_OFFSET), Math::linear2db(peak.r + AUDIO_PEAK_OFFSET)); if (!bus->channels[k].used) { //see if any audio is contained, because channel was not used @@ -931,9 +939,9 @@ void AudioServer::init_channels_and_buffers() { } void AudioServer::init() { - channel_disable_threshold_db = GLOBAL_DEF_RST("audio/channel_disable_threshold_db", -60.0); - channel_disable_frames = float(GLOBAL_DEF_RST("audio/channel_disable_time", 2.0)) * get_mix_rate(); - ProjectSettings::get_singleton()->set_custom_property_info("audio/channel_disable_time", PropertyInfo(Variant::FLOAT, "audio/channel_disable_time", PROPERTY_HINT_RANGE, "0,5,0.01,or_greater")); + channel_disable_threshold_db = GLOBAL_DEF_RST("audio/buses/channel_disable_threshold_db", -60.0); + channel_disable_frames = float(GLOBAL_DEF_RST("audio/buses/channel_disable_time", 2.0)) * get_mix_rate(); + ProjectSettings::get_singleton()->set_custom_property_info("audio/buses/channel_disable_time", PropertyInfo(Variant::FLOAT, "audio/buses/channel_disable_time", PROPERTY_HINT_RANGE, "0,5,0.01,or_greater")); buffer_size = 1024; //hardcoded for now init_channels_and_buffers(); @@ -950,7 +958,7 @@ void AudioServer::init() { set_edited(false); //avoid editors from thinking this was edited #endif - GLOBAL_DEF_RST("audio/video_delay_compensation_ms", 0); + GLOBAL_DEF_RST("audio/video/video_delay_compensation_ms", 0); } void AudioServer::update() { @@ -1027,7 +1035,7 @@ void AudioServer::update() { } void AudioServer::load_default_bus_layout() { - String layout_path = ProjectSettings::get_singleton()->get("audio/default_bus_layout"); + String layout_path = ProjectSettings::get_singleton()->get("audio/buses/default_bus_layout"); if (ResourceLoader::exists(layout_path)) { Ref<AudioBusLayout> default_layout = ResourceLoader::load(layout_path); diff --git a/servers/audio_server.h b/servers/audio_server.h index 631fef0f65..a1a373e1ca 100644 --- a/servers/audio_server.h +++ b/servers/audio_server.h @@ -70,8 +70,8 @@ protected: #endif public: - double get_time_since_last_mix() const; //useful for video -> audio sync - double get_time_to_next_mix() const; + double get_time_since_last_mix(); //useful for video -> audio sync + double get_time_to_next_mix(); enum SpeakerMode { SPEAKER_MODE_STEREO, @@ -199,7 +199,7 @@ private: last_mix_with_audio = 0; used = false; active = false; - peak_volume = AudioFrame(0, 0); + peak_volume = AudioFrame(AUDIO_MIN_PEAK_DB, AUDIO_MIN_PEAK_DB); } }; diff --git a/servers/camera/camera_feed.cpp b/servers/camera/camera_feed.cpp index be812cf62d..eab4c61591 100644 --- a/servers/camera/camera_feed.cpp +++ b/servers/camera/camera_feed.cpp @@ -184,9 +184,10 @@ CameraFeed::~CameraFeed() { #endif } -void CameraFeed::set_RGB_img(Ref<Image> p_rgb_img) { +void CameraFeed::set_RGB_img(const Ref<Image> &p_rgb_img) { // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 + ERR_FAIL_COND(p_rgb_img.is_null()); if (active) { RenderingServer *vs = RenderingServer::get_singleton(); @@ -207,9 +208,10 @@ void CameraFeed::set_RGB_img(Ref<Image> p_rgb_img) { #endif } -void CameraFeed::set_YCbCr_img(Ref<Image> p_ycbcr_img) { +void CameraFeed::set_YCbCr_img(const Ref<Image> &p_ycbcr_img) { // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 + ERR_FAIL_COND(p_ycbcr_img.is_null()); if (active) { RenderingServer *vs = RenderingServer::get_singleton(); @@ -230,9 +232,11 @@ void CameraFeed::set_YCbCr_img(Ref<Image> p_ycbcr_img) { #endif } -void CameraFeed::set_YCbCr_imgs(Ref<Image> p_y_img, Ref<Image> p_cbcr_img) { +void CameraFeed::set_YCbCr_imgs(const Ref<Image> &p_y_img, const Ref<Image> &p_cbcr_img) { // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 + ERR_FAIL_COND(p_y_img.is_null()); + ERR_FAIL_COND(p_cbcr_img.is_null()); if (active) { RenderingServer *vs = RenderingServer::get_singleton(); diff --git a/servers/camera/camera_feed.h b/servers/camera/camera_feed.h index fc02af4249..eb4ef155bc 100644 --- a/servers/camera/camera_feed.h +++ b/servers/camera/camera_feed.h @@ -100,9 +100,9 @@ public: virtual ~CameraFeed(); FeedDataType get_datatype() const; - void set_RGB_img(Ref<Image> p_rgb_img); - void set_YCbCr_img(Ref<Image> p_ycbcr_img); - void set_YCbCr_imgs(Ref<Image> p_y_img, Ref<Image> p_cbcr_img); + void set_RGB_img(const Ref<Image> &p_rgb_img); + void set_YCbCr_img(const Ref<Image> &p_ycbcr_img); + void set_YCbCr_imgs(const Ref<Image> &p_y_img, const Ref<Image> &p_cbcr_img); // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 void allocate_texture(int p_width, int p_height, Image::Format p_format, RenderingServer::TextureType p_texture_type, FeedDataType p_data_type); diff --git a/servers/camera_server.cpp b/servers/camera_server.cpp index b06f32417c..ee4a2e148b 100644 --- a/servers/camera_server.cpp +++ b/servers/camera_server.cpp @@ -99,6 +99,8 @@ Ref<CameraFeed> CameraServer::get_feed_by_id(int p_id) { }; void CameraServer::add_feed(const Ref<CameraFeed> &p_feed) { + ERR_FAIL_COND(p_feed.is_null()); + // add our feed feeds.push_back(p_feed); diff --git a/servers/display_server.cpp b/servers/display_server.cpp index 29c1c9fc60..2fa333cc05 100644 --- a/servers/display_server.cpp +++ b/servers/display_server.cpp @@ -477,7 +477,7 @@ void DisplayServer::_bind_methods() { ClassDB::bind_method(D_METHOD("enable_for_stealing_focus", "process_id"), &DisplayServer::enable_for_stealing_focus); - ClassDB::bind_method(D_METHOD("native_video_play", "path", "volume", "audio_track", "subtitle_track"), &DisplayServer::native_video_play); + ClassDB::bind_method(D_METHOD("native_video_play", "path", "volume", "audio_track", "subtitle_track", "screen"), &DisplayServer::native_video_play); ClassDB::bind_method(D_METHOD("native_video_is_playing"), &DisplayServer::native_video_is_playing); ClassDB::bind_method(D_METHOD("native_video_stop"), &DisplayServer::native_video_stop); ClassDB::bind_method(D_METHOD("native_video_pause"), &DisplayServer::native_video_pause); @@ -504,6 +504,11 @@ void DisplayServer::_bind_methods() { ClassDB::bind_method(D_METHOD("set_native_icon", "filename"), &DisplayServer::set_native_icon); ClassDB::bind_method(D_METHOD("set_icon", "image"), &DisplayServer::set_icon); + ClassDB::bind_method(D_METHOD("tablet_get_driver_count"), &DisplayServer::tablet_get_driver_count); + ClassDB::bind_method(D_METHOD("tablet_get_driver_name", "idx"), &DisplayServer::tablet_get_driver_name); + ClassDB::bind_method(D_METHOD("tablet_get_current_driver"), &DisplayServer::tablet_get_current_driver); + ClassDB::bind_method(D_METHOD("tablet_set_current_driver", "name"), &DisplayServer::tablet_set_current_driver); + BIND_ENUM_CONSTANT(FEATURE_GLOBAL_MENU); BIND_ENUM_CONSTANT(FEATURE_SUBWINDOWS); BIND_ENUM_CONSTANT(FEATURE_TOUCHSCREEN); diff --git a/servers/display_server.h b/servers/display_server.h index fc34a2a228..3aab572120 100644 --- a/servers/display_server.h +++ b/servers/display_server.h @@ -83,7 +83,7 @@ protected: static DisplayServerCreate server_create_functions[MAX_SERVERS]; static int server_create_count; - friend class RenderingServerDefault; + friend class RendererViewport; virtual void _set_use_vsync(bool p_enable); public: @@ -340,6 +340,11 @@ public: virtual String keyboard_get_layout_language(int p_index) const; virtual String keyboard_get_layout_name(int p_index) const; + virtual int tablet_get_driver_count() const { return 1; }; + virtual String tablet_get_driver_name(int p_driver) const { return "default"; }; + virtual String tablet_get_current_driver() const { return "default"; }; + virtual void tablet_set_current_driver(const String &p_driver){}; + virtual void process_events() = 0; virtual void force_process_and_drop_events(); diff --git a/servers/navigation_server_2d.cpp b/servers/navigation_server_2d.cpp index df348d2add..9e32bc209b 100644 --- a/servers/navigation_server_2d.cpp +++ b/servers/navigation_server_2d.cpp @@ -80,6 +80,18 @@ NavigationServer2D *NavigationServer2D::singleton = nullptr; return NavigationServer3D::get_singleton()->FUNC_NAME(CONV_0(D_0), CONV_1(D_1), CONV_2(D_2), CONV_3(D_3)); \ } +#define FORWARD_5_R_C(CONV_R, FUNC_NAME, T_0, D_0, T_1, D_1, T_2, D_2, T_3, D_3, T_4, D_4, CONV_0, CONV_1, CONV_2, CONV_3, CONV_4) \ + NavigationServer2D::FUNC_NAME(T_0 D_0, T_1 D_1, T_2 D_2, T_3 D_3, T_4 D_4) \ + const { \ + return CONV_R(NavigationServer3D::get_singleton()->FUNC_NAME(CONV_0(D_0), CONV_1(D_1), CONV_2(D_2), CONV_3(D_3), CONV_4(D_4))); \ + } + +#define FORWARD_5_C(FUNC_NAME, T_0, D_0, T_1, D_1, T_2, D_2, T_3, D_3, T_4, D_4, CONV_0, CONV_1, CONV_2, CONV_3, CONV_4) \ + NavigationServer2D::FUNC_NAME(T_0 D_0, T_1 D_1, T_2 D_2, T_3 D_3, T_4 D_4) \ + const { \ + return NavigationServer3D::get_singleton()->FUNC_NAME(CONV_0(D_0), CONV_1(D_1), CONV_2(D_2), CONV_3(D_3), CONV_4(D_4)); \ + } + static RID rid_to_rid(const RID d) { return d; } @@ -92,6 +104,10 @@ static int int_to_int(const int d) { return d; } +static uint32_t uint32_to_uint32(const uint32_t d) { + return d; +} + static real_t real_to_real(const real_t d) { return d; } @@ -116,7 +132,8 @@ static Vector<Vector2> vector_v3_to_v2(const Vector<Vector3> &d) { static Transform trf2_to_trf3(const Transform2D &d) { Vector3 o(v2_to_v3(d.get_origin())); Basis b; - b.rotate(Vector3(0, 1, 0), d.get_rotation()); + b.rotate(Vector3(0, -1, 0), d.get_rotation()); + b.scale(v2_to_v3(d.get_scale())); return Transform(b, o); } @@ -140,6 +157,10 @@ static Ref<NavigationMesh> poly_to_mesh(Ref<NavigationPolygon> d) { } } +void NavigationServer2D::_emit_map_changed(RID p_map) { + emit_signal("map_changed", p_map); +} + void NavigationServer2D::_bind_methods() { ClassDB::bind_method(D_METHOD("map_create"), &NavigationServer2D::map_create); ClassDB::bind_method(D_METHOD("map_set_active", "map", "active"), &NavigationServer2D::map_set_active); @@ -148,14 +169,19 @@ void NavigationServer2D::_bind_methods() { ClassDB::bind_method(D_METHOD("map_get_cell_size", "map"), &NavigationServer2D::map_get_cell_size); ClassDB::bind_method(D_METHOD("map_set_edge_connection_margin", "map", "margin"), &NavigationServer2D::map_set_edge_connection_margin); ClassDB::bind_method(D_METHOD("map_get_edge_connection_margin", "map"), &NavigationServer2D::map_get_edge_connection_margin); - ClassDB::bind_method(D_METHOD("map_get_path", "map", "origin", "destination", "optimize"), &NavigationServer2D::map_get_path); + ClassDB::bind_method(D_METHOD("map_get_path", "map", "origin", "destination", "optimize", "layers"), &NavigationServer2D::map_get_path, DEFVAL(1)); ClassDB::bind_method(D_METHOD("map_get_closest_point", "map", "to_point"), &NavigationServer2D::map_get_closest_point); ClassDB::bind_method(D_METHOD("map_get_closest_point_owner", "map", "to_point"), &NavigationServer2D::map_get_closest_point_owner); ClassDB::bind_method(D_METHOD("region_create"), &NavigationServer2D::region_create); ClassDB::bind_method(D_METHOD("region_set_map", "region", "map"), &NavigationServer2D::region_set_map); + ClassDB::bind_method(D_METHOD("region_set_layers", "region", "layers"), &NavigationServer2D::region_set_layers); + ClassDB::bind_method(D_METHOD("region_get_layers", "region"), &NavigationServer2D::region_get_layers); ClassDB::bind_method(D_METHOD("region_set_transform", "region", "transform"), &NavigationServer2D::region_set_transform); ClassDB::bind_method(D_METHOD("region_set_navpoly", "region", "nav_poly"), &NavigationServer2D::region_set_navpoly); + ClassDB::bind_method(D_METHOD("region_get_connections_count", "region"), &NavigationServer2D::region_get_connections_count); + ClassDB::bind_method(D_METHOD("region_get_connection_pathway_start", "region", "connection"), &NavigationServer2D::region_get_connection_pathway_start); + ClassDB::bind_method(D_METHOD("region_get_connection_pathway_end", "region", "connection"), &NavigationServer2D::region_get_connection_pathway_end); ClassDB::bind_method(D_METHOD("agent_create"), &NavigationServer2D::agent_create); ClassDB::bind_method(D_METHOD("agent_set_map", "agent", "map"), &NavigationServer2D::agent_set_map); @@ -171,10 +197,14 @@ void NavigationServer2D::_bind_methods() { ClassDB::bind_method(D_METHOD("agent_set_callback", "agent", "receiver", "method", "userdata"), &NavigationServer2D::agent_set_callback, DEFVAL(Variant())); ClassDB::bind_method(D_METHOD("free", "object"), &NavigationServer2D::free); + + ADD_SIGNAL(MethodInfo("map_changed", PropertyInfo(Variant::RID, "map"))); } NavigationServer2D::NavigationServer2D() { singleton = this; + ERR_FAIL_COND_MSG(!NavigationServer3D::get_singleton(), "The Navigation3D singleton should be initialized before the 2D one."); + NavigationServer3D::get_singleton()->connect("map_changed", callable_mp(this, &NavigationServer2D::_emit_map_changed)); } NavigationServer2D::~NavigationServer2D() { @@ -193,20 +223,25 @@ real_t FORWARD_1_C(map_get_cell_size, RID, p_map, rid_to_rid); void FORWARD_2_C(map_set_edge_connection_margin, RID, p_map, real_t, p_connection_margin, rid_to_rid, real_to_real); real_t FORWARD_1_C(map_get_edge_connection_margin, RID, p_map, rid_to_rid); -Vector<Vector2> FORWARD_4_R_C(vector_v3_to_v2, map_get_path, RID, p_map, Vector2, p_origin, Vector2, p_destination, bool, p_optimize, rid_to_rid, v2_to_v3, v2_to_v3, bool_to_bool); +Vector<Vector2> FORWARD_5_R_C(vector_v3_to_v2, map_get_path, RID, p_map, Vector2, p_origin, Vector2, p_destination, bool, p_optimize, uint32_t, p_layers, rid_to_rid, v2_to_v3, v2_to_v3, bool_to_bool, uint32_to_uint32); Vector2 FORWARD_2_R_C(v3_to_v2, map_get_closest_point, RID, p_map, const Vector2 &, p_point, rid_to_rid, v2_to_v3); RID FORWARD_2_C(map_get_closest_point_owner, RID, p_map, const Vector2 &, p_point, rid_to_rid, v2_to_v3); RID FORWARD_0_C(region_create); void FORWARD_2_C(region_set_map, RID, p_region, RID, p_map, rid_to_rid, rid_to_rid); - +void FORWARD_2_C(region_set_layers, RID, p_region, uint32_t, p_layers, rid_to_rid, uint32_to_uint32); +uint32_t FORWARD_1_C(region_get_layers, RID, p_region, rid_to_rid); void FORWARD_2_C(region_set_transform, RID, p_region, Transform2D, p_transform, rid_to_rid, trf2_to_trf3); void NavigationServer2D::region_set_navpoly(RID p_region, Ref<NavigationPolygon> p_nav_mesh) const { NavigationServer3D::get_singleton()->region_set_navmesh(p_region, poly_to_mesh(p_nav_mesh)); } +int FORWARD_1_C(region_get_connections_count, RID, p_region, rid_to_rid); +Vector2 FORWARD_2_R_C(v3_to_v2, region_get_connection_pathway_start, RID, p_region, int, p_connection_id, rid_to_rid, int_to_int); +Vector2 FORWARD_2_R_C(v3_to_v2, region_get_connection_pathway_end, RID, p_region, int, p_connection_id, rid_to_rid, int_to_int); + RID NavigationServer2D::agent_create() const { RID agent = NavigationServer3D::get_singleton()->agent_create(); NavigationServer3D::get_singleton()->agent_set_ignore_y(agent, true); diff --git a/servers/navigation_server_2d.h b/servers/navigation_server_2d.h index 7be5a74cb3..d56c719839 100644 --- a/servers/navigation_server_2d.h +++ b/servers/navigation_server_2d.h @@ -45,12 +45,14 @@ class NavigationServer2D : public Object { static NavigationServer2D *singleton; + void _emit_map_changed(RID p_map); + protected: static void _bind_methods(); public: /// Thread safe, can be used across many threads. - static const NavigationServer2D *get_singleton() { return singleton; } + static NavigationServer2D *get_singleton() { return singleton; } /// MUST be used in single thread! static NavigationServer2D *get_singleton_mut() { return singleton; } @@ -77,7 +79,7 @@ public: virtual real_t map_get_edge_connection_margin(RID p_map) const; /// Returns the navigation path to reach the destination from the origin. - virtual Vector<Vector2> map_get_path(RID p_map, Vector2 p_origin, Vector2 p_destination, bool p_optimize) const; + virtual Vector<Vector2> map_get_path(RID p_map, Vector2 p_origin, Vector2 p_destination, bool p_optimize, uint32_t p_layers = 1) const; virtual Vector2 map_get_closest_point(RID p_map, const Vector2 &p_point) const; virtual RID map_get_closest_point_owner(RID p_map, const Vector2 &p_point) const; @@ -88,12 +90,21 @@ public: /// Set the map of this region. virtual void region_set_map(RID p_region, RID p_map) const; + /// Set the region's layers + virtual void region_set_layers(RID p_region, uint32_t p_layers) const; + virtual uint32_t region_get_layers(RID p_region) const; + /// Set the global transformation of this region. virtual void region_set_transform(RID p_region, Transform2D p_transform) const; /// Set the navigation poly of this region. virtual void region_set_navpoly(RID p_region, Ref<NavigationPolygon> p_nav_mesh) const; + /// Get a list of a region's connection to other regions. + virtual int region_get_connections_count(RID p_region) const; + virtual Vector2 region_get_connection_pathway_start(RID p_region, int p_connection_id) const; + virtual Vector2 region_get_connection_pathway_end(RID p_region, int p_connection_id) const; + /// Creates the agent. virtual RID agent_create() const; diff --git a/servers/navigation_server_3d.cpp b/servers/navigation_server_3d.cpp index 0e5ae82b0d..b0047a250a 100644 --- a/servers/navigation_server_3d.cpp +++ b/servers/navigation_server_3d.cpp @@ -46,7 +46,7 @@ void NavigationServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("map_get_cell_size", "map"), &NavigationServer3D::map_get_cell_size); ClassDB::bind_method(D_METHOD("map_set_edge_connection_margin", "map", "margin"), &NavigationServer3D::map_set_edge_connection_margin); ClassDB::bind_method(D_METHOD("map_get_edge_connection_margin", "map"), &NavigationServer3D::map_get_edge_connection_margin); - ClassDB::bind_method(D_METHOD("map_get_path", "map", "origin", "destination", "optimize"), &NavigationServer3D::map_get_path); + ClassDB::bind_method(D_METHOD("map_get_path", "map", "origin", "destination", "optimize", "layers"), &NavigationServer3D::map_get_path, DEFVAL(1)); ClassDB::bind_method(D_METHOD("map_get_closest_point_to_segment", "map", "start", "end", "use_collision"), &NavigationServer3D::map_get_closest_point_to_segment, DEFVAL(false)); ClassDB::bind_method(D_METHOD("map_get_closest_point", "map", "to_point"), &NavigationServer3D::map_get_closest_point); ClassDB::bind_method(D_METHOD("map_get_closest_point_normal", "map", "to_point"), &NavigationServer3D::map_get_closest_point_normal); @@ -54,9 +54,14 @@ void NavigationServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("region_create"), &NavigationServer3D::region_create); ClassDB::bind_method(D_METHOD("region_set_map", "region", "map"), &NavigationServer3D::region_set_map); + ClassDB::bind_method(D_METHOD("region_set_layers", "region", "layers"), &NavigationServer3D::region_set_layers); + ClassDB::bind_method(D_METHOD("region_get_layers", "region"), &NavigationServer3D::region_get_layers); ClassDB::bind_method(D_METHOD("region_set_transform", "region", "transform"), &NavigationServer3D::region_set_transform); ClassDB::bind_method(D_METHOD("region_set_navmesh", "region", "nav_mesh"), &NavigationServer3D::region_set_navmesh); ClassDB::bind_method(D_METHOD("region_bake_navmesh", "mesh", "node"), &NavigationServer3D::region_bake_navmesh); + ClassDB::bind_method(D_METHOD("region_get_connections_count", "region"), &NavigationServer3D::region_get_connections_count); + ClassDB::bind_method(D_METHOD("region_get_connection_pathway_start", "region", "connection"), &NavigationServer3D::region_get_connection_pathway_start); + ClassDB::bind_method(D_METHOD("region_get_connection_pathway_end", "region", "connection"), &NavigationServer3D::region_get_connection_pathway_end); ClassDB::bind_method(D_METHOD("agent_create"), &NavigationServer3D::agent_create); ClassDB::bind_method(D_METHOD("agent_set_map", "agent", "map"), &NavigationServer3D::agent_set_map); @@ -75,9 +80,11 @@ void NavigationServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("set_active", "active"), &NavigationServer3D::set_active); ClassDB::bind_method(D_METHOD("process", "delta_time"), &NavigationServer3D::process); + + ADD_SIGNAL(MethodInfo("map_changed", PropertyInfo(Variant::RID, "map"))); } -const NavigationServer3D *NavigationServer3D::get_singleton() { +NavigationServer3D *NavigationServer3D::get_singleton() { return singleton; } diff --git a/servers/navigation_server_3d.h b/servers/navigation_server_3d.h index 3761c3871a..420f9c9c18 100644 --- a/servers/navigation_server_3d.h +++ b/servers/navigation_server_3d.h @@ -55,7 +55,7 @@ protected: public: /// Thread safe, can be used across many threads. - static const NavigationServer3D *get_singleton(); + static NavigationServer3D *get_singleton(); /// MUST be used in single thread! static NavigationServer3D *get_singleton_mut(); @@ -88,7 +88,7 @@ public: virtual real_t map_get_edge_connection_margin(RID p_map) const = 0; /// Returns the navigation path to reach the destination from the origin. - virtual Vector<Vector3> map_get_path(RID p_map, Vector3 p_origin, Vector3 p_destination, bool p_optimize) const = 0; + virtual Vector<Vector3> map_get_path(RID p_map, Vector3 p_origin, Vector3 p_destination, bool p_optimize, uint32_t p_navigable_layers = 1) const = 0; virtual Vector3 map_get_closest_point_to_segment(RID p_map, const Vector3 &p_from, const Vector3 &p_to, const bool p_use_collision = false) const = 0; virtual Vector3 map_get_closest_point(RID p_map, const Vector3 &p_point) const = 0; @@ -101,15 +101,24 @@ public: /// Set the map of this region. virtual void region_set_map(RID p_region, RID p_map) const = 0; + /// Set the region's layers + virtual void region_set_layers(RID p_region, uint32_t p_layers) const = 0; + virtual uint32_t region_get_layers(RID p_region) const = 0; + /// Set the global transformation of this region. virtual void region_set_transform(RID p_region, Transform p_transform) const = 0; /// Set the navigation mesh of this region. virtual void region_set_navmesh(RID p_region, Ref<NavigationMesh> p_nav_mesh) const = 0; - /// Bake the navigation mesh + /// Bake the navigation mesh. virtual void region_bake_navmesh(Ref<NavigationMesh> r_mesh, Node *p_node) const = 0; + /// Get a list of a region's connection to other regions. + virtual int region_get_connections_count(RID p_region) const = 0; + virtual Vector3 region_get_connection_pathway_start(RID p_region, int p_connection_id) const = 0; + virtual Vector3 region_get_connection_pathway_end(RID p_region, int p_connection_id) const = 0; + /// Creates the agent. virtual RID agent_create() const = 0; diff --git a/servers/physics_2d/area_2d_sw.cpp b/servers/physics_2d/area_2d_sw.cpp index 6485c8d1e9..532cb259b3 100644 --- a/servers/physics_2d/area_2d_sw.cpp +++ b/servers/physics_2d/area_2d_sw.cpp @@ -215,7 +215,9 @@ void Area2DSW::call_queries() { for (Map<BodyKey, BodyState>::Element *E = monitored_bodies.front(); E;) { if (E->get().state == 0) { // Nothing happened - E = E->next(); + Map<BodyKey, BodyState>::Element *next = E->next(); + monitored_bodies.erase(E); + E = next; continue; } @@ -250,7 +252,9 @@ void Area2DSW::call_queries() { for (Map<BodyKey, BodyState>::Element *E = monitored_areas.front(); E;) { if (E->get().state == 0) { // Nothing happened - E = E->next(); + Map<BodyKey, BodyState>::Element *next = E->next(); + monitored_areas.erase(E); + E = next; continue; } diff --git a/servers/physics_2d/body_pair_2d_sw.cpp b/servers/physics_2d/body_pair_2d_sw.cpp index 6a13453f9f..da70ac7d4b 100644 --- a/servers/physics_2d/body_pair_2d_sw.cpp +++ b/servers/physics_2d/body_pair_2d_sw.cpp @@ -221,11 +221,21 @@ real_t combine_friction(Body2DSW *A, Body2DSW *B) { bool BodyPair2DSW::setup(real_t p_step) { //cannot collide - if (!A->test_collision_mask(B) || A->has_exception(B->get_self()) || B->has_exception(A->get_self()) || (A->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC && B->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC && A->get_max_contacts_reported() == 0 && B->get_max_contacts_reported() == 0)) { + if (!A->test_collision_mask(B) || A->has_exception(B->get_self()) || B->has_exception(A->get_self())) { collided = false; return false; } + bool report_contacts_only = false; + if ((A->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC) && (B->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC)) { + if ((A->get_max_contacts_reported() > 0) || (B->get_max_contacts_reported() > 0)) { + report_contacts_only = true; + } else { + collided = false; + return false; + } + } + if (A->is_shape_set_as_disabled(shape_A) || B->is_shape_set_as_disabled(shape_B)) { collided = false; return false; @@ -288,21 +298,17 @@ bool BodyPair2DSW::setup(real_t p_step) { if (A->is_shape_set_as_one_way_collision(shape_A)) { Vector2 direction = xform_A.get_axis(1).normalized(); bool valid = false; - if (B->get_linear_velocity().dot(direction) >= 0) { - for (int i = 0; i < contact_count; i++) { - Contact &c = contacts[i]; - if (!c.reused) { - continue; - } - if (c.normal.dot(direction) > 0) { //greater (normal inverted) - continue; - } - - valid = true; - break; + for (int i = 0; i < contact_count; i++) { + Contact &c = contacts[i]; + if (!c.reused) { + continue; + } + if (c.normal.dot(direction) > -CMP_EPSILON) { //greater (normal inverted) + continue; } + valid = true; + break; } - if (!valid) { collided = false; oneway_disabled = true; @@ -313,19 +319,16 @@ bool BodyPair2DSW::setup(real_t p_step) { if (B->is_shape_set_as_one_way_collision(shape_B)) { Vector2 direction = xform_B.get_axis(1).normalized(); bool valid = false; - if (A->get_linear_velocity().dot(direction) >= 0) { - for (int i = 0; i < contact_count; i++) { - Contact &c = contacts[i]; - if (!c.reused) { - continue; - } - if (c.normal.dot(direction) < 0) { //less (normal ok) - continue; - } - - valid = true; - break; + for (int i = 0; i < contact_count; i++) { + Contact &c = contacts[i]; + if (!c.reused) { + continue; } + if (c.normal.dot(direction) < CMP_EPSILON) { //less (normal ok) + continue; + } + valid = true; + break; } if (!valid) { collided = false; @@ -357,51 +360,44 @@ bool BodyPair2DSW::setup(real_t p_step) { for (int i = 0; i < contact_count; i++) { Contact &c = contacts[i]; + c.active = false; + Vector2 global_A = xform_Au.xform(c.local_A); Vector2 global_B = xform_Bu.xform(c.local_B); real_t depth = c.normal.dot(global_A - global_B); if (depth <= 0 || !c.reused) { - c.active = false; continue; } - c.active = true; #ifdef DEBUG_ENABLED if (space->is_debugging_contacts()) { space->add_debug_contact(global_A + offset_A); space->add_debug_contact(global_B + offset_A); } #endif - int gather_A = A->can_report_contacts(); - int gather_B = B->can_report_contacts(); c.rA = global_A; c.rB = global_B - offset_B; - if (gather_A | gather_B) { - //Vector2 crB( -B->get_angular_velocity() * c.rB.y, B->get_angular_velocity() * c.rB.x ); - - global_A += offset_A; - global_B += offset_A; + if (A->can_report_contacts()) { + Vector2 crB(-B->get_angular_velocity() * c.rB.y, B->get_angular_velocity() * c.rB.x); + A->add_contact(global_A + offset_A, -c.normal, depth, shape_A, global_B + offset_A, shape_B, B->get_instance_id(), B->get_self(), crB + B->get_linear_velocity()); + } - if (gather_A) { - Vector2 crB(-B->get_angular_velocity() * c.rB.y, B->get_angular_velocity() * c.rB.x); - A->add_contact(global_A, -c.normal, depth, shape_A, global_B, shape_B, B->get_instance_id(), B->get_self(), crB + B->get_linear_velocity()); - } - if (gather_B) { - Vector2 crA(-A->get_angular_velocity() * c.rA.y, A->get_angular_velocity() * c.rA.x); - B->add_contact(global_B, c.normal, depth, shape_B, global_A, shape_A, A->get_instance_id(), A->get_self(), crA + A->get_linear_velocity()); - } + if (B->can_report_contacts()) { + Vector2 crA(-A->get_angular_velocity() * c.rA.y, A->get_angular_velocity() * c.rA.x); + B->add_contact(global_B + offset_A, c.normal, depth, shape_B, global_A + offset_A, shape_A, A->get_instance_id(), A->get_self(), crA + A->get_linear_velocity()); } - if ((A->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC && B->get_mode() <= PhysicsServer2D::BODY_MODE_KINEMATIC)) { - c.active = false; + if (report_contacts_only) { collided = false; continue; } + c.active = true; + // Precompute normal mass, tangent mass, and bias. real_t rnA = c.rA.dot(c.normal); real_t rnB = c.rB.dot(c.normal); diff --git a/servers/physics_2d/collision_object_2d_sw.h b/servers/physics_2d/collision_object_2d_sw.h index 2939b4b99f..2db3961f41 100644 --- a/servers/physics_2d/collision_object_2d_sw.h +++ b/servers/physics_2d/collision_object_2d_sw.h @@ -61,7 +61,7 @@ private: Variant metadata; bool disabled; bool one_way_collision; - float one_way_collision_margin; + real_t one_way_collision_margin; Shape() { disabled = false; one_way_collision = false; @@ -153,7 +153,7 @@ public: return shapes[p_idx].disabled; } - _FORCE_INLINE_ void set_shape_as_one_way_collision(int p_idx, bool p_one_way_collision, float p_margin) { + _FORCE_INLINE_ void set_shape_as_one_way_collision(int p_idx, bool p_one_way_collision, real_t p_margin) { CRASH_BAD_INDEX(p_idx, shapes.size()); shapes.write[p_idx].one_way_collision = p_one_way_collision; shapes.write[p_idx].one_way_collision_margin = p_margin; @@ -163,7 +163,7 @@ public: return shapes[p_idx].one_way_collision; } - _FORCE_INLINE_ float get_shape_one_way_collision_margin(int p_idx) const { + _FORCE_INLINE_ real_t get_shape_one_way_collision_margin(int p_idx) const { CRASH_BAD_INDEX(p_idx, shapes.size()); return shapes[p_idx].one_way_collision_margin; } diff --git a/servers/physics_2d/joints_2d_sw.cpp b/servers/physics_2d/joints_2d_sw.cpp index 3558848dac..c7b556deba 100644 --- a/servers/physics_2d/joints_2d_sw.cpp +++ b/servers/physics_2d/joints_2d_sw.cpp @@ -55,6 +55,14 @@ * SOFTWARE. */ +void Joint2DSW::copy_settings_from(Joint2DSW *p_joint) { + set_self(p_joint->get_self()); + set_max_force(p_joint->get_max_force()); + set_bias(p_joint->get_bias()); + set_max_bias(p_joint->get_max_bias()); + disable_collisions_between_bodies(p_joint->is_disabled_collisions_between_bodies()); +} + static inline real_t k_scalar(Body2DSW *a, Body2DSW *b, const Vector2 &rA, const Vector2 &rB, const Vector2 &n) { real_t value = 0; @@ -197,15 +205,6 @@ PinJoint2DSW::PinJoint2DSW(const Vector2 &p_pos, Body2DSW *p_body_a, Body2DSW *p } } -PinJoint2DSW::~PinJoint2DSW() { - if (A) { - A->remove_constraint(this, 0); - } - if (B) { - B->remove_constraint(this, 1); - } -} - ////////////////////////////////////////////// ////////////////////////////////////////////// ////////////////////////////////////////////// @@ -338,11 +337,6 @@ GrooveJoint2DSW::GrooveJoint2DSW(const Vector2 &p_a_groove1, const Vector2 &p_a_ B->add_constraint(this, 1); } -GrooveJoint2DSW::~GrooveJoint2DSW() { - A->remove_constraint(this, 0); - B->remove_constraint(this, 1); -} - ////////////////////////////////////////////// ////////////////////////////////////////////// ////////////////////////////////////////////// @@ -434,8 +428,3 @@ DampedSpringJoint2DSW::DampedSpringJoint2DSW(const Vector2 &p_anchor_a, const Ve A->add_constraint(this, 0); B->add_constraint(this, 1); } - -DampedSpringJoint2DSW::~DampedSpringJoint2DSW() { - A->remove_constraint(this, 0); - B->remove_constraint(this, 1); -} diff --git a/servers/physics_2d/joints_2d_sw.h b/servers/physics_2d/joints_2d_sw.h index 53e436b539..628de972ae 100644 --- a/servers/physics_2d/joints_2d_sw.h +++ b/servers/physics_2d/joints_2d_sw.h @@ -49,12 +49,26 @@ public: _FORCE_INLINE_ void set_max_bias(real_t p_bias) { max_bias = p_bias; } _FORCE_INLINE_ real_t get_max_bias() const { return max_bias; } - virtual PhysicsServer2D::JointType get_type() const = 0; + virtual bool setup(real_t p_step) { return false; } + virtual void solve(real_t p_step) {} + + void copy_settings_from(Joint2DSW *p_joint); + + virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_TYPE_MAX; } Joint2DSW(Body2DSW **p_body_ptr = nullptr, int p_body_count = 0) : Constraint2DSW(p_body_ptr, p_body_count) { bias = 0; max_force = max_bias = 3.40282e+38; }; + + virtual ~Joint2DSW() { + for (int i = 0; i < get_body_count(); i++) { + Body2DSW *body = get_body_ptr()[i]; + if (body) { + body->remove_constraint(this, i); + } + } + }; }; class PinJoint2DSW : public Joint2DSW { @@ -76,7 +90,7 @@ class PinJoint2DSW : public Joint2DSW { real_t softness; public: - virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_PIN; } + virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_TYPE_PIN; } virtual bool setup(real_t p_step); virtual void solve(real_t p_step); @@ -85,7 +99,6 @@ public: real_t get_param(PhysicsServer2D::PinJointParam p_param) const; PinJoint2DSW(const Vector2 &p_pos, Body2DSW *p_body_a, Body2DSW *p_body_b = nullptr); - ~PinJoint2DSW(); }; class GrooveJoint2DSW : public Joint2DSW { @@ -113,13 +126,12 @@ class GrooveJoint2DSW : public Joint2DSW { bool correct; public: - virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_GROOVE; } + virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_TYPE_GROOVE; } virtual bool setup(real_t p_step); virtual void solve(real_t p_step); GrooveJoint2DSW(const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, Body2DSW *p_body_a, Body2DSW *p_body_b); - ~GrooveJoint2DSW(); }; class DampedSpringJoint2DSW : public Joint2DSW { @@ -146,7 +158,7 @@ class DampedSpringJoint2DSW : public Joint2DSW { real_t v_coef; public: - virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_DAMPED_SPRING; } + virtual PhysicsServer2D::JointType get_type() const { return PhysicsServer2D::JOINT_TYPE_DAMPED_SPRING; } virtual bool setup(real_t p_step); virtual void solve(real_t p_step); @@ -155,7 +167,6 @@ public: real_t get_param(PhysicsServer2D::DampedSpringParam p_param) const; DampedSpringJoint2DSW(const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, Body2DSW *p_body_a, Body2DSW *p_body_b); - ~DampedSpringJoint2DSW(); }; #endif // JOINTS_2D_SW_H diff --git a/servers/physics_2d/physics_server_2d_sw.cpp b/servers/physics_2d/physics_server_2d_sw.cpp index 85e24ca537..1040437ca7 100644 --- a/servers/physics_2d/physics_server_2d_sw.cpp +++ b/servers/physics_2d/physics_server_2d_sw.cpp @@ -149,24 +149,19 @@ void PhysicsServer2DSW::_shape_col_cbk(const Vector2 &p_point_A, const Vector2 & return; } + Vector2 rel_dir = (p_point_A - p_point_B); + real_t rel_length2 = rel_dir.length_squared(); if (cbk->valid_dir != Vector2()) { - if (p_point_A.distance_squared_to(p_point_B) > cbk->valid_depth * cbk->valid_depth) { - cbk->invalid_by_dir++; - return; - } - Vector2 rel_dir = (p_point_A - p_point_B).normalized(); - - if (cbk->valid_dir.dot(rel_dir) < Math_SQRT12) { //sqrt(2)/2.0 - 45 degrees - cbk->invalid_by_dir++; - - /* - print_line("A: "+p_point_A); - print_line("B: "+p_point_B); - print_line("discard too angled "+rtos(cbk->valid_dir.dot((p_point_A-p_point_B)))); - print_line("resnorm: "+(p_point_A-p_point_B).normalized()); - print_line("distance: "+rtos(p_point_A.distance_to(p_point_B))); - */ - return; + if (cbk->valid_depth < 10e20) { + if (rel_length2 > cbk->valid_depth * cbk->valid_depth || + (rel_length2 > CMP_EPSILON && cbk->valid_dir.dot(rel_dir.normalized()) < CMP_EPSILON)) { + cbk->invalid_by_dir++; + return; + } + } else { + if (rel_length2 > 0 && cbk->valid_dir.dot(rel_dir.normalized()) < CMP_EPSILON) { + return; + } } } @@ -182,8 +177,7 @@ void PhysicsServer2DSW::_shape_col_cbk(const Vector2 &p_point_A, const Vector2 & } } - real_t d = p_point_A.distance_squared_to(p_point_B); - if (d < min_depth) { + if (rel_length2 < min_depth) { return; } cbk->ptr[min_depth_idx * 2 + 0] = p_point_A; @@ -673,7 +667,7 @@ void PhysicsServer2DSW::body_set_shape_disabled(RID p_body, int p_shape_idx, boo body->set_shape_as_disabled(p_shape_idx, p_disabled); } -void PhysicsServer2DSW::body_set_shape_as_one_way_collision(RID p_body, int p_shape_idx, bool p_enable, float p_margin) { +void PhysicsServer2DSW::body_set_shape_as_one_way_collision(RID p_body, int p_shape_idx, bool p_enable, real_t p_margin) { Body2DSW *body = body_owner.getornull(p_body); ERR_FAIL_COND(!body); ERR_FAIL_INDEX(p_shape_idx, body->get_shape_count()); @@ -964,7 +958,7 @@ bool PhysicsServer2DSW::body_test_motion(RID p_body, const Transform2D &p_from, return body->get_space()->test_body_motion(body, p_from, p_motion, p_infinite_inertia, p_margin, r_result, p_exclude_raycast_shapes); } -int PhysicsServer2DSW::body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin) { +int PhysicsServer2DSW::body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin) { Body2DSW *body = body_owner.getornull(p_body); ERR_FAIL_COND_V(!body, false); ERR_FAIL_COND_V(!body->get_space(), false); @@ -991,6 +985,24 @@ PhysicsDirectBodyState2D *PhysicsServer2DSW::body_get_direct_state(RID p_body) { /* JOINT API */ +RID PhysicsServer2DSW::joint_create() { + Joint2DSW *joint = memnew(Joint2DSW); + RID joint_rid = joint_owner.make_rid(joint); + joint->set_self(joint_rid); + return joint_rid; +} + +void PhysicsServer2DSW::joint_clear(RID p_joint) { + Joint2DSW *joint = joint_owner.getornull(p_joint); + if (joint->get_type() != JOINT_TYPE_MAX) { + Joint2DSW *empty_joint = memnew(Joint2DSW); + empty_joint->copy_settings_from(joint); + + joint_owner.replace(p_joint, empty_joint); + memdelete(joint); + } +} + void PhysicsServer2DSW::joint_set_param(RID p_joint, JointParam p_param, real_t p_value) { Joint2DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); @@ -1054,52 +1066,63 @@ bool PhysicsServer2DSW::joint_is_disabled_collisions_between_bodies(RID p_joint) return joint->is_disabled_collisions_between_bodies(); } -RID PhysicsServer2DSW::pin_joint_create(const Vector2 &p_pos, RID p_body_a, RID p_body_b) { +void PhysicsServer2DSW::joint_make_pin(RID p_joint, const Vector2 &p_pos, RID p_body_a, RID p_body_b) { Body2DSW *A = body_owner.getornull(p_body_a); - ERR_FAIL_COND_V(!A, RID()); + ERR_FAIL_COND(!A); Body2DSW *B = nullptr; if (body_owner.owns(p_body_b)) { B = body_owner.getornull(p_body_b); - ERR_FAIL_COND_V(!B, RID()); + ERR_FAIL_COND(!B); } + Joint2DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); + Joint2DSW *joint = memnew(PinJoint2DSW(p_pos, A, B)); - RID self = joint_owner.make_rid(joint); - joint->set_self(self); - return self; + joint_owner.replace(p_joint, joint); + joint->copy_settings_from(prev_joint); + memdelete(prev_joint); } -RID PhysicsServer2DSW::groove_joint_create(const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) { +void PhysicsServer2DSW::joint_make_groove(RID p_joint, const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) { Body2DSW *A = body_owner.getornull(p_body_a); - ERR_FAIL_COND_V(!A, RID()); + ERR_FAIL_COND(!A); Body2DSW *B = body_owner.getornull(p_body_b); - ERR_FAIL_COND_V(!B, RID()); + ERR_FAIL_COND(!B); + + Joint2DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint2DSW *joint = memnew(GrooveJoint2DSW(p_a_groove1, p_a_groove2, p_b_anchor, A, B)); - RID self = joint_owner.make_rid(joint); - joint->set_self(self); - return self; + + joint_owner.replace(p_joint, joint); + joint->copy_settings_from(prev_joint); + memdelete(prev_joint); } -RID PhysicsServer2DSW::damped_spring_joint_create(const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b) { +void PhysicsServer2DSW::joint_make_damped_spring(RID p_joint, const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b) { Body2DSW *A = body_owner.getornull(p_body_a); - ERR_FAIL_COND_V(!A, RID()); + ERR_FAIL_COND(!A); Body2DSW *B = body_owner.getornull(p_body_b); - ERR_FAIL_COND_V(!B, RID()); + ERR_FAIL_COND(!B); + + Joint2DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint2DSW *joint = memnew(DampedSpringJoint2DSW(p_anchor_a, p_anchor_b, A, B)); - RID self = joint_owner.make_rid(joint); - joint->set_self(self); - return self; + + joint_owner.replace(p_joint, joint); + joint->copy_settings_from(prev_joint); + memdelete(prev_joint); } void PhysicsServer2DSW::pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) { Joint2DSW *j = joint_owner.getornull(p_joint); ERR_FAIL_COND(!j); - ERR_FAIL_COND(j->get_type() != JOINT_PIN); + ERR_FAIL_COND(j->get_type() != JOINT_TYPE_PIN); PinJoint2DSW *pin_joint = static_cast<PinJoint2DSW *>(j); pin_joint->set_param(p_param, p_value); @@ -1108,7 +1131,7 @@ void PhysicsServer2DSW::pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t PhysicsServer2DSW::pin_joint_get_param(RID p_joint, PinJointParam p_param) const { Joint2DSW *j = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!j, 0); - ERR_FAIL_COND_V(j->get_type() != JOINT_PIN, 0); + ERR_FAIL_COND_V(j->get_type() != JOINT_TYPE_PIN, 0); PinJoint2DSW *pin_joint = static_cast<PinJoint2DSW *>(j); return pin_joint->get_param(p_param); @@ -1117,7 +1140,7 @@ real_t PhysicsServer2DSW::pin_joint_get_param(RID p_joint, PinJointParam p_param void PhysicsServer2DSW::damped_spring_joint_set_param(RID p_joint, DampedSpringParam p_param, real_t p_value) { Joint2DSW *j = joint_owner.getornull(p_joint); ERR_FAIL_COND(!j); - ERR_FAIL_COND(j->get_type() != JOINT_DAMPED_SPRING); + ERR_FAIL_COND(j->get_type() != JOINT_TYPE_DAMPED_SPRING); DampedSpringJoint2DSW *dsj = static_cast<DampedSpringJoint2DSW *>(j); dsj->set_param(p_param, p_value); @@ -1126,7 +1149,7 @@ void PhysicsServer2DSW::damped_spring_joint_set_param(RID p_joint, DampedSpringP real_t PhysicsServer2DSW::damped_spring_joint_get_param(RID p_joint, DampedSpringParam p_param) const { Joint2DSW *j = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!j, 0); - ERR_FAIL_COND_V(j->get_type() != JOINT_DAMPED_SPRING, 0); + ERR_FAIL_COND_V(j->get_type() != JOINT_TYPE_DAMPED_SPRING, 0); DampedSpringJoint2DSW *dsj = static_cast<DampedSpringJoint2DSW *>(j); return dsj->get_param(p_param); @@ -1134,7 +1157,7 @@ real_t PhysicsServer2DSW::damped_spring_joint_get_param(RID p_joint, DampedSprin PhysicsServer2D::JointType PhysicsServer2DSW::joint_get_type(RID p_joint) const { Joint2DSW *joint = joint_owner.getornull(p_joint); - ERR_FAIL_COND_V(!joint, JOINT_PIN); + ERR_FAIL_COND_V(!joint, JOINT_TYPE_PIN); return joint->get_type(); } @@ -1331,7 +1354,7 @@ int PhysicsServer2DSW::get_process_info(ProcessInfo p_info) { PhysicsServer2DSW *PhysicsServer2DSW::singletonsw = nullptr; -PhysicsServer2DSW::PhysicsServer2DSW() { +PhysicsServer2DSW::PhysicsServer2DSW(bool p_using_threads) { singletonsw = this; BroadPhase2DSW::create_func = BroadPhase2DHashGrid::_create; //BroadPhase2DSW::create_func=BroadPhase2DBasic::_create; @@ -1340,10 +1363,6 @@ PhysicsServer2DSW::PhysicsServer2DSW() { island_count = 0; active_objects = 0; collision_pairs = 0; -#ifdef NO_THREADS - using_threads = false; -#else - using_threads = int(ProjectSettings::get_singleton()->get("physics/2d/thread_model")) == 2; -#endif + using_threads = p_using_threads; flushing_queries = false; }; diff --git a/servers/physics_2d/physics_server_2d_sw.h b/servers/physics_2d/physics_server_2d_sw.h index 3305c0bd3d..65c5df0fce 100644 --- a/servers/physics_2d/physics_server_2d_sw.h +++ b/servers/physics_2d/physics_server_2d_sw.h @@ -61,11 +61,11 @@ class PhysicsServer2DSW : public PhysicsServer2D { PhysicsDirectBodyState2DSW *direct_state; - mutable RID_PtrOwner<Shape2DSW> shape_owner; - mutable RID_PtrOwner<Space2DSW> space_owner; - mutable RID_PtrOwner<Area2DSW> area_owner; - mutable RID_PtrOwner<Body2DSW> body_owner; - mutable RID_PtrOwner<Joint2DSW> joint_owner; + mutable RID_PtrOwner<Shape2DSW, true> shape_owner; + mutable RID_PtrOwner<Space2DSW, true> space_owner; + mutable RID_PtrOwner<Area2DSW, true> area_owner; + mutable RID_PtrOwner<Body2DSW, true> body_owner; + mutable RID_PtrOwner<Joint2DSW, true> joint_owner; static PhysicsServer2DSW *singletonsw; @@ -191,7 +191,7 @@ public: virtual void body_clear_shapes(RID p_body) override; virtual void body_set_shape_disabled(RID p_body, int p_shape_idx, bool p_disabled) override; - virtual void body_set_shape_as_one_way_collision(RID p_body, int p_shape_idx, bool p_enable, float p_margin) override; + virtual void body_set_shape_as_one_way_collision(RID p_body, int p_shape_idx, bool p_enable, real_t p_margin) override; virtual void body_attach_object_instance_id(RID p_body, ObjectID p_id) override; virtual ObjectID body_get_object_instance_id(RID p_body) const override; @@ -248,22 +248,27 @@ public: virtual void body_set_pickable(RID p_body, bool p_pickable) override; virtual bool body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin = 0.001, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) override; - virtual int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin = 0.001) override; + virtual int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) override; // this function only works on physics process, errors and returns null otherwise virtual PhysicsDirectBodyState2D *body_get_direct_state(RID p_body) override; /* JOINT API */ + virtual RID joint_create() override; + + virtual void joint_clear(RID p_joint) override; + virtual void joint_set_param(RID p_joint, JointParam p_param, real_t p_value) override; virtual real_t joint_get_param(RID p_joint, JointParam p_param) const override; virtual void joint_disable_collisions_between_bodies(RID p_joint, const bool p_disabled) override; virtual bool joint_is_disabled_collisions_between_bodies(RID p_joint) const override; - virtual RID pin_joint_create(const Vector2 &p_pos, RID p_body_a, RID p_body_b = RID()) override; - virtual RID groove_joint_create(const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) override; - virtual RID damped_spring_joint_create(const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b = RID()) override; + virtual void joint_make_pin(RID p_joint, const Vector2 &p_anchor, RID p_body_a, RID p_body_b = RID()) override; + virtual void joint_make_groove(RID p_joint, const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) override; + virtual void joint_make_damped_spring(RID p_joint, const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b = RID()) override; + virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) override; virtual real_t pin_joint_get_param(RID p_joint, PinJointParam p_param) const override; virtual void damped_spring_joint_set_param(RID p_joint, DampedSpringParam p_param, real_t p_value) override; @@ -287,7 +292,7 @@ public: int get_process_info(ProcessInfo p_info) override; - PhysicsServer2DSW(); + PhysicsServer2DSW(bool p_using_threads = false); ~PhysicsServer2DSW() {} }; diff --git a/servers/physics_2d/physics_server_2d_wrap_mt.cpp b/servers/physics_2d/physics_server_2d_wrap_mt.cpp index 15d875b3b7..790c87cc44 100644 --- a/servers/physics_2d/physics_server_2d_wrap_mt.cpp +++ b/servers/physics_2d/physics_server_2d_wrap_mt.cpp @@ -33,7 +33,7 @@ #include "core/os/os.h" void PhysicsServer2DWrapMT::thread_exit() { - exit = true; + exit.set(); } void PhysicsServer2DWrapMT::thread_step(real_t p_delta) { @@ -52,9 +52,9 @@ void PhysicsServer2DWrapMT::thread_loop() { physics_2d_server->init(); - exit = false; - step_thread_up = true; - while (!exit) { + exit.clear(); + step_thread_up.set(); + while (!exit.is_set()) { // flush commands one by one, until exit is requested command_queue.wait_and_flush_one(); } @@ -76,7 +76,7 @@ void PhysicsServer2DWrapMT::step(real_t p_step) { } void PhysicsServer2DWrapMT::sync() { - if (thread) { + if (create_thread) { if (first_frame) { first_frame = false; } else { @@ -97,8 +97,8 @@ void PhysicsServer2DWrapMT::end_sync() { void PhysicsServer2DWrapMT::init() { if (create_thread) { //OS::get_singleton()->release_rendering_thread(); - thread = Thread::create(_thread_callback, this); - while (!step_thread_up) { + thread.start(_thread_callback, this); + while (!step_thread_up.is_set()) { OS::get_singleton()->delay_usec(1000); } } else { @@ -107,37 +107,19 @@ void PhysicsServer2DWrapMT::init() { } void PhysicsServer2DWrapMT::finish() { - if (thread) { + if (thread.is_started()) { command_queue.push(this, &PhysicsServer2DWrapMT::thread_exit); - Thread::wait_to_finish(thread); - memdelete(thread); - - thread = nullptr; + thread.wait_to_finish(); } else { physics_2d_server->finish(); } - - line_shape_free_cached_ids(); - ray_shape_free_cached_ids(); - segment_shape_free_cached_ids(); - circle_shape_free_cached_ids(); - rectangle_shape_free_cached_ids(); - capsule_shape_free_cached_ids(); - convex_polygon_shape_free_cached_ids(); - concave_polygon_shape_free_cached_ids(); - - space_free_cached_ids(); - area_free_cached_ids(); - body_free_cached_ids(); } PhysicsServer2DWrapMT::PhysicsServer2DWrapMT(PhysicsServer2D *p_contained, bool p_create_thread) : command_queue(p_create_thread) { physics_2d_server = p_contained; create_thread = p_create_thread; - thread = nullptr; step_pending = 0; - step_thread_up = false; pool_max_size = GLOBAL_GET("memory/limits/multithreaded_server/rid_pool_prealloc"); diff --git a/servers/physics_2d/physics_server_2d_wrap_mt.h b/servers/physics_2d/physics_server_2d_wrap_mt.h index 9207081a51..3577f706de 100644 --- a/servers/physics_2d/physics_server_2d_wrap_mt.h +++ b/servers/physics_2d/physics_server_2d_wrap_mt.h @@ -34,6 +34,7 @@ #include "core/config/project_settings.h" #include "core/os/thread.h" #include "core/templates/command_queue_mt.h" +#include "core/templates/safe_refcount.h" #include "servers/physics_server_2d.h" #ifdef DEBUG_SYNC @@ -52,9 +53,9 @@ class PhysicsServer2DWrapMT : public PhysicsServer2D { Thread::ID server_thread; Thread::ID main_thread; - volatile bool exit; - Thread *thread; - volatile bool step_thread_up; + SafeFlag exit; + Thread thread; + SafeFlag step_thread_up; bool create_thread; Semaphore step_sem; @@ -73,6 +74,8 @@ public: #define ServerName PhysicsServer2D #define ServerNameWrapMT PhysicsServer2DWrapMT #define server_name physics_2d_server +#define WRITE_ACTION + #include "servers/server_wrap_mt_common.h" //FUNC1RID(shape,ShapeType); todo fix @@ -93,7 +96,7 @@ public: FUNC1RC(real_t, shape_get_custom_solver_bias, RID); //these work well, but should be used from the main thread only - bool shape_collide(RID p_shape_A, const Transform2D &p_xform_A, const Vector2 &p_motion_A, RID p_shape_B, const Transform2D &p_xform_B, const Vector2 &p_motion_B, Vector2 *r_results, int p_result_max, int &r_result_count) { + bool shape_collide(RID p_shape_A, const Transform2D &p_xform_A, const Vector2 &p_motion_A, RID p_shape_B, const Transform2D &p_xform_B, const Vector2 &p_motion_B, Vector2 *r_results, int p_result_max, int &r_result_count) override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); return physics_2d_server->shape_collide(p_shape_A, p_xform_A, p_motion_A, p_shape_B, p_xform_B, p_motion_B, r_results, p_result_max, r_result_count); } @@ -108,18 +111,18 @@ public: FUNC2RC(real_t, space_get_param, RID, SpaceParameter); // this function only works on physics process, errors and returns null otherwise - PhysicsDirectSpaceState2D *space_get_direct_state(RID p_space) { + PhysicsDirectSpaceState2D *space_get_direct_state(RID p_space) override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), nullptr); return physics_2d_server->space_get_direct_state(p_space); } FUNC2(space_set_debug_contacts, RID, int); - virtual Vector<Vector2> space_get_contacts(RID p_space) const { + virtual Vector<Vector2> space_get_contacts(RID p_space) const override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), Vector<Vector2>()); return physics_2d_server->space_get_contacts(p_space); } - virtual int space_get_contact_count(RID p_space) const { + virtual int space_get_contact_count(RID p_space) const override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), 0); return physics_2d_server->space_get_contact_count(p_space); } @@ -189,7 +192,7 @@ public: FUNC2RC(RID, body_get_shape, RID, int); FUNC3(body_set_shape_disabled, RID, int, bool); - FUNC4(body_set_shape_as_one_way_collision, RID, int, bool, float); + FUNC4(body_set_shape_as_one_way_collision, RID, int, bool, real_t); FUNC2(body_remove_shape, RID, int); FUNC1(body_clear_shapes, RID); @@ -244,30 +247,34 @@ public: FUNC4(body_set_force_integration_callback, RID, Object *, const StringName &, const Variant &); - bool body_collide_shape(RID p_body, int p_body_shape, RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, Vector2 *r_results, int p_result_max, int &r_result_count) { + bool body_collide_shape(RID p_body, int p_body_shape, RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, Vector2 *r_results, int p_result_max, int &r_result_count) override { return physics_2d_server->body_collide_shape(p_body, p_body_shape, p_shape, p_shape_xform, p_motion, r_results, p_result_max, r_result_count); } FUNC2(body_set_pickable, RID, bool); - bool body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin = 0.001, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) { + bool body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin = 0.001, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); return physics_2d_server->body_test_motion(p_body, p_from, p_motion, p_infinite_inertia, p_margin, r_result, p_exclude_raycast_shapes); } - int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin = 0.001) { + int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); return physics_2d_server->body_test_ray_separation(p_body, p_transform, p_infinite_inertia, r_recover_motion, r_results, p_result_max, p_margin); } // this function only works on physics process, errors and returns null otherwise - PhysicsDirectBodyState2D *body_get_direct_state(RID p_body) { + PhysicsDirectBodyState2D *body_get_direct_state(RID p_body) override { ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), nullptr); return physics_2d_server->body_get_direct_state(p_body); } /* JOINT API */ + FUNCRID(joint) + + FUNC1(joint_clear, RID) + FUNC3(joint_set_param, RID, JointParam, real_t); FUNC2RC(real_t, joint_get_param, RID, JointParam); @@ -280,9 +287,9 @@ public: //TODO need to convert this to FUNCRID, but it's a hassle.. - FUNC3R(RID, pin_joint_create, const Vector2 &, RID, RID); - FUNC5R(RID, groove_joint_create, const Vector2 &, const Vector2 &, const Vector2 &, RID, RID); - FUNC4R(RID, damped_spring_joint_create, const Vector2 &, const Vector2 &, RID, RID); + FUNC4(joint_make_pin, RID, const Vector2 &, RID, RID); + FUNC6(joint_make_groove, RID, const Vector2 &, const Vector2 &, const Vector2 &, RID, RID); + FUNC5(joint_make_damped_spring, RID, const Vector2 &, const Vector2 &, RID, RID); FUNC3(pin_joint_set_param, RID, PinJointParam, real_t); FUNC2RC(real_t, pin_joint_get_param, RID, PinJointParam); @@ -297,43 +304,28 @@ public: FUNC1(free, RID); FUNC1(set_active, bool); - virtual void init(); - virtual void step(real_t p_step); - virtual void sync(); - virtual void end_sync(); - virtual void flush_queries(); - virtual void finish(); + virtual void init() override; + virtual void step(real_t p_step) override; + virtual void sync() override; + virtual void end_sync() override; + virtual void flush_queries() override; + virtual void finish() override; - virtual bool is_flushing_queries() const { + virtual bool is_flushing_queries() const override { return physics_2d_server->is_flushing_queries(); } - int get_process_info(ProcessInfo p_info) { + int get_process_info(ProcessInfo p_info) override { return physics_2d_server->get_process_info(p_info); } PhysicsServer2DWrapMT(PhysicsServer2D *p_contained, bool p_create_thread); ~PhysicsServer2DWrapMT(); - template <class T> - static PhysicsServer2D *init_server() { -#ifdef NO_THREADS - return memnew(T); // Always single unsafe when no threads are available. -#else - int tm = GLOBAL_DEF("physics/2d/thread_model", 1); - if (tm == 0) { // single unsafe - return memnew(T); - } else if (tm == 1) { // single safe - return memnew(PhysicsServer2DWrapMT(memnew(T), false)); - } else { // multi threaded - return memnew(PhysicsServer2DWrapMT(memnew(T), true)); - } -#endif - } - #undef ServerNameWrapMT #undef ServerName #undef server_name +#undef WRITE_ACTION }; #ifdef DEBUG_SYNC diff --git a/servers/physics_2d/shape_2d_sw.cpp b/servers/physics_2d/shape_2d_sw.cpp index 24c73314d8..6cc086b9b7 100644 --- a/servers/physics_2d/shape_2d_sw.cpp +++ b/servers/physics_2d/shape_2d_sw.cpp @@ -339,10 +339,10 @@ void RectangleShape2DSW::get_supports(const Vector2 &p_normal, Vector2 *r_suppor } bool RectangleShape2DSW::contains_point(const Vector2 &p_point) const { - float x = p_point.x; - float y = p_point.y; - float edge_x = half_extents.x; - float edge_y = half_extents.y; + real_t x = p_point.x; + real_t y = p_point.y; + real_t edge_x = half_extents.x; + real_t edge_y = half_extents.y; return (x >= -edge_x) && (x < edge_x) && (y >= -edge_y) && (y < edge_y); } @@ -502,6 +502,7 @@ Variant CapsuleShape2DSW::get_data() const { void ConvexPolygonShape2DSW::get_supports(const Vector2 &p_normal, Vector2 *r_supports, int &r_amount) const { int support_idx = -1; real_t d = -1e10; + r_amount = 0; for (int i = 0; i < point_count; i++) { //test point @@ -520,7 +521,7 @@ void ConvexPolygonShape2DSW::get_supports(const Vector2 &p_normal, Vector2 *r_su } } - ERR_FAIL_COND(support_idx == -1); + ERR_FAIL_COND_MSG(support_idx == -1, "Convex polygon shape support not found."); r_amount = 1; r_supports[0] = points[support_idx].pos; @@ -580,6 +581,7 @@ bool ConvexPolygonShape2DSW::intersect_segment(const Vector2 &p_begin, const Vec } real_t ConvexPolygonShape2DSW::get_moment_of_inertia(real_t p_mass, const Size2 &p_scale) const { + ERR_FAIL_COND_V_MSG(point_count == 0, 0, "Convex polygon shape has no points."); Rect2 aabb; aabb.position = points[0].pos * p_scale; for (int i = 0; i < point_count; i++) { @@ -590,7 +592,11 @@ real_t ConvexPolygonShape2DSW::get_moment_of_inertia(real_t p_mass, const Size2 } void ConvexPolygonShape2DSW::set_data(const Variant &p_data) { +#ifdef REAL_T_IS_DOUBLE + ERR_FAIL_COND(p_data.get_type() != Variant::PACKED_VECTOR2_ARRAY && p_data.get_type() != Variant::PACKED_FLOAT64_ARRAY); +#else ERR_FAIL_COND(p_data.get_type() != Variant::PACKED_VECTOR2_ARRAY && p_data.get_type() != Variant::PACKED_FLOAT32_ARRAY); +#endif if (points) { memdelete_arr(points); @@ -687,6 +693,10 @@ bool ConcavePolygonShape2DSW::contains_point(const Vector2 &p_point) const { } bool ConcavePolygonShape2DSW::intersect_segment(const Vector2 &p_begin, const Vector2 &p_end, Vector2 &r_point, Vector2 &r_normal) const { + if (segments.size() == 0 || points.size() == 0) { + return false; + } + uint32_t *stack = (uint32_t *)alloca(sizeof(int) * bvh_depth); enum { @@ -829,7 +839,11 @@ int ConcavePolygonShape2DSW::_generate_bvh(BVH *p_bvh, int p_len, int p_depth) { } void ConcavePolygonShape2DSW::set_data(const Variant &p_data) { +#ifdef REAL_T_IS_DOUBLE + ERR_FAIL_COND(p_data.get_type() != Variant::PACKED_VECTOR2_ARRAY && p_data.get_type() != Variant::PACKED_FLOAT64_ARRAY); +#else ERR_FAIL_COND(p_data.get_type() != Variant::PACKED_VECTOR2_ARRAY && p_data.get_type() != Variant::PACKED_FLOAT32_ARRAY); +#endif Rect2 aabb; diff --git a/servers/physics_2d/space_2d_sw.cpp b/servers/physics_2d/space_2d_sw.cpp index c2a6dc828e..4f12248c3e 100644 --- a/servers/physics_2d/space_2d_sw.cpp +++ b/servers/physics_2d/space_2d_sw.cpp @@ -84,6 +84,10 @@ int PhysicsDirectSpaceState2DSW::_intersect_point_impl(const Vector2 &p_point, S int shape_idx = space->intersection_query_subindex_results[i]; + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + Shape2DSW *shape = col_obj->get_shape(shape_idx); Vector2 local_point = (col_obj->get_transform() * col_obj->get_shape_transform(shape_idx)).affine_inverse().xform(p_point); @@ -229,6 +233,10 @@ int PhysicsDirectSpaceState2DSW::intersect_shape(const RID &p_shape, const Trans const CollisionObject2DSW *col_obj = space->intersection_query_results[i]; int shape_idx = space->intersection_query_subindex_results[i]; + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + if (!CollisionSolver2DSW::solve(shape, p_xform, p_motion, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), Vector2(), nullptr, nullptr, nullptr, p_margin)) { continue; } @@ -272,15 +280,19 @@ bool PhysicsDirectSpaceState2DSW::cast_motion(const RID &p_shape, const Transfor const CollisionObject2DSW *col_obj = space->intersection_query_results[i]; int shape_idx = space->intersection_query_subindex_results[i]; + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + Transform2D col_obj_xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx); //test initial overlap, does it collide if going all the way? if (!CollisionSolver2DSW::solve(shape, p_xform, p_motion, col_obj->get_shape(shape_idx), col_obj_xform, Vector2(), nullptr, nullptr, nullptr, p_margin)) { continue; } - //test initial overlap + //test initial overlap, ignore objects it's inside of. if (CollisionSolver2DSW::solve(shape, p_xform, Vector2(), col_obj->get_shape(shape_idx), col_obj_xform, Vector2(), nullptr, nullptr, nullptr, p_margin)) { - return false; + continue; } //just do kinematic solving @@ -346,12 +358,17 @@ bool PhysicsDirectSpaceState2DSW::collide_shape(RID p_shape, const Transform2D & } const CollisionObject2DSW *col_obj = space->intersection_query_results[i]; - int shape_idx = space->intersection_query_subindex_results[i]; if (p_exclude.has(col_obj->get_self())) { continue; } + int shape_idx = space->intersection_query_subindex_results[i]; + + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + cbk.valid_dir = Vector2(); cbk.valid_depth = 0; @@ -383,15 +400,6 @@ struct _RestCallbackData2D { static void _rest_cbk_result(const Vector2 &p_point_A, const Vector2 &p_point_B, void *p_userdata) { _RestCallbackData2D *rd = (_RestCallbackData2D *)p_userdata; - if (rd->valid_dir != Vector2()) { - if (p_point_A.distance_squared_to(p_point_B) > rd->valid_depth * rd->valid_depth) { - return; - } - if (rd->valid_dir.dot((p_point_A - p_point_B).normalized()) < Math_PI * 0.25) { - return; - } - } - Vector2 contact_rel = p_point_B - p_point_A; real_t len = contact_rel.length(); @@ -403,9 +411,21 @@ static void _rest_cbk_result(const Vector2 &p_point_A, const Vector2 &p_point_B, return; } + Vector2 normal = contact_rel / len; + + if (rd->valid_dir != Vector2()) { + if (len > rd->valid_depth) { + return; + } + + if (rd->valid_dir.dot(normal) > -CMP_EPSILON) { + return; + } + } + rd->best_len = len; rd->best_contact = p_point_B; - rd->best_normal = contact_rel / len; + rd->best_normal = normal; rd->best_object = rd->object; rd->best_shape = rd->shape; rd->best_local_shape = rd->local_shape; @@ -433,14 +453,18 @@ bool PhysicsDirectSpaceState2DSW::rest_info(RID p_shape, const Transform2D &p_sh } const CollisionObject2DSW *col_obj = space->intersection_query_results[i]; - int shape_idx = space->intersection_query_subindex_results[i]; if (p_exclude.has(col_obj->get_self())) { continue; } + int shape_idx = space->intersection_query_subindex_results[i]; + + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + rcd.valid_dir = Vector2(); - rcd.valid_depth = 0; rcd.object = col_obj; rcd.shape = shape_idx; rcd.local_shape = 0; @@ -643,9 +667,9 @@ int Space2DSW::test_body_ray_separation(Body2DSW *p_body, const Transform2D &p_t Vector2 a = sr[k * 2 + 0]; Vector2 b = sr[k * 2 + 1]; - recover_motion += (b - a) * 0.4; + recover_motion += (b - a) / cbk.amount; - float depth = a.distance_to(b); + real_t depth = a.distance_to(b); if (depth > result.collision_depth) { result.collision_depth = depth; result.collision_point = b; @@ -740,10 +764,13 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co ExcludedShapeSW excluded_shape_pairs[max_excluded_shape_pairs]; int excluded_shape_pair_count = 0; - float separation_margin = MIN(p_margin, MAX(0.0, p_motion.length() - CMP_EPSILON)); //don't separate by more than the intended motion + real_t motion_length = p_motion.length(); + Vector2 motion_normal = p_motion / motion_length; Transform2D body_transform = p_from; + bool recovered = false; + { //STEP 1, FREE BODY IF STUCK @@ -794,7 +821,7 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co if (col_obj->is_shape_set_as_one_way_collision(shape_idx)) { cbk.valid_dir = col_obj_shape_xform.get_axis(1).normalized(); - float owc_margin = col_obj->get_shape_one_way_collision_margin(shape_idx); + real_t owc_margin = col_obj->get_shape_one_way_collision_margin(shape_idx); cbk.valid_depth = MAX(owc_margin, p_margin); //user specified, but never less than actual margin or it won't work cbk.invalid_by_dir = 0; @@ -805,7 +832,7 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co Vector2 lv = b->get_linear_velocity(); //compute displacement from linear velocity Vector2 motion = lv * PhysicsDirectBodyState2DSW::singleton->step; - float motion_len = motion.length(); + real_t motion_len = motion.length(); motion.normalize(); cbk.valid_depth += motion_len * MAX(motion.dot(-cbk.valid_dir), 0.0); } @@ -820,7 +847,7 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co bool did_collide = false; Shape2DSW *against_shape = col_obj->get_shape(shape_idx); - if (CollisionSolver2DSW::solve(body_shape, body_shape_xform, Vector2(), against_shape, col_obj_shape_xform, Vector2(), cbkres, cbkptr, nullptr, separation_margin)) { + if (CollisionSolver2DSW::solve(body_shape, body_shape_xform, Vector2(), against_shape, col_obj_shape_xform, Vector2(), cbkres, cbkptr, nullptr, p_margin)) { did_collide = cbk.passed > current_passed; //more passed, so collision actually existed } @@ -846,11 +873,20 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co } Vector2 recover_motion; - for (int i = 0; i < cbk.amount; i++) { Vector2 a = sr[i * 2 + 0]; Vector2 b = sr[i * 2 + 1]; - recover_motion += (b - a) * 0.4; + + // Compute plane on b towards a. + Vector2 n = (a - b).normalized(); + real_t d = n.dot(b); + + // Compute depth on recovered motion. + real_t depth = n.dot(a + recover_motion) - d; + if (depth > 0.0) { + // Only recover if there is penetration. + recover_motion -= n * depth * 0.4; + } } if (recover_motion == Vector2()) { @@ -858,6 +894,8 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co break; } + recovered = true; + body_transform.elements[2] += recover_motion; body_aabb.position += recover_motion; @@ -930,7 +968,10 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co //test initial overlap if (CollisionSolver2DSW::solve(body_shape, body_shape_xform, Vector2(), against_shape, col_obj_shape_xform, Vector2(), nullptr, nullptr, nullptr, 0)) { if (col_obj->is_shape_set_as_one_way_collision(col_shape_idx)) { - continue; + Vector2 direction = col_obj_shape_xform.get_axis(1).normalized(); + if (motion_normal.dot(direction) < 0) { + continue; + } } stuck = true; @@ -940,13 +981,12 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co //just do kinematic solving real_t low = 0; real_t hi = 1; - Vector2 mnormal = p_motion.normalized(); for (int k = 0; k < 8; k++) { //steps should be customizable.. real_t ofs = (low + hi) * 0.5; - Vector2 sep = mnormal; //important optimization for this to work fast enough + Vector2 sep = motion_normal; //important optimization for this to work fast enough bool collided = CollisionSolver2DSW::solve(body_shape, body_shape_xform, p_motion * ofs, against_shape, col_obj_shape_xform, Vector2(), nullptr, nullptr, &sep, 0); if (collided) { @@ -967,7 +1007,7 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co cbk.valid_depth = 10e20; - Vector2 sep = mnormal; //important optimization for this to work fast enough + Vector2 sep = motion_normal; //important optimization for this to work fast enough bool collided = CollisionSolver2DSW::solve(body_shape, body_shape_xform, p_motion * (hi + contact_max_allowed_penetration), col_obj->get_shape(col_shape_idx), col_obj_shape_xform, Vector2(), PhysicsServer2DSW::_shape_col_cbk, &cbk, &sep, 0); if (!collided || cbk.amount == 0) { continue; @@ -998,11 +1038,12 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co } bool collided = false; - if (safe >= 1) { - best_shape = -1; //no best shape with cast, reset to -1 - } - { + if (recovered || (safe < 1)) { + if (safe >= 1) { + best_shape = -1; //no best shape with cast, reset to -1 + } + //it collided, let's get the rest info in unsafe advance Transform2D ugt = body_transform; ugt.elements[2] += p_motion * unsafe; @@ -1011,9 +1052,10 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co rcd.best_len = 0; rcd.best_object = nullptr; rcd.best_shape = 0; - rcd.min_allowed_depth = test_motion_min_contact_depth; - //optimization + // Allowed depth can't be lower than motion length, in order to handle contacts at low speed. + rcd.min_allowed_depth = MIN(motion_length, test_motion_min_contact_depth); + int from_shape = best_shape != -1 ? best_shape : 0; int to_shape = best_shape != -1 ? best_shape + 1 : p_body->get_shape_count(); @@ -1061,7 +1103,22 @@ bool Space2DSW::test_body_motion(Body2DSW *p_body, const Transform2D &p_from, co if (col_obj->is_shape_set_as_one_way_collision(shape_idx)) { rcd.valid_dir = col_obj_shape_xform.get_axis(1).normalized(); - rcd.valid_depth = 10e20; + + real_t owc_margin = col_obj->get_shape_one_way_collision_margin(shape_idx); + rcd.valid_depth = MAX(owc_margin, p_margin); //user specified, but never less than actual margin or it won't work + + if (col_obj->get_type() == CollisionObject2DSW::TYPE_BODY) { + const Body2DSW *b = static_cast<const Body2DSW *>(col_obj); + if (b->get_mode() == PhysicsServer2D::BODY_MODE_KINEMATIC || b->get_mode() == PhysicsServer2D::BODY_MODE_RIGID) { + //fix for moving platforms (kinematic and dynamic), margin is increased by how much it moved in the given direction + Vector2 lv = b->get_linear_velocity(); + //compute displacement from linear velocity + Vector2 motion = lv * PhysicsDirectBodyState2DSW::singleton->step; + real_t motion_len = motion.length(); + motion.normalize(); + rcd.valid_depth += motion_len * MAX(motion.dot(-rcd.valid_dir), 0.0); + } + } } else { rcd.valid_dir = Vector2(); rcd.valid_depth = 0; @@ -1331,7 +1388,7 @@ Space2DSW::Space2DSW() { constraint_bias = 0.2; body_linear_velocity_sleep_threshold = GLOBAL_DEF("physics/2d/sleep_threshold_linear", 2.0); - body_angular_velocity_sleep_threshold = GLOBAL_DEF("physics/2d/sleep_threshold_angular", (8.0 / 180.0 * Math_PI)); + body_angular_velocity_sleep_threshold = GLOBAL_DEF("physics/2d/sleep_threshold_angular", Math::deg2rad(8.0)); body_time_to_sleep = GLOBAL_DEF("physics/2d/time_before_sleep", 0.5); ProjectSettings::get_singleton()->set_custom_property_info("physics/2d/time_before_sleep", PropertyInfo(Variant::FLOAT, "physics/2d/time_before_sleep", PROPERTY_HINT_RANGE, "0,5,0.01,or_greater")); diff --git a/servers/physics_3d/area_3d_sw.cpp b/servers/physics_3d/area_3d_sw.cpp index b6c5b3003c..bb4e0ed752 100644 --- a/servers/physics_3d/area_3d_sw.cpp +++ b/servers/physics_3d/area_3d_sw.cpp @@ -215,7 +215,9 @@ void Area3DSW::call_queries() { for (Map<BodyKey, BodyState>::Element *E = monitored_bodies.front(); E;) { if (E->get().state == 0) { // Nothing happened - E = E->next(); + Map<BodyKey, BodyState>::Element *next = E->next(); + monitored_bodies.erase(E); + E = next; continue; } @@ -250,7 +252,9 @@ void Area3DSW::call_queries() { for (Map<BodyKey, BodyState>::Element *E = monitored_areas.front(); E;) { if (E->get().state == 0) { // Nothing happened - E = E->next(); + Map<BodyKey, BodyState>::Element *next = E->next(); + monitored_areas.erase(E); + E = next; continue; } diff --git a/servers/physics_3d/body_3d_sw.cpp b/servers/physics_3d/body_3d_sw.cpp index 82356e77ef..64ba0cb09d 100644 --- a/servers/physics_3d/body_3d_sw.cpp +++ b/servers/physics_3d/body_3d_sw.cpp @@ -51,18 +51,18 @@ void Body3DSW::_update_transform_dependant() { } void Body3DSW::update_inertias() { - //update shapes and motions + // Update shapes and motions. switch (mode) { case PhysicsServer3D::BODY_MODE_RIGID: { - //update tensor for all shapes, not the best way but should be somehow OK. (inspired from bullet) + // Update tensor for all shapes, not the best way but should be somehow OK. (inspired from bullet) real_t total_area = 0; for (int i = 0; i < get_shape_count(); i++) { total_area += get_shape_area(i); } - // We have to recompute the center of mass + // We have to recompute the center of mass. center_of_mass_local.zero(); for (int i = 0; i < get_shape_count(); i++) { @@ -70,21 +70,24 @@ void Body3DSW::update_inertias() { real_t mass = area * this->mass / total_area; - // NOTE: we assume that the shape origin is also its center of mass + // NOTE: we assume that the shape origin is also its center of mass. center_of_mass_local += mass * get_shape_transform(i).origin; } center_of_mass_local /= mass; - // Recompute the inertia tensor + // Recompute the inertia tensor. Basis inertia_tensor; inertia_tensor.set_zero(); + bool inertia_set = false; for (int i = 0; i < get_shape_count(); i++) { if (is_shape_disabled(i)) { continue; } + inertia_set = true; + const Shape3DSW *shape = get_shape(i); real_t area = get_shape_area(i); @@ -102,7 +105,12 @@ void Body3DSW::update_inertias() { inertia_tensor += shape_inertia_tensor + (Basis() * shape_origin.dot(shape_origin) - shape_origin.outer(shape_origin)) * mass; } - // Compute the principal axes of inertia + // Set the inertia to a valid value when there are no valid shapes. + if (!inertia_set) { + inertia_tensor.set_diagonal(Vector3(1.0, 1.0, 1.0)); + } + + // Compute the principal axes of inertia. principal_inertia_axes_local = inertia_tensor.diagonalize().transposed(); _inv_inertia = inertia_tensor.get_main_diagonal().inverse(); @@ -493,20 +501,18 @@ void Body3DSW::integrate_forces(real_t p_step) { if (mode == PhysicsServer3D::BODY_MODE_KINEMATIC) { //compute motion, angular and etc. velocities from prev transform - linear_velocity = (new_transform.origin - get_transform().origin) / p_step; + motion = new_transform.origin - get_transform().origin; + do_motion = true; + linear_velocity = motion / p_step; //compute a FAKE angular velocity, not so easy - Basis rot = new_transform.basis.orthonormalized().transposed() * get_transform().basis.orthonormalized(); + Basis rot = new_transform.basis.orthonormalized() * get_transform().basis.orthonormalized().transposed(); Vector3 axis; real_t angle; rot.get_axis_angle(axis, angle); axis.normalize(); - angular_velocity = axis.normalized() * (angle / p_step); - - motion = new_transform.origin - get_transform().origin; - do_motion = true; - + angular_velocity = axis * (angle / p_step); } else { if (!omit_force_integration && !first_integration) { //overridden by direct state query diff --git a/servers/physics_3d/body_3d_sw.h b/servers/physics_3d/body_3d_sw.h index 41578778f6..e87ff2364b 100644 --- a/servers/physics_3d/body_3d_sw.h +++ b/servers/physics_3d/body_3d_sw.h @@ -290,10 +290,10 @@ public: void update_inertias(); _FORCE_INLINE_ real_t get_inv_mass() const { return _inv_mass; } - _FORCE_INLINE_ Vector3 get_inv_inertia() const { return _inv_inertia; } - _FORCE_INLINE_ Basis get_inv_inertia_tensor() const { return _inv_inertia_tensor; } + _FORCE_INLINE_ const Vector3 &get_inv_inertia() const { return _inv_inertia; } + _FORCE_INLINE_ const Basis &get_inv_inertia_tensor() const { return _inv_inertia_tensor; } _FORCE_INLINE_ real_t get_friction() const { return friction; } - _FORCE_INLINE_ Vector3 get_gravity() const { return gravity; } + _FORCE_INLINE_ const Vector3 &get_gravity() const { return gravity; } _FORCE_INLINE_ real_t get_bounce() const { return bounce; } void set_axis_lock(PhysicsServer3D::BodyAxis p_axis, bool lock); @@ -426,7 +426,7 @@ public: ERR_FAIL_INDEX_V(p_contact_idx, body->contact_count, Vector3()); return body->contacts[p_contact_idx].local_normal; } - virtual float get_contact_impulse(int p_contact_idx) const override { + virtual real_t get_contact_impulse(int p_contact_idx) const override { return 0.0f; // Only implemented for bullet } virtual int get_contact_local_shape(int p_contact_idx) const override { diff --git a/servers/physics_3d/body_pair_3d_sw.cpp b/servers/physics_3d/body_pair_3d_sw.cpp index 6012ff1522..36114c0c91 100644 --- a/servers/physics_3d/body_pair_3d_sw.cpp +++ b/servers/physics_3d/body_pair_3d_sw.cpp @@ -49,12 +49,12 @@ #define MIN_VELOCITY 0.0001 #define MAX_BIAS_ROTATION (Math_PI / 8) -void BodyPair3DSW::_contact_added_callback(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata) { +void BodyPair3DSW::_contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { BodyPair3DSW *pair = (BodyPair3DSW *)p_userdata; - pair->contact_added_callback(p_point_A, p_point_B); + pair->contact_added_callback(p_point_A, p_index_A, p_point_B, p_index_B); } -void BodyPair3DSW::contact_added_callback(const Vector3 &p_point_A, const Vector3 &p_point_B) { +void BodyPair3DSW::contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B) { // check if we already have the contact //Vector3 local_A = A->get_inv_transform().xform(p_point_A); @@ -73,6 +73,8 @@ void BodyPair3DSW::contact_added_callback(const Vector3 &p_point_A, const Vector contact.acc_bias_impulse = 0; contact.acc_bias_impulse_center_of_mass = 0; contact.acc_tangent_impulse = Vector3(); + contact.index_A = p_index_A; + contact.index_B = p_index_B; contact.local_A = local_A; contact.local_B = local_B; contact.normal = (p_point_A - p_point_B).normalized(); @@ -211,11 +213,21 @@ real_t combine_friction(Body3DSW *A, Body3DSW *B) { bool BodyPair3DSW::setup(real_t p_step) { //cannot collide - if (!A->test_collision_mask(B) || A->has_exception(B->get_self()) || B->has_exception(A->get_self()) || (A->get_mode() <= PhysicsServer3D::BODY_MODE_KINEMATIC && B->get_mode() <= PhysicsServer3D::BODY_MODE_KINEMATIC && A->get_max_contacts_reported() == 0 && B->get_max_contacts_reported() == 0)) { + if (!A->test_collision_mask(B) || A->has_exception(B->get_self()) || B->has_exception(A->get_self())) { collided = false; return false; } + bool report_contacts_only = false; + if ((A->get_mode() <= PhysicsServer3D::BODY_MODE_KINEMATIC) && (B->get_mode() <= PhysicsServer3D::BODY_MODE_KINEMATIC)) { + if ((A->get_max_contacts_reported() > 0) || (B->get_max_contacts_reported() > 0)) { + report_contacts_only = true; + } else { + collided = false; + return false; + } + } + if (A->is_shape_set_as_disabled(shape_A) || B->is_shape_set_as_disabled(shape_B)) { collided = false; return false; @@ -279,12 +291,9 @@ bool BodyPair3DSW::setup(real_t p_step) { real_t depth = c.normal.dot(global_A - global_B); if (depth <= 0) { - c.active = false; continue; } - c.active = true; - #ifdef DEBUG_ENABLED if (space->is_debugging_contacts()) { @@ -308,6 +317,11 @@ bool BodyPair3DSW::setup(real_t p_step) { B->add_contact(global_B, c.normal, depth, shape_B, global_A, shape_A, A->get_instance_id(), A->get_self(), crB); } + if (report_contacts_only) { + collided = false; + continue; + } + c.active = true; // Precompute normal mass, tangent mass, and bias. @@ -456,7 +470,7 @@ void BodyPair3DSW::solve(real_t p_step) { } BodyPair3DSW::BodyPair3DSW(Body3DSW *p_A, int p_shape_A, Body3DSW *p_B, int p_shape_B) : - Constraint3DSW(_arr, 2) { + BodyContact3DSW(_arr, 2) { A = p_A; B = p_B; shape_A = p_shape_A; @@ -472,3 +486,305 @@ BodyPair3DSW::~BodyPair3DSW() { A->remove_constraint(this); B->remove_constraint(this); } + +void BodySoftBodyPair3DSW::_contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { + BodySoftBodyPair3DSW *pair = (BodySoftBodyPair3DSW *)p_userdata; + pair->contact_added_callback(p_point_A, p_index_A, p_point_B, p_index_B); +} + +void BodySoftBodyPair3DSW::contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B) { + Vector3 local_A = body->get_inv_transform().xform(p_point_A); + Vector3 local_B = p_point_B - soft_body->get_node_position(p_index_B); + + Contact contact; + contact.index_A = p_index_A; + contact.index_B = p_index_B; + contact.acc_normal_impulse = 0; + contact.acc_bias_impulse = 0; + contact.acc_bias_impulse_center_of_mass = 0; + contact.acc_tangent_impulse = Vector3(); + contact.local_A = local_A; + contact.local_B = local_B; + contact.normal = (p_point_A - p_point_B).normalized(); + contact.mass_normal = 0; + + // Attempt to determine if the contact will be reused. + real_t contact_recycle_radius = space->get_contact_recycle_radius(); + + uint32_t contact_count = contacts.size(); + for (uint32_t contact_index = 0; contact_index < contact_count; ++contact_index) { + Contact &c = contacts[contact_index]; + if (c.index_B == p_index_B) { + if (c.local_A.distance_squared_to(local_A) < (contact_recycle_radius * contact_recycle_radius) && + c.local_B.distance_squared_to(local_B) < (contact_recycle_radius * contact_recycle_radius)) { + contact.acc_normal_impulse = c.acc_normal_impulse; + contact.acc_bias_impulse = c.acc_bias_impulse; + contact.acc_bias_impulse_center_of_mass = c.acc_bias_impulse_center_of_mass; + contact.acc_tangent_impulse = c.acc_tangent_impulse; + } + c = contact; + return; + } + } + + contacts.push_back(contact); +} + +void BodySoftBodyPair3DSW::validate_contacts() { + // Make sure to erase contacts that are no longer valid. + const Transform &transform_A = body->get_transform(); + + real_t contact_max_separation = space->get_contact_max_separation(); + + uint32_t contact_count = contacts.size(); + for (uint32_t contact_index = 0; contact_index < contact_count; ++contact_index) { + Contact &c = contacts[contact_index]; + + Vector3 global_A = transform_A.xform(c.local_A); + Vector3 global_B = soft_body->get_node_position(c.index_B) + c.local_B; + Vector3 axis = global_A - global_B; + real_t depth = axis.dot(c.normal); + + if (depth < -contact_max_separation || (global_B + c.normal * depth - global_A).length() > contact_max_separation) { + // Contact no longer needed, remove. + if ((contact_index + 1) < contact_count) { + // Swap with the last one. + SWAP(c, contacts[contact_count - 1]); + } + + contact_index--; + contact_count--; + } + } + + contacts.resize(contact_count); +} + +bool BodySoftBodyPair3DSW::setup(real_t p_step) { + if (!body->test_collision_mask(soft_body) || body->has_exception(soft_body->get_self()) || soft_body->has_exception(body->get_self())) { + collided = false; + return false; + } + + if (body->is_shape_set_as_disabled(body_shape)) { + collided = false; + return false; + } + + const Transform &xform_Au = body->get_transform(); + Transform xform_A = xform_Au * body->get_shape_transform(body_shape); + + Transform xform_Bu = soft_body->get_transform(); + Transform xform_B = xform_Bu * soft_body->get_shape_transform(0); + + validate_contacts(); + + Shape3DSW *shape_A_ptr = body->get_shape(body_shape); + Shape3DSW *shape_B_ptr = soft_body->get_shape(0); + + bool collided = CollisionSolver3DSW::solve_static(shape_A_ptr, xform_A, shape_B_ptr, xform_B, _contact_added_callback, this, &sep_axis); + this->collided = collided; + + real_t max_penetration = space->get_contact_max_allowed_penetration(); + + real_t bias = (real_t)0.3; + if (shape_A_ptr->get_custom_bias()) { + bias = shape_A_ptr->get_custom_bias(); + } + + real_t inv_dt = 1.0 / p_step; + + uint32_t contact_count = contacts.size(); + for (uint32_t contact_index = 0; contact_index < contact_count; ++contact_index) { + Contact &c = contacts[contact_index]; + c.active = false; + + real_t node_inv_mass = soft_body->get_node_inv_mass(c.index_B); + if (node_inv_mass == 0.0) { + continue; + } + + Vector3 global_A = xform_Au.xform(c.local_A); + Vector3 global_B = soft_body->get_node_position(c.index_B) + c.local_B; + + real_t depth = c.normal.dot(global_A - global_B); + + if (depth <= 0) { + continue; + } + + c.active = true; + +#ifdef DEBUG_ENABLED + + if (space->is_debugging_contacts()) { + space->add_debug_contact(global_A); + space->add_debug_contact(global_B); + } +#endif + + c.rA = global_A - xform_Au.origin - body->get_center_of_mass(); + c.rB = global_B; + + if (body->can_report_contacts()) { + Vector3 crA = body->get_angular_velocity().cross(c.rA) + body->get_linear_velocity(); + body->add_contact(global_A, -c.normal, depth, body_shape, global_B, 0, soft_body->get_instance_id(), soft_body->get_self(), crA); + } + + if (body->get_mode() > PhysicsServer3D::BODY_MODE_KINEMATIC) { + body->set_active(true); + } + + // Precompute normal mass, tangent mass, and bias. + Vector3 inertia_A = body->get_inv_inertia_tensor().xform(c.rA.cross(c.normal)); + real_t kNormal = body->get_inv_mass() + node_inv_mass; + kNormal += c.normal.dot(inertia_A.cross(c.rA)); + c.mass_normal = 1.0f / kNormal; + + c.bias = -bias * inv_dt * MIN(0.0f, -depth + max_penetration); + c.depth = depth; + + Vector3 j_vec = c.normal * c.acc_normal_impulse + c.acc_tangent_impulse; + body->apply_impulse(c.rA + body->get_center_of_mass(), -j_vec); + soft_body->apply_node_impulse(c.index_B, j_vec); + c.acc_bias_impulse = 0; + c.acc_bias_impulse_center_of_mass = 0; + + c.bounce = body->get_bounce(); + + if (c.bounce) { + Vector3 crA = body->get_angular_velocity().cross(c.rA); + Vector3 dv = soft_body->get_node_velocity(c.index_B) - body->get_linear_velocity() - crA; + + // Normal impulse. + c.bounce = c.bounce * dv.dot(c.normal); + } + } + + return true; +} + +void BodySoftBodyPair3DSW::solve(real_t p_step) { + if (!collided) { + return; + } + + uint32_t contact_count = contacts.size(); + for (uint32_t contact_index = 0; contact_index < contact_count; ++contact_index) { + Contact &c = contacts[contact_index]; + if (!c.active) { + continue; + } + + c.active = false; + + // Bias impulse. + Vector3 crbA = body->get_biased_angular_velocity().cross(c.rA); + Vector3 dbv = soft_body->get_node_biased_velocity(c.index_B) - body->get_biased_linear_velocity() - crbA; + + real_t vbn = dbv.dot(c.normal); + + if (Math::abs(-vbn + c.bias) > MIN_VELOCITY) { + real_t jbn = (-vbn + c.bias) * c.mass_normal; + real_t jbnOld = c.acc_bias_impulse; + c.acc_bias_impulse = MAX(jbnOld + jbn, 0.0f); + + Vector3 jb = c.normal * (c.acc_bias_impulse - jbnOld); + + body->apply_bias_impulse(c.rA + body->get_center_of_mass(), -jb, MAX_BIAS_ROTATION / p_step); + soft_body->apply_node_bias_impulse(c.index_B, jb); + + crbA = body->get_biased_angular_velocity().cross(c.rA); + dbv = soft_body->get_node_biased_velocity(c.index_B) - body->get_biased_linear_velocity() - crbA; + + vbn = dbv.dot(c.normal); + + if (Math::abs(-vbn + c.bias) > MIN_VELOCITY) { + real_t jbn_com = (-vbn + c.bias) / (body->get_inv_mass() + soft_body->get_node_inv_mass(c.index_B)); + real_t jbnOld_com = c.acc_bias_impulse_center_of_mass; + c.acc_bias_impulse_center_of_mass = MAX(jbnOld_com + jbn_com, 0.0f); + + Vector3 jb_com = c.normal * (c.acc_bias_impulse_center_of_mass - jbnOld_com); + + body->apply_bias_impulse(body->get_center_of_mass(), -jb_com, 0.0f); + soft_body->apply_node_bias_impulse(c.index_B, -jb_com); + } + + c.active = true; + } + + Vector3 crA = body->get_angular_velocity().cross(c.rA); + Vector3 dv = soft_body->get_node_velocity(c.index_B) - body->get_linear_velocity() - crA; + + // Normal impulse. + real_t vn = dv.dot(c.normal); + + if (Math::abs(vn) > MIN_VELOCITY) { + real_t jn = -(c.bounce + vn) * c.mass_normal; + real_t jnOld = c.acc_normal_impulse; + c.acc_normal_impulse = MAX(jnOld + jn, 0.0f); + + Vector3 j = c.normal * (c.acc_normal_impulse - jnOld); + + body->apply_impulse(c.rA + body->get_center_of_mass(), -j); + soft_body->apply_node_impulse(c.index_B, j); + + c.active = true; + } + + // Friction impulse. + real_t friction = body->get_friction(); + + Vector3 lvA = body->get_linear_velocity() + body->get_angular_velocity().cross(c.rA); + Vector3 lvB = soft_body->get_node_velocity(c.index_B); + Vector3 dtv = lvB - lvA; + + real_t tn = c.normal.dot(dtv); + + // Tangential velocity. + Vector3 tv = dtv - c.normal * tn; + real_t tvl = tv.length(); + + if (tvl > MIN_VELOCITY) { + tv /= tvl; + + Vector3 temp1 = body->get_inv_inertia_tensor().xform(c.rA.cross(tv)); + + real_t t = -tvl / + (body->get_inv_mass() + soft_body->get_node_inv_mass(c.index_B) + tv.dot(temp1.cross(c.rA))); + + Vector3 jt = t * tv; + + Vector3 jtOld = c.acc_tangent_impulse; + c.acc_tangent_impulse += jt; + + real_t fi_len = c.acc_tangent_impulse.length(); + real_t jtMax = c.acc_normal_impulse * friction; + + if (fi_len > CMP_EPSILON && fi_len > jtMax) { + c.acc_tangent_impulse *= jtMax / fi_len; + } + + jt = c.acc_tangent_impulse - jtOld; + + body->apply_impulse(c.rA + body->get_center_of_mass(), -jt); + soft_body->apply_node_impulse(c.index_B, jt); + + c.active = true; + } + } +} + +BodySoftBodyPair3DSW::BodySoftBodyPair3DSW(Body3DSW *p_A, int p_shape_A, SoftBody3DSW *p_B) { + body = p_A; + soft_body = p_B; + body_shape = p_shape_A; + space = p_A->get_space(); + body->add_constraint(this, 0); + soft_body->add_constraint(this); +} + +BodySoftBodyPair3DSW::~BodySoftBodyPair3DSW() { + body->remove_constraint(this); + soft_body->remove_constraint(this); +} diff --git a/servers/physics_3d/body_pair_3d_sw.h b/servers/physics_3d/body_pair_3d_sw.h index 4d049eafdc..74dddfa6aa 100644 --- a/servers/physics_3d/body_pair_3d_sw.h +++ b/servers/physics_3d/body_pair_3d_sw.h @@ -28,32 +28,20 @@ /* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ /*************************************************************************/ -#ifndef BODY_PAIR_SW_H -#define BODY_PAIR_SW_H +#ifndef BODY_PAIR_3D_SW_H +#define BODY_PAIR_3D_SW_H #include "body_3d_sw.h" #include "constraint_3d_sw.h" +#include "core/templates/local_vector.h" +#include "soft_body_3d_sw.h" -class BodyPair3DSW : public Constraint3DSW { - enum { - MAX_CONTACTS = 4 - }; - - union { - struct { - Body3DSW *A; - Body3DSW *B; - }; - - Body3DSW *_arr[2]; - }; - - int shape_A; - int shape_B; - +class BodyContact3DSW : public Constraint3DSW { +protected: struct Contact { Vector3 position; Vector3 normal; + int index_A, index_B; Vector3 local_A, local_B; real_t acc_normal_impulse; // accumulated normal impulse (Pn) Vector3 acc_tangent_impulse; // accumulated tangent impulse (Pt) @@ -68,22 +56,45 @@ class BodyPair3DSW : public Constraint3DSW { Vector3 rA, rB; // Offset in world orientation with respect to center of mass }; + Vector3 sep_axis; + bool collided; + + Space3DSW *space; + + BodyContact3DSW(Body3DSW **p_body_ptr = nullptr, int p_body_count = 0) : + Constraint3DSW(p_body_ptr, p_body_count) { + } +}; + +class BodyPair3DSW : public BodyContact3DSW { + enum { + MAX_CONTACTS = 4 + }; + + union { + struct { + Body3DSW *A; + Body3DSW *B; + }; + + Body3DSW *_arr[2]; + }; + + int shape_A; + int shape_B; + Vector3 offset_B; //use local A coordinates to avoid numerical issues on collision detection - Vector3 sep_axis; Contact contacts[MAX_CONTACTS]; int contact_count; - bool collided; - static void _contact_added_callback(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata); + static void _contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata); - void contact_added_callback(const Vector3 &p_point_A, const Vector3 &p_point_B); + void contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B); void validate_contacts(); bool _test_ccd(real_t p_step, Body3DSW *p_A, int p_shape_A, const Transform &p_xform_A, Body3DSW *p_B, int p_shape_B, const Transform &p_xform_B); - Space3DSW *space; - public: bool setup(real_t p_step); void solve(real_t p_step); @@ -92,4 +103,26 @@ public: ~BodyPair3DSW(); }; -#endif // BODY_PAIR__SW_H +class BodySoftBodyPair3DSW : public BodyContact3DSW { + Body3DSW *body; + SoftBody3DSW *soft_body; + + int body_shape; + + LocalVector<Contact> contacts; + + static void _contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata); + + void contact_added_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B); + + void validate_contacts(); + +public: + bool setup(real_t p_step); + void solve(real_t p_step); + + BodySoftBodyPair3DSW(Body3DSW *p_A, int p_shape_A, SoftBody3DSW *p_B); + ~BodySoftBodyPair3DSW(); +}; + +#endif // BODY_PAIR_3D_SW_H diff --git a/servers/physics_3d/collision_object_3d_sw.cpp b/servers/physics_3d/collision_object_3d_sw.cpp index b06ade5ed3..293a7e6606 100644 --- a/servers/physics_3d/collision_object_3d_sw.cpp +++ b/servers/physics_3d/collision_object_3d_sw.cpp @@ -43,7 +43,7 @@ void CollisionObject3DSW::add_shape(Shape3DSW *p_shape, const Transform &p_trans p_shape->add_owner(this); if (!pending_shape_update_list.in_list()) { - PhysicsServer3DSW::singleton->pending_shape_update_list.add(&pending_shape_update_list); + PhysicsServer3DSW::singletonsw->pending_shape_update_list.add(&pending_shape_update_list); } //_update_shapes(); //_shapes_changed(); @@ -56,7 +56,7 @@ void CollisionObject3DSW::set_shape(int p_index, Shape3DSW *p_shape) { p_shape->add_owner(this); if (!pending_shape_update_list.in_list()) { - PhysicsServer3DSW::singleton->pending_shape_update_list.add(&pending_shape_update_list); + PhysicsServer3DSW::singletonsw->pending_shape_update_list.add(&pending_shape_update_list); } //_update_shapes(); //_shapes_changed(); @@ -68,7 +68,7 @@ void CollisionObject3DSW::set_shape_transform(int p_index, const Transform &p_tr shapes.write[p_index].xform = p_transform; shapes.write[p_index].xform_inv = p_transform.affine_inverse(); if (!pending_shape_update_list.in_list()) { - PhysicsServer3DSW::singleton->pending_shape_update_list.add(&pending_shape_update_list); + PhysicsServer3DSW::singletonsw->pending_shape_update_list.add(&pending_shape_update_list); } //_update_shapes(); //_shapes_changed(); @@ -77,7 +77,7 @@ void CollisionObject3DSW::set_shape_transform(int p_index, const Transform &p_tr void CollisionObject3DSW::set_shape_as_disabled(int p_idx, bool p_enable) { shapes.write[p_idx].disabled = p_enable; if (!pending_shape_update_list.in_list()) { - PhysicsServer3DSW::singleton->pending_shape_update_list.add(&pending_shape_update_list); + PhysicsServer3DSW::singletonsw->pending_shape_update_list.add(&pending_shape_update_list); } } @@ -106,7 +106,7 @@ void CollisionObject3DSW::remove_shape(int p_index) { shapes.remove(p_index); if (!pending_shape_update_list.in_list()) { - PhysicsServer3DSW::singleton->pending_shape_update_list.add(&pending_shape_update_list); + PhysicsServer3DSW::singletonsw->pending_shape_update_list.add(&pending_shape_update_list); } //_update_shapes(); //_shapes_changed(); diff --git a/servers/physics_3d/collision_object_3d_sw.h b/servers/physics_3d/collision_object_3d_sw.h index 3847b81381..85221b7746 100644 --- a/servers/physics_3d/collision_object_3d_sw.h +++ b/servers/physics_3d/collision_object_3d_sw.h @@ -48,7 +48,8 @@ class CollisionObject3DSW : public ShapeOwner3DSW { public: enum Type { TYPE_AREA, - TYPE_BODY + TYPE_BODY, + TYPE_SOFT_BODY, }; private: @@ -129,8 +130,8 @@ public: _FORCE_INLINE_ const AABB &get_shape_aabb(int p_index) const { return shapes[p_index].aabb_cache; } _FORCE_INLINE_ real_t get_shape_area(int p_index) const { return shapes[p_index].area_cache; } - _FORCE_INLINE_ Transform get_transform() const { return transform; } - _FORCE_INLINE_ Transform get_inv_transform() const { return inv_transform; } + _FORCE_INLINE_ const Transform &get_transform() const { return transform; } + _FORCE_INLINE_ const Transform &get_inv_transform() const { return inv_transform; } _FORCE_INLINE_ Space3DSW *get_space() const { return space; } _FORCE_INLINE_ void set_ray_pickable(bool p_enable) { ray_pickable = p_enable; } diff --git a/servers/physics_3d/collision_solver_3d_sat.cpp b/servers/physics_3d/collision_solver_3d_sat.cpp index b8e056f1f4..9d5448dbfa 100644 --- a/servers/physics_3d/collision_solver_3d_sat.cpp +++ b/servers/physics_3d/collision_solver_3d_sat.cpp @@ -31,7 +31,38 @@ #include "collision_solver_3d_sat.h" #include "core/math/geometry_3d.h" -#define _EDGE_IS_VALID_SUPPORT_THRESHOLD 0.02 +#include "gjk_epa.h" + +#define fallback_collision_solver gjk_epa_calculate_penetration + +// Cylinder SAT analytic methods and face-circle contact points for cylinder-trimesh and cylinder-box collision are based on ODE colliders. + +/* + * Cylinder-trimesh and Cylinder-box colliders by Alen Ladavac + * Ported to ODE by Nguyen Binh + */ + +/************************************************************************* + * * + * Open Dynamics Engine, Copyright (C) 2001-2003 Russell L. Smith. * + * All rights reserved. Email: russ@q12.org Web: www.q12.org * + * * + * This library is free software; you can redistribute it and/or * + * modify it under the terms of EITHER: * + * (1) The GNU Lesser General Public License as published by the Free * + * Software Foundation; either version 2.1 of the License, or (at * + * your option) any later version. The text of the GNU Lesser * + * General Public License is included with this library in the * + * file LICENSE.TXT. * + * (2) The BSD-style license that is included with this library in * + * the file LICENSE-BSD.TXT. * + * * + * This library is distributed in the hope that it will be useful, * + * but WITHOUT ANY WARRANTY; without even the implied warranty of * + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the files * + * LICENSE.TXT and LICENSE-BSD.TXT for more details. * + * * + *************************************************************************/ struct _CollectorCallback { CollisionSolver3DSW::CallbackResult callback; @@ -43,9 +74,9 @@ struct _CollectorCallback { _FORCE_INLINE_ void call(const Vector3 &p_point_A, const Vector3 &p_point_B) { if (swap) { - callback(p_point_B, p_point_A, userdata); + callback(p_point_B, 0, p_point_A, 0, userdata); } else { - callback(p_point_A, p_point_B, userdata); + callback(p_point_A, 0, p_point_B, 0, userdata); } } }; @@ -82,6 +113,17 @@ static void _generate_contacts_point_face(const Vector3 *p_points_A, int p_point p_callback->call(*p_points_A, closest_B); } +static void _generate_contacts_point_circle(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { +#ifdef DEBUG_ENABLED + ERR_FAIL_COND(p_point_count_A != 1); + ERR_FAIL_COND(p_point_count_B != 3); +#endif + + Vector3 closest_B = Plane(p_points_B[0], p_points_B[1], p_points_B[2]).project(*p_points_A); + + p_callback->call(*p_points_A, closest_B); +} + static void _generate_contacts_edge_edge(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { #ifdef DEBUG_ENABLED ERR_FAIL_COND(p_point_count_A != 2); @@ -128,6 +170,104 @@ static void _generate_contacts_edge_edge(const Vector3 *p_points_A, int p_point_ p_callback->call(closest_A, closest_B); } +static void _generate_contacts_edge_circle(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { +#ifdef DEBUG_ENABLED + ERR_FAIL_COND(p_point_count_A != 2); + ERR_FAIL_COND(p_point_count_B != 3); +#endif + + const Vector3 &circle_B_pos = p_points_B[0]; + Vector3 circle_B_line_1 = p_points_B[1] - circle_B_pos; + Vector3 circle_B_line_2 = p_points_B[2] - circle_B_pos; + + real_t circle_B_radius = circle_B_line_1.length(); + Vector3 circle_B_normal = circle_B_line_1.cross(circle_B_line_2).normalized(); + + Plane circle_plane(circle_B_pos, circle_B_normal); + + static const int max_clip = 2; + Vector3 contact_points[max_clip]; + int num_points = 0; + + // Project edge point in circle plane. + const Vector3 &edge_A_1 = p_points_A[0]; + Vector3 proj_point_1 = circle_plane.project(edge_A_1); + + Vector3 dist_vec = proj_point_1 - circle_B_pos; + real_t dist_sq = dist_vec.length_squared(); + + // Point 1 is inside disk, add as contact point. + if (dist_sq <= circle_B_radius * circle_B_radius) { + contact_points[num_points] = edge_A_1; + ++num_points; + } + + const Vector3 &edge_A_2 = p_points_A[1]; + Vector3 proj_point_2 = circle_plane.project(edge_A_2); + + Vector3 dist_vec_2 = proj_point_2 - circle_B_pos; + real_t dist_sq_2 = dist_vec_2.length_squared(); + + // Point 2 is inside disk, add as contact point. + if (dist_sq_2 <= circle_B_radius * circle_B_radius) { + contact_points[num_points] = edge_A_2; + ++num_points; + } + + if (num_points < 2) { + Vector3 line_vec = proj_point_2 - proj_point_1; + real_t line_length_sq = line_vec.length_squared(); + + // Create a quadratic formula of the form ax^2 + bx + c = 0 + real_t a, b, c; + + a = line_length_sq; + b = 2.0 * dist_vec.dot(line_vec); + c = dist_sq - circle_B_radius * circle_B_radius; + + // Solve for t. + real_t sqrtterm = b * b - 4.0 * a * c; + + // If the term we intend to square root is less than 0 then the answer won't be real, + // so the line doesn't intersect. + if (sqrtterm >= 0) { + sqrtterm = Math::sqrt(sqrtterm); + + Vector3 edge_dir = edge_A_2 - edge_A_1; + + real_t fraction_1 = (-b - sqrtterm) / (2.0 * a); + if ((fraction_1 > 0.0) && (fraction_1 < 1.0)) { + Vector3 face_point_1 = edge_A_1 + fraction_1 * edge_dir; + ERR_FAIL_COND(num_points >= max_clip); + contact_points[num_points] = face_point_1; + ++num_points; + } + + real_t fraction_2 = (-b + sqrtterm) / (2.0 * a); + if ((fraction_2 > 0.0) && (fraction_2 < 1.0) && !Math::is_equal_approx(fraction_1, fraction_2)) { + Vector3 face_point_2 = edge_A_1 + fraction_2 * edge_dir; + ERR_FAIL_COND(num_points >= max_clip); + contact_points[num_points] = face_point_2; + ++num_points; + } + } + } + + // Generate contact points. + for (int i = 0; i < num_points; i++) { + const Vector3 &contact_point_A = contact_points[i]; + + real_t d = circle_plane.distance_to(contact_point_A); + Vector3 closest_B = contact_point_A - circle_plane.normal * d; + + if (p_callback->normal.dot(contact_point_A) >= p_callback->normal.dot(closest_B)) { + continue; + } + + p_callback->call(contact_point_A, closest_B); + } +} + static void _generate_contacts_face_face(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { #ifdef DEBUG_ENABLED ERR_FAIL_COND(p_point_count_A < 2); @@ -217,36 +357,229 @@ static void _generate_contacts_face_face(const Vector3 *p_points_A, int p_point_ } } -static void _generate_contacts_from_supports(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { +static void _generate_contacts_face_circle(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { +#ifdef DEBUG_ENABLED + ERR_FAIL_COND(p_point_count_A < 3); + ERR_FAIL_COND(p_point_count_B != 3); +#endif + + const Vector3 &circle_B_pos = p_points_B[0]; + Vector3 circle_B_line_1 = p_points_B[1] - circle_B_pos; + Vector3 circle_B_line_2 = p_points_B[2] - circle_B_pos; + + // Clip face with circle segments. + static const int circle_segments = 8; + Vector3 circle_points[circle_segments]; + + real_t angle_delta = 2.0 * Math_PI / circle_segments; + + for (int i = 0; i < circle_segments; ++i) { + Vector3 point_pos = circle_B_pos; + point_pos += circle_B_line_1 * Math::cos(i * angle_delta); + point_pos += circle_B_line_2 * Math::sin(i * angle_delta); + circle_points[i] = point_pos; + } + + _generate_contacts_face_face(p_points_A, p_point_count_A, circle_points, circle_segments, p_callback); + + // Clip face with circle plane. + Vector3 circle_B_normal = circle_B_line_1.cross(circle_B_line_2).normalized(); + + Plane circle_plane(circle_B_pos, circle_B_normal); + + static const int max_clip = 32; + Vector3 contact_points[max_clip]; + int num_points = 0; + + for (int i = 0; i < p_point_count_A; i++) { + int i_n = (i + 1) % p_point_count_A; + + const Vector3 &edge0_A = p_points_A[i]; + const Vector3 &edge1_A = p_points_A[i_n]; + + real_t dist0 = circle_plane.distance_to(edge0_A); + real_t dist1 = circle_plane.distance_to(edge1_A); + + // First point in front of plane, generate contact point. + if (dist0 * circle_plane.d >= 0) { + ERR_FAIL_COND(num_points >= max_clip); + contact_points[num_points] = edge0_A; + ++num_points; + } + + // Points on different sides, generate contact point. + if (dist0 * dist1 < 0) { + // calculate intersection + Vector3 rel = edge1_A - edge0_A; + real_t den = circle_plane.normal.dot(rel); + real_t dist = -(circle_plane.normal.dot(edge0_A) - circle_plane.d) / den; + Vector3 inters = edge0_A + rel * dist; + + ERR_FAIL_COND(num_points >= max_clip); + contact_points[num_points] = inters; + ++num_points; + } + } + + // Generate contact points. + for (int i = 0; i < num_points; i++) { + const Vector3 &contact_point_A = contact_points[i]; + + real_t d = circle_plane.distance_to(contact_point_A); + Vector3 closest_B = contact_point_A - circle_plane.normal * d; + + if (p_callback->normal.dot(contact_point_A) >= p_callback->normal.dot(closest_B)) { + continue; + } + + p_callback->call(contact_point_A, closest_B); + } +} + +static void _generate_contacts_circle_circle(const Vector3 *p_points_A, int p_point_count_A, const Vector3 *p_points_B, int p_point_count_B, _CollectorCallback *p_callback) { +#ifdef DEBUG_ENABLED + ERR_FAIL_COND(p_point_count_A != 3); + ERR_FAIL_COND(p_point_count_B != 3); +#endif + + const Vector3 &circle_A_pos = p_points_A[0]; + Vector3 circle_A_line_1 = p_points_A[1] - circle_A_pos; + Vector3 circle_A_line_2 = p_points_A[2] - circle_A_pos; + + real_t circle_A_radius = circle_A_line_1.length(); + Vector3 circle_A_normal = circle_A_line_1.cross(circle_A_line_2).normalized(); + + const Vector3 &circle_B_pos = p_points_B[0]; + Vector3 circle_B_line_1 = p_points_B[1] - circle_B_pos; + Vector3 circle_B_line_2 = p_points_B[2] - circle_B_pos; + + real_t circle_B_radius = circle_B_line_1.length(); + Vector3 circle_B_normal = circle_B_line_1.cross(circle_B_line_2).normalized(); + + static const int max_clip = 4; + Vector3 contact_points[max_clip]; + int num_points = 0; + + Vector3 centers_diff = circle_B_pos - circle_A_pos; + Vector3 norm_proj = circle_A_normal.dot(centers_diff) * circle_A_normal; + Vector3 comp_proj = centers_diff - norm_proj; + real_t proj_dist = comp_proj.length(); + if (!Math::is_zero_approx(proj_dist)) { + comp_proj /= proj_dist; + if ((proj_dist > circle_A_radius - circle_B_radius) && (proj_dist > circle_B_radius - circle_A_radius)) { + // Circles are overlapping, use the 2 points of intersection as contacts. + real_t radius_a_sqr = circle_A_radius * circle_A_radius; + real_t radius_b_sqr = circle_B_radius * circle_B_radius; + real_t d_sqr = proj_dist * proj_dist; + real_t s = (1.0 + (radius_a_sqr - radius_b_sqr) / d_sqr) * 0.5; + real_t h = Math::sqrt(MAX(radius_a_sqr - d_sqr * s * s, 0.0)); + Vector3 midpoint = circle_A_pos + s * comp_proj * proj_dist; + Vector3 h_vec = h * circle_A_normal.cross(comp_proj); + + Vector3 point_A = midpoint + h_vec; + contact_points[num_points] = point_A; + ++num_points; + + point_A = midpoint - h_vec; + contact_points[num_points] = point_A; + ++num_points; + + // Add 2 points from circle A and B along the line between the centers. + point_A = circle_A_pos + comp_proj * circle_A_radius; + contact_points[num_points] = point_A; + ++num_points; + + point_A = circle_B_pos - comp_proj * circle_B_radius - norm_proj; + contact_points[num_points] = point_A; + ++num_points; + } // Otherwise one circle is inside the other one, use 3 arbitrary equidistant points. + } // Otherwise circles are concentric, use 3 arbitrary equidistant points. + + if (num_points == 0) { + // Generate equidistant points. + if (circle_A_radius < circle_B_radius) { + // Circle A inside circle B. + for (int i = 0; i < 3; ++i) { + Vector3 circle_A_point = circle_A_pos; + circle_A_point += circle_A_line_1 * Math::cos(2.0 * Math_PI * i / 3.0); + circle_A_point += circle_A_line_2 * Math::sin(2.0 * Math_PI * i / 3.0); + + contact_points[num_points] = circle_A_point; + ++num_points; + } + } else { + // Circle B inside circle A. + for (int i = 0; i < 3; ++i) { + Vector3 circle_B_point = circle_B_pos; + circle_B_point += circle_B_line_1 * Math::cos(2.0 * Math_PI * i / 3.0); + circle_B_point += circle_B_line_2 * Math::sin(2.0 * Math_PI * i / 3.0); + + Vector3 circle_A_point = circle_B_point - norm_proj; + + contact_points[num_points] = circle_A_point; + ++num_points; + } + } + } + + Plane circle_B_plane(circle_B_pos, circle_B_normal); + + // Generate contact points. + for (int i = 0; i < num_points; i++) { + const Vector3 &contact_point_A = contact_points[i]; + + real_t d = circle_B_plane.distance_to(contact_point_A); + Vector3 closest_B = contact_point_A - circle_B_plane.normal * d; + + if (p_callback->normal.dot(contact_point_A) >= p_callback->normal.dot(closest_B)) { + continue; + } + + p_callback->call(contact_point_A, closest_B); + } +} + +static void _generate_contacts_from_supports(const Vector3 *p_points_A, int p_point_count_A, Shape3DSW::FeatureType p_feature_type_A, const Vector3 *p_points_B, int p_point_count_B, Shape3DSW::FeatureType p_feature_type_B, _CollectorCallback *p_callback) { #ifdef DEBUG_ENABLED ERR_FAIL_COND(p_point_count_A < 1); ERR_FAIL_COND(p_point_count_B < 1); #endif - static const GenerateContactsFunc generate_contacts_func_table[3][3] = { + static const GenerateContactsFunc generate_contacts_func_table[4][4] = { { _generate_contacts_point_point, _generate_contacts_point_edge, _generate_contacts_point_face, + _generate_contacts_point_circle, }, { nullptr, _generate_contacts_edge_edge, _generate_contacts_face_face, + _generate_contacts_edge_circle, }, { nullptr, nullptr, _generate_contacts_face_face, - } + _generate_contacts_face_circle, + }, + { + nullptr, + nullptr, + nullptr, + _generate_contacts_circle_circle, + }, }; int pointcount_B; int pointcount_A; const Vector3 *points_A; const Vector3 *points_B; + int version_A; + int version_B; - if (p_point_count_A > p_point_count_B) { + if (p_feature_type_A > p_feature_type_B) { //swap p_callback->swap = !p_callback->swap; p_callback->normal = -p_callback->normal; @@ -255,16 +588,17 @@ static void _generate_contacts_from_supports(const Vector3 *p_points_A, int p_po pointcount_A = p_point_count_B; points_A = p_points_B; points_B = p_points_A; + version_A = p_feature_type_B; + version_B = p_feature_type_A; } else { pointcount_B = p_point_count_B; pointcount_A = p_point_count_A; points_A = p_points_A; points_B = p_points_B; + version_A = p_feature_type_A; + version_B = p_feature_type_B; } - int version_A = (pointcount_A > 3 ? 3 : pointcount_A) - 1; - int version_B = (pointcount_B > 3 ? 3 : pointcount_B) - 1; - GenerateContactsFunc contacts_func = generate_contacts_func_table[version_A][version_B]; ERR_FAIL_COND(!contacts_func); contacts_func(points_A, pointcount_A, points_B, pointcount_B, p_callback); @@ -292,7 +626,7 @@ public: } } - _FORCE_INLINE_ bool test_axis(const Vector3 &p_axis) { + _FORCE_INLINE_ bool test_axis(const Vector3 &p_axis, bool p_directional = false) { Vector3 axis = p_axis; if (Math::abs(axis.x) < CMP_EPSILON && @@ -328,7 +662,12 @@ public: //use the smallest depth if (min_B < 0.0) { // could be +0.0, we don't want it to become -0.0 - min_B = -min_B; + if (p_directional) { + min_B = max_B; + axis = -axis; + } else { + min_B = -min_B; + } } if (max_B < min_B) { @@ -346,6 +685,17 @@ public: return true; } + static _FORCE_INLINE_ void test_contact_points(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { + SeparatorAxisTest<ShapeA, ShapeB, withMargin> *separator = (SeparatorAxisTest<ShapeA, ShapeB, withMargin> *)p_userdata; + Vector3 axis = (p_point_B - p_point_A); + real_t depth = axis.length(); + + // Filter out bogus directions with a treshold and re-testing axis. + if (separator->best_depth - depth > 0.001) { + separator->test_axis(axis / depth); + } + } + _FORCE_INLINE_ void generate_contacts() { // nothing to do, don't generate if (best_axis == Vector3(0.0, 0.0, 0.0)) { @@ -365,7 +715,8 @@ public: Vector3 supports_A[max_supports]; int support_count_A; - shape_A->get_supports(transform_A->basis.xform_inv(-best_axis).normalized(), max_supports, supports_A, support_count_A); + Shape3DSW::FeatureType support_type_A; + shape_A->get_supports(transform_A->basis.xform_inv(-best_axis).normalized(), max_supports, supports_A, support_count_A, support_type_A); for (int i = 0; i < support_count_A; i++) { supports_A[i] = transform_A->xform(supports_A[i]); } @@ -378,7 +729,8 @@ public: Vector3 supports_B[max_supports]; int support_count_B; - shape_B->get_supports(transform_B->basis.xform_inv(best_axis).normalized(), max_supports, supports_B, support_count_B); + Shape3DSW::FeatureType support_type_B; + shape_B->get_supports(transform_B->basis.xform_inv(best_axis).normalized(), max_supports, supports_B, support_count_B, support_type_B); for (int i = 0; i < support_count_B; i++) { supports_B[i] = transform_B->xform(supports_B[i]); } @@ -393,7 +745,7 @@ public: if (callback->prev_axis) { *callback->prev_axis = best_axis; } - _generate_contacts_from_supports(supports_A, support_count_A, supports_B, support_count_B, callback); + _generate_contacts_from_supports(supports_A, support_count_A, support_type_A, supports_B, support_count_B, support_type_B, callback); callback->collided = true; } @@ -498,7 +850,7 @@ static void _collision_sphere_capsule(const Shape3DSW *p_a, const Transform &p_t //capsule sphere 1, sphere - Vector3 capsule_axis = p_transform_b.basis.get_axis(2) * (capsule_B->get_height() * 0.5); + Vector3 capsule_axis = p_transform_b.basis.get_axis(1) * (capsule_B->get_height() * 0.5); Vector3 capsule_ball_1 = p_transform_b.origin + capsule_axis; @@ -529,6 +881,61 @@ static void _collision_sphere_capsule(const Shape3DSW *p_a, const Transform &p_t template <bool withMargin> static void _collision_sphere_cylinder(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const SphereShape3DSW *sphere_A = static_cast<const SphereShape3DSW *>(p_a); + const CylinderShape3DSW *cylinder_B = static_cast<const CylinderShape3DSW *>(p_b); + + SeparatorAxisTest<SphereShape3DSW, CylinderShape3DSW, withMargin> separator(sphere_A, p_transform_a, cylinder_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + if (!separator.test_previous_axis()) { + return; + } + + // Cylinder B end caps. + Vector3 cylinder_B_axis = p_transform_b.basis.get_axis(1).normalized(); + if (!separator.test_axis(cylinder_B_axis)) { + return; + } + + Vector3 cylinder_diff = p_transform_b.origin - p_transform_a.origin; + + // Cylinder B lateral surface. + if (!separator.test_axis(cylinder_B_axis.cross(cylinder_diff).cross(cylinder_B_axis).normalized())) { + return; + } + + // Closest point to cylinder caps. + const Vector3 &sphere_center = p_transform_a.origin; + Vector3 cyl_axis = p_transform_b.basis.get_axis(1); + Vector3 cap_axis = p_transform_b.basis.get_axis(0); + real_t height_scale = cyl_axis.length(); + real_t cap_dist = cylinder_B->get_height() * 0.5 * height_scale; + cyl_axis /= height_scale; + real_t radius_scale = cap_axis.length(); + real_t cap_radius = cylinder_B->get_radius() * radius_scale; + + for (int i = 0; i < 2; i++) { + Vector3 cap_dir = ((i == 0) ? cyl_axis : -cyl_axis); + Vector3 cap_pos = p_transform_b.origin + cap_dir * cap_dist; + + Vector3 closest_point; + + Vector3 diff = sphere_center - cap_pos; + Vector3 proj = diff - cap_dir.dot(diff) * cap_dir; + + real_t proj_len = proj.length(); + if (Math::is_zero_approx(proj_len)) { + // Point is equidistant to all circle points. + continue; + } + + closest_point = cap_pos + (cap_radius / proj_len) * proj; + + if (!separator.test_axis((closest_point - sphere_center).normalized())) { + return; + } + } + + separator.generate_contacts(); } template <bool withMargin> @@ -604,23 +1011,31 @@ static void _collision_sphere_face(const Shape3DSW *p_a, const Transform &p_tran p_transform_b.xform(face_B->vertex[2]), }; - if (!separator.test_axis((vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized())) { + Vector3 normal = (vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized(); + + if (!separator.test_axis(normal, !face_B->backface_collision)) { return; } // edges and points of B for (int i = 0; i < 3; i++) { Vector3 n1 = vertex[i] - p_transform_a.origin; + if (n1.dot(normal) < 0.0) { + n1 *= -1.0; + } - if (!separator.test_axis(n1.normalized())) { + if (!separator.test_axis(n1.normalized(), !face_B->backface_collision)) { return; } Vector3 n2 = vertex[(i + 1) % 3] - vertex[i]; Vector3 axis = n1.cross(n2).cross(n2).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis)) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -739,14 +1154,14 @@ static void _collision_box_capsule(const Shape3DSW *p_a, const Transform &p_tran // faces of A for (int i = 0; i < 3; i++) { - Vector3 axis = p_transform_a.basis.get_axis(i); + Vector3 axis = p_transform_a.basis.get_axis(i).normalized(); if (!separator.test_axis(axis)) { return; } } - Vector3 cyl_axis = p_transform_b.basis.get_axis(2).normalized(); + Vector3 cyl_axis = p_transform_b.basis.get_axis(1).normalized(); // edges of A, capsule cylinder @@ -791,7 +1206,7 @@ static void _collision_box_capsule(const Shape3DSW *p_a, const Transform &p_tran // capsule balls, edges of A for (int i = 0; i < 2; i++) { - Vector3 capsule_axis = p_transform_b.basis.get_axis(2) * (capsule_B->get_height() * 0.5); + Vector3 capsule_axis = p_transform_b.basis.get_axis(1) * (capsule_B->get_height() * 0.5); Vector3 sphere_pos = p_transform_b.origin + ((i == 0) ? capsule_axis : -capsule_axis); @@ -826,6 +1241,115 @@ static void _collision_box_capsule(const Shape3DSW *p_a, const Transform &p_tran template <bool withMargin> static void _collision_box_cylinder(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const BoxShape3DSW *box_A = static_cast<const BoxShape3DSW *>(p_a); + const CylinderShape3DSW *cylinder_B = static_cast<const CylinderShape3DSW *>(p_b); + + SeparatorAxisTest<BoxShape3DSW, CylinderShape3DSW, withMargin> separator(box_A, p_transform_a, cylinder_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + if (!separator.test_previous_axis()) { + return; + } + + // Faces of A. + for (int i = 0; i < 3; i++) { + Vector3 axis = p_transform_a.basis.get_axis(i).normalized(); + + if (!separator.test_axis(axis)) { + return; + } + } + + Vector3 cyl_axis = p_transform_b.basis.get_axis(1).normalized(); + + // Cylinder end caps. + { + if (!separator.test_axis(cyl_axis)) { + return; + } + } + + // Edges of A, cylinder lateral surface. + for (int i = 0; i < 3; i++) { + Vector3 box_axis = p_transform_a.basis.get_axis(i); + Vector3 axis = box_axis.cross(cyl_axis); + if (Math::is_zero_approx(axis.length_squared())) { + continue; + } + + if (!separator.test_axis(axis.normalized())) { + return; + } + } + + // Gather points of A. + Vector3 vertices_A[8]; + Vector3 box_extent = box_A->get_half_extents(); + for (int i = 0; i < 2; i++) { + for (int j = 0; j < 2; j++) { + for (int k = 0; k < 2; k++) { + Vector3 extent = box_extent; + extent.x *= (i * 2 - 1); + extent.y *= (j * 2 - 1); + extent.z *= (k * 2 - 1); + Vector3 &point = vertices_A[i * 2 * 2 + j * 2 + k]; + point = p_transform_a.origin; + for (int l = 0; l < 3; l++) { + point += p_transform_a.basis.get_axis(l) * extent[l]; + } + } + } + } + + // Points of A, cylinder lateral surface. + for (int i = 0; i < 8; i++) { + const Vector3 &point = vertices_A[i]; + Vector3 axis = Plane(cyl_axis, 0).project(point).normalized(); + + if (!separator.test_axis(axis)) { + return; + } + } + + // Edges of A, cylinder end caps rim. + int edges_start_A[12] = { 0, 2, 4, 6, 0, 1, 4, 5, 0, 1, 2, 3 }; + int edges_end_A[12] = { 1, 3, 5, 7, 2, 3, 6, 7, 4, 5, 6, 7 }; + + Vector3 cap_axis = cyl_axis * (cylinder_B->get_height() * 0.5); + + for (int i = 0; i < 2; i++) { + Vector3 cap_pos = p_transform_b.origin + ((i == 0) ? cap_axis : -cap_axis); + + for (int e = 0; e < 12; e++) { + const Vector3 &edge_start = vertices_A[edges_start_A[e]]; + const Vector3 &edge_end = vertices_A[edges_end_A[e]]; + + Vector3 edge_dir = (edge_end - edge_start); + edge_dir.normalize(); + + real_t edge_dot = edge_dir.dot(cyl_axis); + if (Math::is_zero_approx(edge_dot)) { + // Edge is perpendicular to cylinder axis. + continue; + } + + // Calculate intersection between edge and circle plane. + Vector3 edge_diff = cap_pos - edge_start; + real_t diff_dot = edge_diff.dot(cyl_axis); + Vector3 intersection = edge_start + edge_dir * diff_dot / edge_dot; + + // Calculate tangent that touches intersection. + Vector3 tangent = (cap_pos - intersection).cross(cyl_axis); + + // Axis is orthogonal both to tangent and edge direction. + Vector3 axis = tangent.cross(edge_dir); + + if (!separator.test_axis(axis.normalized())) { + return; + } + } + } + + separator.generate_contacts(); } template <bool withMargin> @@ -956,15 +1480,20 @@ static void _collision_box_face(const Shape3DSW *p_a, const Transform &p_transfo p_transform_b.xform(face_B->vertex[2]), }; - if (!separator.test_axis((vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized())) { + Vector3 normal = (vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized(); + + if (!separator.test_axis(normal, !face_B->backface_collision)) { return; } // faces of A for (int i = 0; i < 3; i++) { Vector3 axis = p_transform_a.basis.get_axis(i).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis)) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -975,9 +1504,12 @@ static void _collision_box_face(const Shape3DSW *p_a, const Transform &p_transfo Vector3 e = vertex[i] - vertex[(i + 1) % 3]; for (int j = 0; j < 3; j++) { - Vector3 axis = p_transform_a.basis.get_axis(j); + Vector3 axis = e.cross(p_transform_a.basis.get_axis(j)).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(e.cross(axis).normalized())) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -997,8 +1529,11 @@ static void _collision_box_face(const Shape3DSW *p_a, const Transform &p_transfo (cnormal_a.z < 0) ? -box_A->get_half_extents().z : box_A->get_half_extents().z)); Vector3 axis_ab = support_a - vertex[v]; + if (axis_ab.dot(normal) < 0.0) { + axis_ab *= -1.0; + } - if (!separator.test_axis(axis_ab.normalized())) { + if (!separator.test_axis(axis_ab.normalized(), !face_B->backface_collision)) { return; } @@ -1008,7 +1543,12 @@ static void _collision_box_face(const Shape3DSW *p_a, const Transform &p_transfo //a ->b Vector3 axis_a = p_transform_a.basis.get_axis(i); - if (!separator.test_axis(axis_ab.cross(axis_a).cross(axis_a).normalized())) { + Vector3 axis = axis_ab.cross(axis_a).cross(axis_a).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1033,7 +1573,12 @@ static void _collision_box_face(const Shape3DSW *p_a, const Transform &p_transfo Vector3 n = (p2 - p1); - if (!separator.test_axis((point - p2).cross(n).cross(n).normalized())) { + Vector3 axis = (point - p2).cross(n).cross(n).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1058,8 +1603,8 @@ static void _collision_capsule_capsule(const Shape3DSW *p_a, const Transform &p_ // some values - Vector3 capsule_A_axis = p_transform_a.basis.get_axis(2) * (capsule_A->get_height() * 0.5); - Vector3 capsule_B_axis = p_transform_b.basis.get_axis(2) * (capsule_B->get_height() * 0.5); + Vector3 capsule_A_axis = p_transform_a.basis.get_axis(1) * (capsule_A->get_height() * 0.5); + Vector3 capsule_B_axis = p_transform_b.basis.get_axis(1) * (capsule_B->get_height() * 0.5); Vector3 capsule_A_ball_1 = p_transform_a.origin + capsule_A_axis; Vector3 capsule_A_ball_2 = p_transform_a.origin - capsule_A_axis; @@ -1111,6 +1656,64 @@ static void _collision_capsule_capsule(const Shape3DSW *p_a, const Transform &p_ template <bool withMargin> static void _collision_capsule_cylinder(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const CapsuleShape3DSW *capsule_A = static_cast<const CapsuleShape3DSW *>(p_a); + const CylinderShape3DSW *cylinder_B = static_cast<const CylinderShape3DSW *>(p_b); + + SeparatorAxisTest<CapsuleShape3DSW, CylinderShape3DSW, withMargin> separator(capsule_A, p_transform_a, cylinder_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + if (!separator.test_previous_axis()) { + return; + } + + // Cylinder B end caps. + Vector3 cylinder_B_axis = p_transform_b.basis.get_axis(1).normalized(); + if (!separator.test_axis(cylinder_B_axis)) { + return; + } + + // Cylinder edge against capsule balls. + + Vector3 capsule_A_axis = p_transform_a.basis.get_axis(1); + + Vector3 capsule_A_ball_1 = p_transform_a.origin + capsule_A_axis * (capsule_A->get_height() * 0.5); + Vector3 capsule_A_ball_2 = p_transform_a.origin - capsule_A_axis * (capsule_A->get_height() * 0.5); + + if (!separator.test_axis((p_transform_b.origin - capsule_A_ball_1).cross(cylinder_B_axis).cross(cylinder_B_axis).normalized())) { + return; + } + + if (!separator.test_axis((p_transform_b.origin - capsule_A_ball_2).cross(cylinder_B_axis).cross(cylinder_B_axis).normalized())) { + return; + } + + // Cylinder edge against capsule edge. + + Vector3 center_diff = p_transform_b.origin - p_transform_a.origin; + + if (!separator.test_axis(capsule_A_axis.cross(center_diff).cross(capsule_A_axis).normalized())) { + return; + } + + if (!separator.test_axis(cylinder_B_axis.cross(center_diff).cross(cylinder_B_axis).normalized())) { + return; + } + + real_t proj = capsule_A_axis.cross(cylinder_B_axis).cross(cylinder_B_axis).dot(capsule_A_axis); + if (Math::is_zero_approx(proj)) { + // Parallel capsule and cylinder axes, handle with specific axes only. + // Note: GJKEPA with no margin can lead to degenerate cases in this situation. + separator.generate_contacts(); + return; + } + + CollisionSolver3DSW::CallbackResult callback = SeparatorAxisTest<CapsuleShape3DSW, CylinderShape3DSW, withMargin>::test_contact_points; + + // Fallback to generic algorithm to find the best separating axis. + if (!fallback_collision_solver(p_a, p_transform_a, p_b, p_transform_b, callback, &separator, false, p_margin_a, p_margin_b)) { + return; + } + + separator.generate_contacts(); } template <bool withMargin> @@ -1146,7 +1749,7 @@ static void _collision_capsule_convex_polygon(const Shape3DSW *p_a, const Transf for (int i = 0; i < edge_count; i++) { // cylinder Vector3 edge_axis = p_transform_b.basis.xform(vertices[edges[i].a]) - p_transform_b.basis.xform(vertices[edges[i].b]); - Vector3 axis = edge_axis.cross(p_transform_a.basis.get_axis(2)).normalized(); + Vector3 axis = edge_axis.cross(p_transform_a.basis.get_axis(1)).normalized(); if (!separator.test_axis(axis)) { return; @@ -1158,7 +1761,7 @@ static void _collision_capsule_convex_polygon(const Shape3DSW *p_a, const Transf for (int i = 0; i < 2; i++) { // edges of B, capsule cylinder - Vector3 capsule_axis = p_transform_a.basis.get_axis(2) * (capsule_A->get_height() * 0.5); + Vector3 capsule_axis = p_transform_a.basis.get_axis(1) * (capsule_A->get_height() * 0.5); Vector3 sphere_pos = p_transform_a.origin + ((i == 0) ? capsule_axis : -capsule_axis); @@ -1190,24 +1793,35 @@ static void _collision_capsule_face(const Shape3DSW *p_a, const Transform &p_tra p_transform_b.xform(face_B->vertex[2]), }; - if (!separator.test_axis((vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized())) { + Vector3 normal = (vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized(); + + if (!separator.test_axis(normal, !face_B->backface_collision)) { return; } // edges of B, capsule cylinder - Vector3 capsule_axis = p_transform_a.basis.get_axis(2) * (capsule_A->get_height() * 0.5); + Vector3 capsule_axis = p_transform_a.basis.get_axis(1) * (capsule_A->get_height() * 0.5); for (int i = 0; i < 3; i++) { // edge-cylinder Vector3 edge_axis = vertex[i] - vertex[(i + 1) % 3]; + Vector3 axis = edge_axis.cross(capsule_axis).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis)) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } - if (!separator.test_axis((p_transform_a.origin - vertex[i]).cross(capsule_axis).cross(capsule_axis).normalized())) { + Vector3 dir_axis = (p_transform_a.origin - vertex[i]).cross(capsule_axis).cross(capsule_axis).normalized(); + if (dir_axis.dot(normal) < 0.0) { + dir_axis *= -1.0; + } + + if (!separator.test_axis(dir_axis, !face_B->backface_collision)) { return; } @@ -1216,16 +1830,22 @@ static void _collision_capsule_face(const Shape3DSW *p_a, const Transform &p_tra Vector3 sphere_pos = p_transform_a.origin + ((j == 0) ? capsule_axis : -capsule_axis); Vector3 n1 = sphere_pos - vertex[i]; + if (n1.dot(normal) < 0.0) { + n1 *= -1.0; + } - if (!separator.test_axis(n1.normalized())) { + if (!separator.test_axis(n1.normalized(), !face_B->backface_collision)) { return; } Vector3 n2 = edge_axis; axis = n1.cross(n2).cross(n2); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis.normalized())) { + if (!separator.test_axis(axis.normalized(), !face_B->backface_collision)) { return; } } @@ -1236,14 +1856,178 @@ static void _collision_capsule_face(const Shape3DSW *p_a, const Transform &p_tra template <bool withMargin> static void _collision_cylinder_cylinder(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const CylinderShape3DSW *cylinder_A = static_cast<const CylinderShape3DSW *>(p_a); + const CylinderShape3DSW *cylinder_B = static_cast<const CylinderShape3DSW *>(p_b); + + SeparatorAxisTest<CylinderShape3DSW, CylinderShape3DSW, withMargin> separator(cylinder_A, p_transform_a, cylinder_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + Vector3 cylinder_A_axis = p_transform_a.basis.get_axis(1); + Vector3 cylinder_B_axis = p_transform_b.basis.get_axis(1); + + if (!separator.test_previous_axis()) { + return; + } + + // Cylinder A end caps. + if (!separator.test_axis(cylinder_A_axis.normalized())) { + return; + } + + // Cylinder B end caps. + if (!separator.test_axis(cylinder_A_axis.normalized())) { + return; + } + + Vector3 cylinder_diff = p_transform_b.origin - p_transform_a.origin; + + // Cylinder A lateral surface. + if (!separator.test_axis(cylinder_A_axis.cross(cylinder_diff).cross(cylinder_A_axis).normalized())) { + return; + } + + // Cylinder B lateral surface. + if (!separator.test_axis(cylinder_B_axis.cross(cylinder_diff).cross(cylinder_B_axis).normalized())) { + return; + } + + real_t proj = cylinder_A_axis.cross(cylinder_B_axis).cross(cylinder_B_axis).dot(cylinder_A_axis); + if (Math::is_zero_approx(proj)) { + // Parallel cylinders, handle with specific axes only. + // Note: GJKEPA with no margin can lead to degenerate cases in this situation. + separator.generate_contacts(); + return; + } + + CollisionSolver3DSW::CallbackResult callback = SeparatorAxisTest<CylinderShape3DSW, CylinderShape3DSW, withMargin>::test_contact_points; + + // Fallback to generic algorithm to find the best separating axis. + if (!fallback_collision_solver(p_a, p_transform_a, p_b, p_transform_b, callback, &separator, false, p_margin_a, p_margin_b)) { + return; + } + + separator.generate_contacts(); } template <bool withMargin> static void _collision_cylinder_convex_polygon(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const CylinderShape3DSW *cylinder_A = static_cast<const CylinderShape3DSW *>(p_a); + const ConvexPolygonShape3DSW *convex_polygon_B = static_cast<const ConvexPolygonShape3DSW *>(p_b); + + SeparatorAxisTest<CylinderShape3DSW, ConvexPolygonShape3DSW, withMargin> separator(cylinder_A, p_transform_a, convex_polygon_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + CollisionSolver3DSW::CallbackResult callback = SeparatorAxisTest<CylinderShape3DSW, ConvexPolygonShape3DSW, withMargin>::test_contact_points; + + // Fallback to generic algorithm to find the best separating axis. + if (!fallback_collision_solver(p_a, p_transform_a, p_b, p_transform_b, callback, &separator, false, p_margin_a, p_margin_b)) { + return; + } + + separator.generate_contacts(); } template <bool withMargin> static void _collision_cylinder_face(const Shape3DSW *p_a, const Transform &p_transform_a, const Shape3DSW *p_b, const Transform &p_transform_b, _CollectorCallback *p_collector, real_t p_margin_a, real_t p_margin_b) { + const CylinderShape3DSW *cylinder_A = static_cast<const CylinderShape3DSW *>(p_a); + const FaceShape3DSW *face_B = static_cast<const FaceShape3DSW *>(p_b); + + SeparatorAxisTest<CylinderShape3DSW, FaceShape3DSW, withMargin> separator(cylinder_A, p_transform_a, face_B, p_transform_b, p_collector, p_margin_a, p_margin_b); + + if (!separator.test_previous_axis()) { + return; + } + + Vector3 vertex[3] = { + p_transform_b.xform(face_B->vertex[0]), + p_transform_b.xform(face_B->vertex[1]), + p_transform_b.xform(face_B->vertex[2]), + }; + + Vector3 normal = (vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized(); + + // Face B normal. + if (!separator.test_axis(normal, !face_B->backface_collision)) { + return; + } + + Vector3 cyl_axis = p_transform_a.basis.get_axis(1).normalized(); + if (cyl_axis.dot(normal) < 0.0) { + cyl_axis *= -1.0; + } + + // Cylinder end caps. + if (!separator.test_axis(cyl_axis, !face_B->backface_collision)) { + return; + } + + // Edges of B, cylinder lateral surface. + for (int i = 0; i < 3; i++) { + Vector3 edge_axis = vertex[i] - vertex[(i + 1) % 3]; + Vector3 axis = edge_axis.cross(cyl_axis); + if (Math::is_zero_approx(axis.length_squared())) { + continue; + } + + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis.normalized(), !face_B->backface_collision)) { + return; + } + } + + // Points of B, cylinder lateral surface. + for (int i = 0; i < 3; i++) { + const Vector3 &point = vertex[i]; + Vector3 axis = Plane(cyl_axis, 0).project(point).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { + return; + } + } + + // Edges of B, cylinder end caps rim. + Vector3 cap_axis = cyl_axis * (cylinder_A->get_height() * 0.5); + + for (int i = 0; i < 2; i++) { + Vector3 cap_pos = p_transform_a.origin + ((i == 0) ? cap_axis : -cap_axis); + + for (int j = 0; j < 3; j++) { + const Vector3 &edge_start = vertex[j]; + const Vector3 &edge_end = vertex[(j + 1) % 3]; + Vector3 edge_dir = edge_end - edge_start; + edge_dir.normalize(); + + real_t edge_dot = edge_dir.dot(cyl_axis); + if (Math::is_zero_approx(edge_dot)) { + // Edge is perpendicular to cylinder axis. + continue; + } + + // Calculate intersection between edge and circle plane. + Vector3 edge_diff = cap_pos - edge_start; + real_t diff_dot = edge_diff.dot(cyl_axis); + Vector3 intersection = edge_start + edge_dir * diff_dot / edge_dot; + + // Calculate tangent that touches intersection. + Vector3 tangent = (cap_pos - intersection).cross(cyl_axis); + + // Axis is orthogonal both to tangent and edge direction. + Vector3 axis = tangent.cross(edge_dir); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis.normalized(), !face_B->backface_collision)) { + return; + } + } + } + + separator.generate_contacts(); } template <bool withMargin> @@ -1377,7 +2161,9 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform p_transform_b.xform(face_B->vertex[2]), }; - if (!separator.test_axis((vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized())) { + Vector3 normal = (vertex[0] - vertex[2]).cross(vertex[0] - vertex[1]).normalized(); + + if (!separator.test_axis(normal, !face_B->backface_collision)) { return; } @@ -1385,8 +2171,11 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform for (int i = 0; i < face_count; i++) { //Vector3 axis = p_transform_a.xform( faces[i].plane ).normal; Vector3 axis = p_transform_a.basis.xform(faces[i].plane.normal).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis)) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1399,8 +2188,11 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform Vector3 e2 = vertex[j] - vertex[(j + 1) % 3]; Vector3 axis = e1.cross(e2).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } - if (!separator.test_axis(axis)) { + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1412,7 +2204,12 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform Vector3 va = p_transform_a.xform(vertices[i]); for (int j = 0; j < 3; j++) { - if (!separator.test_axis((va - vertex[j]).normalized())) { + Vector3 axis = (va - vertex[j]).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1427,7 +2224,12 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform for (int j = 0; j < 3; j++) { Vector3 e3 = vertex[j]; - if (!separator.test_axis((e1 - e3).cross(n).cross(n).normalized())) { + Vector3 axis = (e1 - e3).cross(n).cross(n).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } @@ -1441,7 +2243,12 @@ static void _collision_convex_polygon_face(const Shape3DSW *p_a, const Transform for (int j = 0; j < vertex_count; j++) { Vector3 e3 = p_transform_a.xform(vertices[j]); - if (!separator.test_axis((e1 - e3).cross(n).cross(n).normalized())) { + Vector3 axis = (e1 - e3).cross(n).cross(n).normalized(); + if (axis.dot(normal) < 0.0) { + axis *= -1.0; + } + + if (!separator.test_axis(axis, !face_B->backface_collision)) { return; } } diff --git a/servers/physics_3d/collision_solver_3d_sw.cpp b/servers/physics_3d/collision_solver_3d_sw.cpp index 1150696b84..f655c4626c 100644 --- a/servers/physics_3d/collision_solver_3d_sw.cpp +++ b/servers/physics_3d/collision_solver_3d_sw.cpp @@ -30,6 +30,7 @@ #include "collision_solver_3d_sw.h" #include "collision_solver_3d_sat.h" +#include "soft_body_3d_sw.h" #include "gjk_epa.h" @@ -46,8 +47,24 @@ bool CollisionSolver3DSW::solve_static_plane(const Shape3DSW *p_shape_A, const T static const int max_supports = 16; Vector3 supports[max_supports]; int support_count; - - p_shape_B->get_supports(p_transform_B.basis.xform_inv(-p.normal).normalized(), max_supports, supports, support_count); + Shape3DSW::FeatureType support_type; + p_shape_B->get_supports(p_transform_B.basis.xform_inv(-p.normal).normalized(), max_supports, supports, support_count, support_type); + + if (support_type == Shape3DSW::FEATURE_CIRCLE) { + ERR_FAIL_COND_V(support_count != 3, false); + + Vector3 circle_pos = supports[0]; + Vector3 circle_axis_1 = supports[1] - circle_pos; + Vector3 circle_axis_2 = supports[2] - circle_pos; + + // Use 3 equidistant points on the circle. + for (int i = 0; i < 3; ++i) { + Vector3 vertex_pos = circle_pos; + vertex_pos += circle_axis_1 * Math::cos(2.0 * Math_PI * i / 3.0); + vertex_pos += circle_axis_2 * Math::sin(2.0 * Math_PI * i / 3.0); + supports[i] = vertex_pos; + } + } bool found = false; @@ -62,9 +79,9 @@ bool CollisionSolver3DSW::solve_static_plane(const Shape3DSW *p_shape_A, const T if (p_result_callback) { if (p_swap_result) { - p_result_callback(supports[i], support_A, p_userdata); + p_result_callback(supports[i], 0, support_A, 0, p_userdata); } else { - p_result_callback(support_A, supports[i], p_userdata); + p_result_callback(support_A, 0, supports[i], 0, p_userdata); } } } @@ -97,14 +114,148 @@ bool CollisionSolver3DSW::solve_ray(const Shape3DSW *p_shape_A, const Transform if (p_result_callback) { if (p_swap_result) { - p_result_callback(support_B, support_A, p_userdata); + p_result_callback(support_B, 0, support_A, 0, p_userdata); } else { - p_result_callback(support_A, support_B, p_userdata); + p_result_callback(support_A, 0, support_B, 0, p_userdata); } } return true; } +struct _SoftBodyContactCollisionInfo { + int node_index = 0; + CollisionSolver3DSW::CallbackResult result_callback = nullptr; + void *userdata = nullptr; + bool swap_result = false; + int contact_count = 0; +}; + +void CollisionSolver3DSW::soft_body_contact_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { + _SoftBodyContactCollisionInfo &cinfo = *(_SoftBodyContactCollisionInfo *)(p_userdata); + + ++cinfo.contact_count; + + if (cinfo.swap_result) { + cinfo.result_callback(p_point_B, cinfo.node_index, p_point_A, p_index_A, cinfo.userdata); + } else { + cinfo.result_callback(p_point_A, p_index_A, p_point_B, cinfo.node_index, cinfo.userdata); + } +} + +struct _SoftBodyQueryInfo { + SoftBody3DSW *soft_body = nullptr; + const Shape3DSW *shape_A = nullptr; + const Shape3DSW *shape_B = nullptr; + Transform transform_A; + Transform node_transform; + _SoftBodyContactCollisionInfo contact_info; +#ifdef DEBUG_ENABLED + int node_query_count = 0; + int convex_query_count = 0; +#endif +}; + +bool CollisionSolver3DSW::soft_body_query_callback(uint32_t p_node_index, void *p_userdata) { + _SoftBodyQueryInfo &query_cinfo = *(_SoftBodyQueryInfo *)(p_userdata); + + Vector3 node_position = query_cinfo.soft_body->get_node_position(p_node_index); + + Transform transform_B; + transform_B.origin = query_cinfo.node_transform.xform(node_position); + + query_cinfo.contact_info.node_index = p_node_index; + solve_static(query_cinfo.shape_A, query_cinfo.transform_A, query_cinfo.shape_B, transform_B, soft_body_contact_callback, &query_cinfo.contact_info); + +#ifdef DEBUG_ENABLED + ++query_cinfo.node_query_count; +#endif + + // Continue with the query. + return false; +} + +void CollisionSolver3DSW::soft_body_concave_callback(void *p_userdata, Shape3DSW *p_convex) { + _SoftBodyQueryInfo &query_cinfo = *(_SoftBodyQueryInfo *)(p_userdata); + + query_cinfo.shape_A = p_convex; + + // Calculate AABB for internal soft body query (in world space). + AABB shape_aabb; + for (int i = 0; i < 3; i++) { + Vector3 axis; + axis[i] = 1.0; + + real_t smin, smax; + p_convex->project_range(axis, query_cinfo.transform_A, smin, smax); + + shape_aabb.position[i] = smin; + shape_aabb.size[i] = smax - smin; + } + + shape_aabb.grow_by(query_cinfo.soft_body->get_collision_margin()); + + query_cinfo.soft_body->query_aabb(shape_aabb, soft_body_query_callback, &query_cinfo); + +#ifdef DEBUG_ENABLED + ++query_cinfo.convex_query_count; +#endif +} + +bool CollisionSolver3DSW::solve_soft_body(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CallbackResult p_result_callback, void *p_userdata, bool p_swap_result) { + const SoftBodyShape3DSW *soft_body_shape_B = static_cast<const SoftBodyShape3DSW *>(p_shape_B); + + SoftBody3DSW *soft_body = soft_body_shape_B->get_soft_body(); + const Transform &world_to_local = soft_body->get_inv_transform(); + + const real_t collision_margin = soft_body->get_collision_margin(); + + SphereShape3DSW sphere_shape; + sphere_shape.set_data(collision_margin); + + _SoftBodyQueryInfo query_cinfo; + query_cinfo.contact_info.result_callback = p_result_callback; + query_cinfo.contact_info.userdata = p_userdata; + query_cinfo.contact_info.swap_result = p_swap_result; + query_cinfo.soft_body = soft_body; + query_cinfo.node_transform = p_transform_B * world_to_local; + query_cinfo.shape_A = p_shape_A; + query_cinfo.transform_A = p_transform_A; + query_cinfo.shape_B = &sphere_shape; + + if (p_shape_A->is_concave()) { + // In case of concave shape, query convex shapes first. + const ConcaveShape3DSW *concave_shape_A = static_cast<const ConcaveShape3DSW *>(p_shape_A); + + AABB soft_body_aabb = soft_body->get_bounds(); + soft_body_aabb.grow_by(collision_margin); + + // Calculate AABB for internal concave shape query (in local space). + AABB local_aabb; + for (int i = 0; i < 3; i++) { + Vector3 axis(p_transform_A.basis.get_axis(i)); + real_t axis_scale = 1.0 / axis.length(); + + real_t smin = soft_body_aabb.position[i]; + real_t smax = smin + soft_body_aabb.size[i]; + + smin *= axis_scale; + smax *= axis_scale; + + local_aabb.position[i] = smin; + local_aabb.size[i] = smax - smin; + } + + concave_shape_A->cull(local_aabb, soft_body_concave_callback, &query_cinfo); + } else { + AABB shape_aabb = p_transform_A.xform(p_shape_A->get_aabb()); + shape_aabb.grow_by(collision_margin); + + soft_body->query_aabb(shape_aabb, soft_body_query_callback, &query_cinfo); + } + + return (query_cinfo.contact_info.contact_count > 0); +} + struct _ConcaveCollisionInfo { const Transform *transform_A; const Shape3DSW *shape_A; @@ -199,6 +350,9 @@ bool CollisionSolver3DSW::solve_static(const Shape3DSW *p_shape_A, const Transfo if (type_B == PhysicsServer3D::SHAPE_RAY) { return false; } + if (type_B == PhysicsServer3D::SHAPE_SOFT_BODY) { + return false; + } if (swap) { return solve_static_plane(p_shape_B, p_transform_B, p_shape_A, p_transform_A, p_result_callback, p_userdata, true); @@ -217,6 +371,18 @@ bool CollisionSolver3DSW::solve_static(const Shape3DSW *p_shape_A, const Transfo return solve_ray(p_shape_A, p_transform_A, p_shape_B, p_transform_B, p_result_callback, p_userdata, false); } + } else if (type_B == PhysicsServer3D::SHAPE_SOFT_BODY) { + if (type_A == PhysicsServer3D::SHAPE_SOFT_BODY) { + // Soft Body / Soft Body not supported. + return false; + } + + if (swap) { + return solve_soft_body(p_shape_B, p_transform_B, p_shape_A, p_transform_A, p_result_callback, p_userdata, true); + } else { + return solve_soft_body(p_shape_A, p_transform_A, p_shape_B, p_transform_B, p_result_callback, p_userdata, false); + } + } else if (concave_B) { if (concave_A) { return false; @@ -265,8 +431,25 @@ bool CollisionSolver3DSW::solve_distance_plane(const Shape3DSW *p_shape_A, const static const int max_supports = 16; Vector3 supports[max_supports]; int support_count; + Shape3DSW::FeatureType support_type; + + p_shape_B->get_supports(p_transform_B.basis.xform_inv(-p.normal).normalized(), max_supports, supports, support_count, support_type); - p_shape_B->get_supports(p_transform_B.basis.xform_inv(-p.normal).normalized(), max_supports, supports, support_count); + if (support_type == Shape3DSW::FEATURE_CIRCLE) { + ERR_FAIL_COND_V(support_count != 3, false); + + Vector3 circle_pos = supports[0]; + Vector3 circle_axis_1 = supports[1] - circle_pos; + Vector3 circle_axis_2 = supports[2] - circle_pos; + + // Use 3 equidistant points on the circle. + for (int i = 0; i < 3; ++i) { + Vector3 vertex_pos = circle_pos; + vertex_pos += circle_axis_1 * Math::cos(2.0 * Math_PI * i / 3.0); + vertex_pos += circle_axis_2 * Math::sin(2.0 * Math_PI * i / 3.0); + supports[i] = vertex_pos; + } + } bool collided = false; Vector3 closest; diff --git a/servers/physics_3d/collision_solver_3d_sw.h b/servers/physics_3d/collision_solver_3d_sw.h index 81d87e9773..34ac2c6d3f 100644 --- a/servers/physics_3d/collision_solver_3d_sw.h +++ b/servers/physics_3d/collision_solver_3d_sw.h @@ -35,12 +35,16 @@ class CollisionSolver3DSW { public: - typedef void (*CallbackResult)(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata); + typedef void (*CallbackResult)(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata); private: + static bool soft_body_query_callback(uint32_t p_node_index, void *p_userdata); + static void soft_body_contact_callback(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata); + static void soft_body_concave_callback(void *p_userdata, Shape3DSW *p_convex); static void concave_callback(void *p_userdata, Shape3DSW *p_convex); static bool solve_static_plane(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CallbackResult p_result_callback, void *p_userdata, bool p_swap_result); static bool solve_ray(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CallbackResult p_result_callback, void *p_userdata, bool p_swap_result); + static bool solve_soft_body(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CallbackResult p_result_callback, void *p_userdata, bool p_swap_result); static bool solve_concave(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CallbackResult p_result_callback, void *p_userdata, bool p_swap_result, real_t p_margin_A = 0, real_t p_margin_B = 0); static void concave_distance_callback(void *p_userdata, Shape3DSW *p_convex); static bool solve_distance_plane(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, Vector3 &r_point_A, Vector3 &r_point_B); diff --git a/servers/physics_3d/gjk_epa.cpp b/servers/physics_3d/gjk_epa.cpp index dafd2feb8b..1a8c7f704f 100644 --- a/servers/physics_3d/gjk_epa.cpp +++ b/servers/physics_3d/gjk_epa.cpp @@ -64,7 +64,7 @@ GJK-EPA collision solver by Nathanael Presson, 2008 /* GJK */ #define GJK_MAX_ITERATIONS 128 -#define GJK_ACCURARY ((real_t)0.0001) +#define GJK_ACCURACY ((real_t)0.0001) #define GJK_MIN_DISTANCE ((real_t)0.0001) #define GJK_DUPLICATED_EPS ((real_t)0.0001) #define GJK_SIMPLEX2_EPS ((real_t)0.0) @@ -72,10 +72,13 @@ GJK-EPA collision solver by Nathanael Presson, 2008 #define GJK_SIMPLEX4_EPS ((real_t)0.0) /* EPA */ -#define EPA_MAX_VERTICES 64 +#define EPA_MAX_VERTICES 128 #define EPA_MAX_FACES (EPA_MAX_VERTICES*2) #define EPA_MAX_ITERATIONS 255 -#define EPA_ACCURACY ((real_t)0.0001) +// -- GODOT start -- +//#define EPA_ACCURACY ((real_t)0.0001) +#define EPA_ACCURACY ((real_t)0.00001) +// -- GODOT end -- #define EPA_FALLBACK (10*EPA_ACCURACY) #define EPA_PLANE_EPS ((real_t)0.00001) #define EPA_INSIDE_EPS ((real_t)0.01) @@ -107,26 +110,60 @@ struct MinkowskiDiff { Transform transform_A; Transform transform_B; + real_t margin_A = 0.0; + real_t margin_B = 0.0; + + Vector3 (*get_support)(const Shape3DSW*, const Vector3&, real_t); + + void Initialize(const Shape3DSW* shape0, const Transform& wtrs0, const real_t margin0, + const Shape3DSW* shape1, const Transform& wtrs1, const real_t margin1) { + m_shapes[0] = shape0; + m_shapes[1] = shape1; + transform_A = wtrs0; + transform_B = wtrs1; + margin_A = margin0; + margin_B = margin1; + + if ((margin0 > 0.0) || (margin1 > 0.0)) { + get_support = get_support_with_margin; + } else { + get_support = get_support_without_margin; + } + } + + static Vector3 get_support_without_margin(const Shape3DSW* p_shape, const Vector3& p_dir, real_t p_margin) { + return p_shape->get_support(p_dir.normalized()); + } + + static Vector3 get_support_with_margin(const Shape3DSW* p_shape, const Vector3& p_dir, real_t p_margin) { + Vector3 local_dir_norm = p_dir; + if (local_dir_norm.length_squared() < CMP_EPSILON2) { + local_dir_norm = Vector3(-1.0, -1.0, -1.0); + } + local_dir_norm.normalize(); + + return p_shape->get_support(local_dir_norm) + p_margin * local_dir_norm; + } + // i wonder how this could be sped up... if it can - _FORCE_INLINE_ Vector3 Support0 ( const Vector3& d ) const { - return transform_A.xform( m_shapes[0]->get_support( transform_A.basis.xform_inv(d).normalized() ) ); + _FORCE_INLINE_ Vector3 Support0(const Vector3& d) const { + return transform_A.xform(get_support(m_shapes[0], transform_A.basis.xform_inv(d), margin_A)); } - _FORCE_INLINE_ Vector3 Support1 ( const Vector3& d ) const { - return transform_B.xform( m_shapes[1]->get_support( transform_B.basis.xform_inv(d).normalized() ) ); + _FORCE_INLINE_ Vector3 Support1(const Vector3& d) const { + return transform_B.xform(get_support(m_shapes[1], transform_B.basis.xform_inv(d), margin_B)); } - _FORCE_INLINE_ Vector3 Support ( const Vector3& d ) const { - return ( Support0 ( d )-Support1 ( -d ) ); + _FORCE_INLINE_ Vector3 Support (const Vector3& d) const { + return (Support0(d) - Support1(-d)); } - _FORCE_INLINE_ Vector3 Support ( const Vector3& d,U index ) const - { - if ( index ) { - return ( Support1 ( d ) ); + _FORCE_INLINE_ Vector3 Support(const Vector3& d, U index) const { + if (index) { + return Support1(d); } else { - return ( Support0 ( d ) ); -} + return Support0(d); + } } }; @@ -237,7 +274,7 @@ struct GJK /* Check for termination */ const real_t omega=vec3_dot(m_ray,w)/rl; alpha=MAX(omega,alpha); - if(((rl-alpha)-(GJK_ACCURARY*rl))<=0) + if(((rl-alpha)-(GJK_ACCURACY*rl))<=0) {/* Return old simplex */ removevertice(m_simplices[m_current]); break; @@ -466,7 +503,7 @@ struct GJK if(ng&&(Math::abs(vl)>GJK_SIMPLEX4_EPS)) { real_t mindist=-1; - real_t subw[3]; + real_t subw[3] = {0.f, 0.f, 0.f}; U subm=0; for(U i=0;i<3;++i) { @@ -512,7 +549,6 @@ struct GJK { Vector3 n; real_t d; - real_t p; sSV* c[3]; sFace* f[3]; sFace* l[2]; @@ -661,8 +697,7 @@ struct GJK remove(m_hull,best); append(m_stock,best); best=findbest(); - if(best->p>=outer.p) { outer=*best; -} + outer=*best; } else { m_status=eStatus::InvalidHull;break; } } else { m_status=eStatus::AccuraryReached;break; } } else { m_status=eStatus::OutOfVertices;break; } @@ -688,24 +723,54 @@ struct GJK } } /* Fallback */ - m_status = eStatus::FallBack; - m_normal = -guess; - const real_t nl=m_normal.length(); - if(nl>0) { - m_normal = m_normal/nl; + m_status = eStatus::FallBack; + m_normal = -guess; + const real_t nl = m_normal.length(); + if (nl > 0) { + m_normal = m_normal/nl; } else { - m_normal = Vector3(1,0,0); -} + m_normal = Vector3(1,0,0); + } m_depth = 0; m_result.rank=1; m_result.c[0]=simplex.c[0]; m_result.p[0]=1; return(m_status); } + + bool getedgedist(sFace* face, sSV* a, sSV* b, real_t& dist) + { + const Vector3 ba = b->w - a->w; + const Vector3 n_ab = vec3_cross(ba, face->n); // Outward facing edge normal direction, on triangle plane + const real_t a_dot_nab = vec3_dot(a->w, n_ab); // Only care about the sign to determine inside/outside, so not normalization required + + if (a_dot_nab < 0) { + // Outside of edge a->b + const real_t ba_l2 = ba.length_squared(); + const real_t a_dot_ba = vec3_dot(a->w, ba); + const real_t b_dot_ba = vec3_dot(b->w, ba); + + if (a_dot_ba > 0) { + // Pick distance vertex a + dist = a->w.length(); + } else if (b_dot_ba < 0) { + // Pick distance vertex b + dist = b->w.length(); + } else { + // Pick distance to edge a->b + const real_t a_dot_b = vec3_dot(a->w, b->w); + dist = Math::sqrt(MAX((a->w.length_squared() * b->w.length_squared() - a_dot_b * a_dot_b) / ba_l2, 0.0)); + } + + return true; + } + + return false; + } + sFace* newface(sSV* a,sSV* b,sSV* c,bool forced) { - if(m_stock.root) - { + if (m_stock.root) { sFace* face=m_stock.root; remove(m_stock,face); append(m_hull,face); @@ -716,23 +781,23 @@ struct GJK face->n = vec3_cross(b->w-a->w,c->w-a->w); const real_t l=face->n.length(); const bool v=l>EPA_ACCURACY; - face->p = MIN(MIN( - vec3_dot(a->w,vec3_cross(face->n,a->w-b->w)), - vec3_dot(b->w,vec3_cross(face->n,b->w-c->w))), - vec3_dot(c->w,vec3_cross(face->n,c->w-a->w))) / - (v?l:1); - face->p = face->p>=-EPA_INSIDE_EPS?0:face->p; - if(v) - { - face->d = vec3_dot(a->w,face->n)/l; + if (v) { + if (!(getedgedist(face, a, b, face->d) || + getedgedist(face, b, c, face->d) || + getedgedist(face, c, a, face->d))) { + // Origin projects to the interior of the triangle + // Use distance to triangle plane + face->d = vec3_dot(a->w, face->n) / l; + } face->n /= l; - if(forced||(face->d>=-EPA_PLANE_EPS)) - { + if (forced||(face->d>=-EPA_PLANE_EPS)) { return(face); - } else { m_status=eStatus::NonConvex; -} - } else { m_status=eStatus::Degenerated; -} + } else { + m_status=eStatus::NonConvex; + } + } else { + m_status=eStatus::Degenerated; + } remove(m_hull,face); append(m_stock,face); return(nullptr); @@ -747,15 +812,13 @@ struct GJK { sFace* minf=m_hull.root; real_t mind=minf->d*minf->d; - real_t maxp=minf->p; for(sFace* f=minf->l[1];f;f=f->l[1]) { const real_t sqd=f->d*f->d; - if((f->p>=maxp)&&(sqd<mind)) + if(sqd<mind) { minf=f; mind=sqd; - maxp=f->p; } } return(minf); @@ -799,22 +862,17 @@ struct GJK }; // - static void Initialize( const Shape3DSW* shape0,const Transform& wtrs0, - const Shape3DSW* shape1,const Transform& wtrs1, + static void Initialize( const Shape3DSW* shape0, const Transform& wtrs0, real_t margin0, + const Shape3DSW* shape1, const Transform& wtrs1, real_t margin1, sResults& results, - tShape& shape, - bool withmargins) + tShape& shape) { /* Results */ - results.witnesses[0] = - results.witnesses[1] = Vector3(0,0,0); + results.witnesses[0] = Vector3(0,0,0); + results.witnesses[1] = Vector3(0,0,0); results.status = sResults::Separated; /* Shape */ - shape.m_shapes[0] = shape0; - shape.m_shapes[1] = shape1; - shape.transform_A = wtrs0; - shape.transform_B = wtrs1; - + shape.Initialize(shape0, wtrs0, margin0, shape1, wtrs1, margin1); } @@ -828,13 +886,15 @@ struct GJK // bool Distance( const Shape3DSW* shape0, const Transform& wtrs0, - const Shape3DSW* shape1, + real_t margin0, + const Shape3DSW* shape1, const Transform& wtrs1, + real_t margin1, const Vector3& guess, sResults& results) { tShape shape; - Initialize(shape0,wtrs0,shape1,wtrs1,results,shape,false); + Initialize(shape0, wtrs0, margin0, shape1, wtrs1, margin1, results, shape); GJK gjk; GJK::eStatus::_ gjk_status=gjk.Evaluate(shape,guess); if(gjk_status==GJK::eStatus::Valid) @@ -867,14 +927,16 @@ bool Distance( const Shape3DSW* shape0, // bool Penetration( const Shape3DSW* shape0, const Transform& wtrs0, - const Shape3DSW* shape1, + real_t margin0, + const Shape3DSW* shape1, const Transform& wtrs1, - const Vector3& guess, + real_t margin1, + const Vector3& guess, sResults& results ) { tShape shape; - Initialize(shape0,wtrs0,shape1,wtrs1,results,shape,false); + Initialize(shape0, wtrs0, margin0, shape1, wtrs1, margin1, results, shape); GJK gjk; GJK::eStatus::_ gjk_status=gjk.Evaluate(shape,-guess); switch(gjk_status) @@ -934,7 +996,7 @@ bool Penetration( const Shape3DSW* shape0, bool gjk_epa_calculate_distance(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, Vector3 &r_result_A, Vector3 &r_result_B) { GjkEpa2::sResults res; - if (GjkEpa2::Distance(p_shape_A, p_transform_A, p_shape_B, p_transform_B, p_transform_B.origin - p_transform_A.origin, res)) { + if (GjkEpa2::Distance(p_shape_A, p_transform_A, 0.0, p_shape_B, p_transform_B, 0.0, p_transform_B.origin - p_transform_A.origin, res)) { r_result_A = res.witnesses[0]; r_result_B = res.witnesses[1]; return true; @@ -943,15 +1005,15 @@ bool gjk_epa_calculate_distance(const Shape3DSW *p_shape_A, const Transform &p_t return false; } -bool gjk_epa_calculate_penetration(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CollisionSolver3DSW::CallbackResult p_result_callback, void *p_userdata, bool p_swap) { +bool gjk_epa_calculate_penetration(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CollisionSolver3DSW::CallbackResult p_result_callback, void *p_userdata, bool p_swap, real_t p_margin_A, real_t p_margin_B) { GjkEpa2::sResults res; - if (GjkEpa2::Penetration(p_shape_A, p_transform_A, p_shape_B, p_transform_B, p_transform_B.origin - p_transform_A.origin, res)) { + if (GjkEpa2::Penetration(p_shape_A, p_transform_A, p_margin_A, p_shape_B, p_transform_B, p_margin_B, p_transform_B.origin - p_transform_A.origin, res)) { if (p_result_callback) { if (p_swap) { - p_result_callback(res.witnesses[1], res.witnesses[0], p_userdata); + p_result_callback(res.witnesses[1], 0, res.witnesses[0], 0, p_userdata); } else { - p_result_callback(res.witnesses[0], res.witnesses[1], p_userdata); + p_result_callback(res.witnesses[0], 0, res.witnesses[1], 0, p_userdata); } } return true; diff --git a/servers/physics_3d/gjk_epa.h b/servers/physics_3d/gjk_epa.h index be3ba4e664..a7e2e1719e 100644 --- a/servers/physics_3d/gjk_epa.h +++ b/servers/physics_3d/gjk_epa.h @@ -34,7 +34,7 @@ #include "collision_solver_3d_sw.h" #include "shape_3d_sw.h" -bool gjk_epa_calculate_penetration(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CollisionSolver3DSW::CallbackResult p_result_callback, void *p_userdata, bool p_swap = false); +bool gjk_epa_calculate_penetration(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, CollisionSolver3DSW::CallbackResult p_result_callback, void *p_userdata, bool p_swap = false, real_t p_margin_A = 0.0, real_t p_margin_B = 0.0); bool gjk_epa_calculate_distance(const Shape3DSW *p_shape_A, const Transform &p_transform_A, const Shape3DSW *p_shape_B, const Transform &p_transform_B, Vector3 &r_result_A, Vector3 &r_result_B); #endif diff --git a/servers/physics_3d/joints/cone_twist_joint_3d_sw.cpp b/servers/physics_3d/joints/cone_twist_joint_3d_sw.cpp index 7b10257157..9c4493f4a2 100644 --- a/servers/physics_3d/joints/cone_twist_joint_3d_sw.cpp +++ b/servers/physics_3d/joints/cone_twist_joint_3d_sw.cpp @@ -92,9 +92,9 @@ ConeTwistJoint3DSW::ConeTwistJoint3DSW(Body3DSW *rbA, Body3DSW *rbB, const Trans m_rbAFrame = rbAFrame; m_rbBFrame = rbBFrame; - m_swingSpan1 = Math_PI / 4.0; - m_swingSpan2 = Math_PI / 4.0; - m_twistSpan = Math_PI * 2; + m_swingSpan1 = Math_TAU / 8.0; + m_swingSpan2 = Math_TAU / 8.0; + m_twistSpan = Math_TAU; m_biasFactor = 0.3f; m_relaxationFactor = 1.0f; diff --git a/servers/physics_3d/joints/cone_twist_joint_3d_sw.h b/servers/physics_3d/joints/cone_twist_joint_3d_sw.h index c38edc5737..4e4d4e7f0c 100644 --- a/servers/physics_3d/joints/cone_twist_joint_3d_sw.h +++ b/servers/physics_3d/joints/cone_twist_joint_3d_sw.h @@ -102,7 +102,7 @@ public: bool m_solveSwingLimit; public: - virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_CONE_TWIST; } + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_CONE_TWIST; } virtual bool setup(real_t p_timestep); virtual void solve(real_t p_timestep); diff --git a/servers/physics_3d/joints/generic_6dof_joint_3d_sw.cpp b/servers/physics_3d/joints/generic_6dof_joint_3d_sw.cpp index 7eb49e657b..13b389251f 100644 --- a/servers/physics_3d/joints/generic_6dof_joint_3d_sw.cpp +++ b/servers/physics_3d/joints/generic_6dof_joint_3d_sw.cpp @@ -132,7 +132,7 @@ real_t G6DOFRotationalLimitMotor3DSW::solveAngularLimits( real_t oldaccumImpulse = m_accumulatedImpulse; real_t sum = oldaccumImpulse + clippedMotorImpulse; - m_accumulatedImpulse = sum > hi ? real_t(0.) : sum < lo ? real_t(0.) : sum; + m_accumulatedImpulse = sum > hi ? real_t(0.) : (sum < lo ? real_t(0.) : sum); clippedMotorImpulse = m_accumulatedImpulse - oldaccumImpulse; @@ -201,7 +201,7 @@ real_t G6DOFTranslationalLimitMotor3DSW::solveLinearAxis( real_t oldNormalImpulse = m_accumulatedImpulse[limit_index]; real_t sum = oldNormalImpulse + normalImpulse; - m_accumulatedImpulse[limit_index] = sum > hi ? real_t(0.) : sum < lo ? real_t(0.) : sum; + m_accumulatedImpulse[limit_index] = sum > hi ? real_t(0.) : (sum < lo ? real_t(0.) : sum); normalImpulse = m_accumulatedImpulse[limit_index] - oldNormalImpulse; Vector3 impulse_vector = axis_normal_on_a * normalImpulse; diff --git a/servers/physics_3d/joints/generic_6dof_joint_3d_sw.h b/servers/physics_3d/joints/generic_6dof_joint_3d_sw.h index 2ae6fe85fa..d61a033231 100644 --- a/servers/physics_3d/joints/generic_6dof_joint_3d_sw.h +++ b/servers/physics_3d/joints/generic_6dof_joint_3d_sw.h @@ -234,7 +234,7 @@ protected: public: Generic6DOFJoint3DSW(Body3DSW *rbA, Body3DSW *rbB, const Transform &frameInA, const Transform &frameInB, bool useLinearReferenceFrameA); - virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_6DOF; } + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_6DOF; } virtual bool setup(real_t p_timestep); virtual void solve(real_t p_timestep); diff --git a/servers/physics_3d/joints/hinge_joint_3d_sw.h b/servers/physics_3d/joints/hinge_joint_3d_sw.h index 028a8b8c72..b6117aa0bc 100644 --- a/servers/physics_3d/joints/hinge_joint_3d_sw.h +++ b/servers/physics_3d/joints/hinge_joint_3d_sw.h @@ -96,7 +96,7 @@ class HingeJoint3DSW : public Joint3DSW { real_t m_appliedImpulse; public: - virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_HINGE; } + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_HINGE; } virtual bool setup(real_t p_step); virtual void solve(real_t p_step); diff --git a/servers/physics_3d/joints/pin_joint_3d_sw.h b/servers/physics_3d/joints/pin_joint_3d_sw.h index e28fbec6cd..1875983527 100644 --- a/servers/physics_3d/joints/pin_joint_3d_sw.h +++ b/servers/physics_3d/joints/pin_joint_3d_sw.h @@ -74,7 +74,7 @@ class PinJoint3DSW : public Joint3DSW { Vector3 m_pivotInB; public: - virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_PIN; } + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_PIN; } virtual bool setup(real_t p_step); virtual void solve(real_t p_step); diff --git a/servers/physics_3d/joints/slider_joint_3d_sw.h b/servers/physics_3d/joints/slider_joint_3d_sw.h index 196e60d19d..f52f6ace27 100644 --- a/servers/physics_3d/joints/slider_joint_3d_sw.h +++ b/servers/physics_3d/joints/slider_joint_3d_sw.h @@ -243,7 +243,7 @@ public: bool setup(real_t p_step); void solve(real_t p_step); - virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_SLIDER; } + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_SLIDER; } }; #endif // SLIDER_JOINT_SW_H diff --git a/servers/physics_3d/joints_3d_sw.h b/servers/physics_3d/joints_3d_sw.h index cad05b6702..225a71aca9 100644 --- a/servers/physics_3d/joints_3d_sw.h +++ b/servers/physics_3d/joints_3d_sw.h @@ -36,10 +36,28 @@ class Joint3DSW : public Constraint3DSW { public: - virtual PhysicsServer3D::JointType get_type() const = 0; + virtual bool setup(real_t p_step) { return false; } + virtual void solve(real_t p_step) {} + + void copy_settings_from(Joint3DSW *p_joint) { + set_self(p_joint->get_self()); + set_priority(p_joint->get_priority()); + disable_collisions_between_bodies(p_joint->is_disabled_collisions_between_bodies()); + } + + virtual PhysicsServer3D::JointType get_type() const { return PhysicsServer3D::JOINT_TYPE_MAX; } _FORCE_INLINE_ Joint3DSW(Body3DSW **p_body_ptr = nullptr, int p_body_count = 0) : Constraint3DSW(p_body_ptr, p_body_count) { } + + virtual ~Joint3DSW() { + for (int i = 0; i < get_body_count(); i++) { + Body3DSW *body = get_body_ptr()[i]; + if (body) { + body->remove_constraint(this); + } + } + } }; #endif // JOINTS_SW_H diff --git a/servers/physics_3d/physics_server_3d_sw.cpp b/servers/physics_3d/physics_server_3d_sw.cpp index 274de8411c..3d0063b0fa 100644 --- a/servers/physics_3d/physics_server_3d_sw.cpp +++ b/servers/physics_3d/physics_server_3d_sw.cpp @@ -43,47 +43,63 @@ #define FLUSH_QUERY_CHECK(m_object) \ ERR_FAIL_COND_MSG(m_object->get_space() && flushing_queries, "Can't change this state while flushing queries. Use call_deferred() or set_deferred() to change monitoring state instead."); -RID PhysicsServer3DSW::shape_create(ShapeType p_shape) { - Shape3DSW *shape = nullptr; - switch (p_shape) { - case SHAPE_PLANE: { - shape = memnew(PlaneShape3DSW); - } break; - case SHAPE_RAY: { - shape = memnew(RayShape3DSW); - } break; - case SHAPE_SPHERE: { - shape = memnew(SphereShape3DSW); - } break; - case SHAPE_BOX: { - shape = memnew(BoxShape3DSW); - } break; - case SHAPE_CAPSULE: { - shape = memnew(CapsuleShape3DSW); - } break; - case SHAPE_CYLINDER: { - ERR_FAIL_V_MSG(RID(), "CylinderShape3D is not supported in GodotPhysics3D. Please switch to Bullet in the Project Settings."); - } break; - case SHAPE_CONVEX_POLYGON: { - shape = memnew(ConvexPolygonShape3DSW); - } break; - case SHAPE_CONCAVE_POLYGON: { - shape = memnew(ConcavePolygonShape3DSW); - } break; - case SHAPE_HEIGHTMAP: { - shape = memnew(HeightMapShape3DSW); - } break; - case SHAPE_CUSTOM: { - ERR_FAIL_V(RID()); - - } break; - } - - RID id = shape_owner.make_rid(shape); - shape->set_self(id); - - return id; -}; +RID PhysicsServer3DSW::plane_shape_create() { + Shape3DSW *shape = memnew(PlaneShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::ray_shape_create() { + Shape3DSW *shape = memnew(RayShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::sphere_shape_create() { + Shape3DSW *shape = memnew(SphereShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::box_shape_create() { + Shape3DSW *shape = memnew(BoxShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::capsule_shape_create() { + Shape3DSW *shape = memnew(CapsuleShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::cylinder_shape_create() { + Shape3DSW *shape = memnew(CylinderShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::convex_polygon_shape_create() { + Shape3DSW *shape = memnew(ConvexPolygonShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::concave_polygon_shape_create() { + Shape3DSW *shape = memnew(ConcavePolygonShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::heightmap_shape_create() { + Shape3DSW *shape = memnew(HeightMapShape3DSW); + RID rid = shape_owner.make_rid(shape); + shape->set_self(rid); + return rid; +} +RID PhysicsServer3DSW::custom_shape_create() { + ERR_FAIL_V(RID()); +} void PhysicsServer3DSW::shape_set_data(RID p_shape, const Variant &p_data) { Shape3DSW *shape = shape_owner.getornull(p_shape); @@ -174,7 +190,7 @@ real_t PhysicsServer3DSW::space_get_param(RID p_space, SpaceParameter p_param) c PhysicsDirectSpaceState3D *PhysicsServer3DSW::space_get_direct_state(RID p_space) { Space3DSW *space = space_owner.getornull(p_space); ERR_FAIL_COND_V(!space, nullptr); - ERR_FAIL_COND_V_MSG(space->is_locked(), nullptr, "Space state is inaccessible right now, wait for iteration or physics process notification."); + ERR_FAIL_COND_V_MSG((using_threads && !doing_sync) || space->is_locked(), nullptr, "Space state is inaccessible right now, wait for iteration or physics process notification."); return space->get_direct_state(); } @@ -413,13 +429,6 @@ void PhysicsServer3DSW::area_set_ray_pickable(RID p_area, bool p_enable) { area->set_ray_pickable(p_enable); } -bool PhysicsServer3DSW::area_is_ray_pickable(RID p_area) const { - Area3DSW *area = area_owner.getornull(p_area); - ERR_FAIL_COND_V(!area, false); - - return area->is_ray_pickable(); -} - void PhysicsServer3DSW::area_set_area_monitor_callback(RID p_area, Object *p_receiver, const StringName &p_method) { Area3DSW *area = area_owner.getornull(p_area); ERR_FAIL_COND(!area); @@ -429,14 +438,8 @@ void PhysicsServer3DSW::area_set_area_monitor_callback(RID p_area, Object *p_rec /* BODY API */ -RID PhysicsServer3DSW::body_create(BodyMode p_mode, bool p_init_sleeping) { +RID PhysicsServer3DSW::body_create() { Body3DSW *body = memnew(Body3DSW); - if (p_mode != BODY_MODE_RIGID) { - body->set_mode(p_mode); - } - if (p_init_sleeping) { - body->set_state(BODY_STATE_SLEEPING, p_init_sleeping); - } RID rid = body_owner.make_rid(body); body->set_self(rid); return rid; @@ -608,9 +611,18 @@ uint32_t PhysicsServer3DSW::body_get_collision_mask(RID p_body) const { void PhysicsServer3DSW::body_attach_object_instance_id(RID p_body, ObjectID p_id) { Body3DSW *body = body_owner.getornull(p_body); - ERR_FAIL_COND(!body); + if (body) { + body->set_instance_id(p_id); + return; + } - body->set_instance_id(p_id); + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + if (soft_body) { + soft_body->set_instance_id(p_id); + return; + } + + ERR_FAIL_MSG("Invalid ID."); }; ObjectID PhysicsServer3DSW::body_get_object_instance_id(RID p_body) const { @@ -857,12 +869,6 @@ void PhysicsServer3DSW::body_set_ray_pickable(RID p_body, bool p_enable) { body->set_ray_pickable(p_enable); } -bool PhysicsServer3DSW::body_is_ray_pickable(RID p_body) const { - Body3DSW *body = body_owner.getornull(p_body); - ERR_FAIL_COND_V(!body, false); - return body->is_ray_pickable(); -} - bool PhysicsServer3DSW::body_test_motion(RID p_body, const Transform &p_from, const Vector3 &p_motion, bool p_infinite_inertia, MotionResult *r_result, bool p_exclude_raycast_shapes) { Body3DSW *body = body_owner.getornull(p_body); ERR_FAIL_COND_V(!body, false); @@ -874,7 +880,7 @@ bool PhysicsServer3DSW::body_test_motion(RID p_body, const Transform &p_from, co return body->get_space()->test_body_motion(body, p_from, p_motion, p_infinite_inertia, body->get_kinematic_margin(), r_result, p_exclude_raycast_shapes); } -int PhysicsServer3DSW::body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin) { +int PhysicsServer3DSW::body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin) { Body3DSW *body = body_owner.getornull(p_body); ERR_FAIL_COND_V(!body, false); ERR_FAIL_COND_V(!body->get_space(), false); @@ -886,6 +892,8 @@ int PhysicsServer3DSW::body_test_ray_separation(RID p_body, const Transform &p_t } PhysicsDirectBodyState3D *PhysicsServer3DSW::body_get_direct_state(RID p_body) { + ERR_FAIL_COND_V_MSG((using_threads && !doing_sync), nullptr, "Body state is inaccessible right now, wait for iteration or physics process notification."); + Body3DSW *body = body_owner.getornull(p_body); ERR_FAIL_COND_V(!body, nullptr); ERR_FAIL_COND_V_MSG(body->get_space()->is_locked(), nullptr, "Body state is inaccessible right now, wait for iteration or physics process notification."); @@ -894,32 +902,314 @@ PhysicsDirectBodyState3D *PhysicsServer3DSW::body_get_direct_state(RID p_body) { return direct_state; } +/* SOFT BODY */ + +RID PhysicsServer3DSW::soft_body_create() { + SoftBody3DSW *soft_body = memnew(SoftBody3DSW); + RID rid = soft_body_owner.make_rid(soft_body); + soft_body->set_self(rid); + return rid; +} + +void PhysicsServer3DSW::soft_body_update_rendering_server(RID p_body, RenderingServerHandler *p_rendering_server_handler) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->update_rendering_server(p_rendering_server_handler); +} + +void PhysicsServer3DSW::soft_body_set_space(RID p_body, RID p_space) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + Space3DSW *space = nullptr; + if (p_space.is_valid()) { + space = space_owner.getornull(p_space); + ERR_FAIL_COND(!space); + } + + if (soft_body->get_space() == space) { + return; + } + + soft_body->set_space(space); +} + +RID PhysicsServer3DSW::soft_body_get_space(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, RID()); + + Space3DSW *space = soft_body->get_space(); + if (!space) { + return RID(); + } + return space->get_self(); +} + +void PhysicsServer3DSW::soft_body_set_collision_layer(RID p_body, uint32_t p_layer) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_collision_layer(p_layer); +} + +uint32_t PhysicsServer3DSW::soft_body_get_collision_layer(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0); + + return soft_body->get_collision_layer(); +} + +void PhysicsServer3DSW::soft_body_set_collision_mask(RID p_body, uint32_t p_mask) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_collision_mask(p_mask); +} + +uint32_t PhysicsServer3DSW::soft_body_get_collision_mask(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0); + + return soft_body->get_collision_mask(); +} + +void PhysicsServer3DSW::soft_body_add_collision_exception(RID p_body, RID p_body_b) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->add_exception(p_body_b); +} + +void PhysicsServer3DSW::soft_body_remove_collision_exception(RID p_body, RID p_body_b) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->remove_exception(p_body_b); +} + +void PhysicsServer3DSW::soft_body_get_collision_exceptions(RID p_body, List<RID> *p_exceptions) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + for (int i = 0; i < soft_body->get_exceptions().size(); i++) { + p_exceptions->push_back(soft_body->get_exceptions()[i]); + } +} + +void PhysicsServer3DSW::soft_body_set_state(RID p_body, BodyState p_state, const Variant &p_variant) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_state(p_state, p_variant); +} + +Variant PhysicsServer3DSW::soft_body_get_state(RID p_body, BodyState p_state) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, Variant()); + + return soft_body->get_state(p_state); +} + +void PhysicsServer3DSW::soft_body_set_transform(RID p_body, const Transform &p_transform) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_state(BODY_STATE_TRANSFORM, p_transform); +} + +void PhysicsServer3DSW::soft_body_set_ray_pickable(RID p_body, bool p_enable) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_ray_pickable(p_enable); +} + +void PhysicsServer3DSW::soft_body_set_simulation_precision(RID p_body, int p_simulation_precision) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_iteration_count(p_simulation_precision); +} + +int PhysicsServer3DSW::soft_body_get_simulation_precision(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_iteration_count(); +} + +void PhysicsServer3DSW::soft_body_set_total_mass(RID p_body, real_t p_total_mass) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_total_mass(p_total_mass); +} + +real_t PhysicsServer3DSW::soft_body_get_total_mass(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_total_mass(); +} + +void PhysicsServer3DSW::soft_body_set_linear_stiffness(RID p_body, real_t p_stiffness) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_linear_stiffness(p_stiffness); +} + +real_t PhysicsServer3DSW::soft_body_get_linear_stiffness(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_linear_stiffness(); +} + +void PhysicsServer3DSW::soft_body_set_pressure_coefficient(RID p_body, real_t p_pressure_coefficient) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_pressure_coefficient(p_pressure_coefficient); +} + +real_t PhysicsServer3DSW::soft_body_get_pressure_coefficient(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_pressure_coefficient(); +} + +void PhysicsServer3DSW::soft_body_set_damping_coefficient(RID p_body, real_t p_damping_coefficient) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_damping_coefficient(p_damping_coefficient); +} + +real_t PhysicsServer3DSW::soft_body_get_damping_coefficient(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_damping_coefficient(); +} + +void PhysicsServer3DSW::soft_body_set_drag_coefficient(RID p_body, real_t p_drag_coefficient) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_drag_coefficient(p_drag_coefficient); +} + +real_t PhysicsServer3DSW::soft_body_get_drag_coefficient(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, 0.f); + + return soft_body->get_drag_coefficient(); +} + +void PhysicsServer3DSW::soft_body_set_mesh(RID p_body, const REF &p_mesh) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_mesh(p_mesh); +} + +AABB PhysicsServer3DSW::soft_body_get_bounds(RID p_body) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, AABB()); + + return soft_body->get_bounds(); +} + +void PhysicsServer3DSW::soft_body_move_point(RID p_body, int p_point_index, const Vector3 &p_global_position) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->set_vertex_position(p_point_index, p_global_position); +} + +Vector3 PhysicsServer3DSW::soft_body_get_point_global_position(RID p_body, int p_point_index) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, Vector3()); + + return soft_body->get_vertex_position(p_point_index); +} + +void PhysicsServer3DSW::soft_body_remove_all_pinned_points(RID p_body) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + soft_body->unpin_all_vertices(); +} + +void PhysicsServer3DSW::soft_body_pin_point(RID p_body, int p_point_index, bool p_pin) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND(!soft_body); + + if (p_pin) { + soft_body->pin_vertex(p_point_index); + } else { + soft_body->unpin_vertex(p_point_index); + } +} + +bool PhysicsServer3DSW::soft_body_is_point_pinned(RID p_body, int p_point_index) const { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_body); + ERR_FAIL_COND_V(!soft_body, false); + + return soft_body->is_vertex_pinned(p_point_index); +} + /* JOINT API */ -RID PhysicsServer3DSW::joint_create_pin(RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) { +RID PhysicsServer3DSW::joint_create() { + Joint3DSW *joint = memnew(Joint3DSW); + RID rid = joint_owner.make_rid(joint); + joint->set_self(rid); + return rid; +} + +void PhysicsServer3DSW::joint_clear(RID p_joint) { + Joint3DSW *joint = joint_owner.getornull(p_joint); + if (joint->get_type() != JOINT_TYPE_MAX) { + Joint3DSW *empty_joint = memnew(Joint3DSW); + empty_joint->copy_settings_from(joint); + + joint_owner.replace(p_joint, empty_joint); + memdelete(joint); + } +} + +void PhysicsServer3DSW::joint_make_pin(RID p_joint, RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); + + ERR_FAIL_COND(body_A == body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(PinJoint3DSW(body_A, p_local_A, body_B, p_local_B)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } void PhysicsServer3DSW::pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_PIN); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_PIN); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); pin_joint->set_param(p_param, p_value); } @@ -927,7 +1217,7 @@ void PhysicsServer3DSW::pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t PhysicsServer3DSW::pin_joint_get_param(RID p_joint, PinJointParam p_param) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, 0); - ERR_FAIL_COND_V(joint->get_type() != JOINT_PIN, 0); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_PIN, 0); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); return pin_joint->get_param(p_param); } @@ -935,7 +1225,7 @@ real_t PhysicsServer3DSW::pin_joint_get_param(RID p_joint, PinJointParam p_param void PhysicsServer3DSW::pin_joint_set_local_a(RID p_joint, const Vector3 &p_A) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_PIN); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_PIN); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); pin_joint->set_pos_a(p_A); } @@ -943,7 +1233,7 @@ void PhysicsServer3DSW::pin_joint_set_local_a(RID p_joint, const Vector3 &p_A) { Vector3 PhysicsServer3DSW::pin_joint_get_local_a(RID p_joint) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, Vector3()); - ERR_FAIL_COND_V(joint->get_type() != JOINT_PIN, Vector3()); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_PIN, Vector3()); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); return pin_joint->get_position_a(); } @@ -951,7 +1241,7 @@ Vector3 PhysicsServer3DSW::pin_joint_get_local_a(RID p_joint) const { void PhysicsServer3DSW::pin_joint_set_local_b(RID p_joint, const Vector3 &p_B) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_PIN); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_PIN); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); pin_joint->set_pos_b(p_B); } @@ -959,55 +1249,63 @@ void PhysicsServer3DSW::pin_joint_set_local_b(RID p_joint, const Vector3 &p_B) { Vector3 PhysicsServer3DSW::pin_joint_get_local_b(RID p_joint) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, Vector3()); - ERR_FAIL_COND_V(joint->get_type() != JOINT_PIN, Vector3()); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_PIN, Vector3()); PinJoint3DSW *pin_joint = static_cast<PinJoint3DSW *>(joint); return pin_joint->get_position_b(); } -RID PhysicsServer3DSW::joint_create_hinge(RID p_body_A, const Transform &p_frame_A, RID p_body_B, const Transform &p_frame_B) { +void PhysicsServer3DSW::joint_make_hinge(RID p_joint, RID p_body_A, const Transform &p_frame_A, RID p_body_B, const Transform &p_frame_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); + + ERR_FAIL_COND(body_A == body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(HingeJoint3DSW(body_A, body_B, p_frame_A, p_frame_B)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } -RID PhysicsServer3DSW::joint_create_hinge_simple(RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) { +void PhysicsServer3DSW::joint_make_hinge_simple(RID p_joint, RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + ERR_FAIL_COND(body_A == body_B); + + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(HingeJoint3DSW(body_A, body_B, p_pivot_A, p_pivot_B, p_axis_A, p_axis_B)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } void PhysicsServer3DSW::hinge_joint_set_param(RID p_joint, HingeJointParam p_param, real_t p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_HINGE); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_HINGE); HingeJoint3DSW *hinge_joint = static_cast<HingeJoint3DSW *>(joint); hinge_joint->set_param(p_param, p_value); } @@ -1015,7 +1313,7 @@ void PhysicsServer3DSW::hinge_joint_set_param(RID p_joint, HingeJointParam p_par real_t PhysicsServer3DSW::hinge_joint_get_param(RID p_joint, HingeJointParam p_param) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, 0); - ERR_FAIL_COND_V(joint->get_type() != JOINT_HINGE, 0); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_HINGE, 0); HingeJoint3DSW *hinge_joint = static_cast<HingeJoint3DSW *>(joint); return hinge_joint->get_param(p_param); } @@ -1023,7 +1321,7 @@ real_t PhysicsServer3DSW::hinge_joint_get_param(RID p_joint, HingeJointParam p_p void PhysicsServer3DSW::hinge_joint_set_flag(RID p_joint, HingeJointFlag p_flag, bool p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_HINGE); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_HINGE); HingeJoint3DSW *hinge_joint = static_cast<HingeJoint3DSW *>(joint); hinge_joint->set_flag(p_flag, p_value); } @@ -1031,7 +1329,7 @@ void PhysicsServer3DSW::hinge_joint_set_flag(RID p_joint, HingeJointFlag p_flag, bool PhysicsServer3DSW::hinge_joint_get_flag(RID p_joint, HingeJointFlag p_flag) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, false); - ERR_FAIL_COND_V(joint->get_type() != JOINT_HINGE, false); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_HINGE, false); HingeJoint3DSW *hinge_joint = static_cast<HingeJoint3DSW *>(joint); return hinge_joint->get_flag(p_flag); } @@ -1077,34 +1375,38 @@ bool PhysicsServer3DSW::joint_is_disabled_collisions_between_bodies(RID p_joint) PhysicsServer3DSW::JointType PhysicsServer3DSW::joint_get_type(RID p_joint) const { Joint3DSW *joint = joint_owner.getornull(p_joint); - ERR_FAIL_COND_V(!joint, JOINT_PIN); + ERR_FAIL_COND_V(!joint, JOINT_TYPE_PIN); return joint->get_type(); } -RID PhysicsServer3DSW::joint_create_slider(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { +void PhysicsServer3DSW::joint_make_slider(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); + + ERR_FAIL_COND(body_A == body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(SliderJoint3DSW(body_A, body_B, p_local_frame_A, p_local_frame_B)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } void PhysicsServer3DSW::slider_joint_set_param(RID p_joint, SliderJointParam p_param, real_t p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_SLIDER); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_SLIDER); SliderJoint3DSW *slider_joint = static_cast<SliderJoint3DSW *>(joint); slider_joint->set_param(p_param, p_value); } @@ -1112,35 +1414,39 @@ void PhysicsServer3DSW::slider_joint_set_param(RID p_joint, SliderJointParam p_p real_t PhysicsServer3DSW::slider_joint_get_param(RID p_joint, SliderJointParam p_param) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, 0); - ERR_FAIL_COND_V(joint->get_type() != JOINT_CONE_TWIST, 0); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_CONE_TWIST, 0); SliderJoint3DSW *slider_joint = static_cast<SliderJoint3DSW *>(joint); return slider_joint->get_param(p_param); } -RID PhysicsServer3DSW::joint_create_cone_twist(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { +void PhysicsServer3DSW::joint_make_cone_twist(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); + + ERR_FAIL_COND(body_A == body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(ConeTwistJoint3DSW(body_A, body_B, p_local_frame_A, p_local_frame_B)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } void PhysicsServer3DSW::cone_twist_joint_set_param(RID p_joint, ConeTwistJointParam p_param, real_t p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_CONE_TWIST); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_CONE_TWIST); ConeTwistJoint3DSW *cone_twist_joint = static_cast<ConeTwistJoint3DSW *>(joint); cone_twist_joint->set_param(p_param, p_value); } @@ -1148,43 +1454,47 @@ void PhysicsServer3DSW::cone_twist_joint_set_param(RID p_joint, ConeTwistJointPa real_t PhysicsServer3DSW::cone_twist_joint_get_param(RID p_joint, ConeTwistJointParam p_param) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, 0); - ERR_FAIL_COND_V(joint->get_type() != JOINT_CONE_TWIST, 0); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_CONE_TWIST, 0); ConeTwistJoint3DSW *cone_twist_joint = static_cast<ConeTwistJoint3DSW *>(joint); return cone_twist_joint->get_param(p_param); } -RID PhysicsServer3DSW::joint_create_generic_6dof(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { +void PhysicsServer3DSW::joint_make_generic_6dof(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) { Body3DSW *body_A = body_owner.getornull(p_body_A); - ERR_FAIL_COND_V(!body_A, RID()); + ERR_FAIL_COND(!body_A); if (!p_body_B.is_valid()) { - ERR_FAIL_COND_V(!body_A->get_space(), RID()); + ERR_FAIL_COND(!body_A->get_space()); p_body_B = body_A->get_space()->get_static_global_body(); } Body3DSW *body_B = body_owner.getornull(p_body_B); - ERR_FAIL_COND_V(!body_B, RID()); + ERR_FAIL_COND(!body_B); + + ERR_FAIL_COND(body_A == body_B); - ERR_FAIL_COND_V(body_A == body_B, RID()); + Joint3DSW *prev_joint = joint_owner.getornull(p_joint); + ERR_FAIL_COND(prev_joint == nullptr); Joint3DSW *joint = memnew(Generic6DOFJoint3DSW(body_A, body_B, p_local_frame_A, p_local_frame_B, true)); - RID rid = joint_owner.make_rid(joint); - joint->set_self(rid); - return rid; + + joint->copy_settings_from(prev_joint); + joint_owner.replace(p_joint, joint); + memdelete(prev_joint); } void PhysicsServer3DSW::generic_6dof_joint_set_param(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisParam p_param, real_t p_value) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_6DOF); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_6DOF); Generic6DOFJoint3DSW *generic_6dof_joint = static_cast<Generic6DOFJoint3DSW *>(joint); generic_6dof_joint->set_param(p_axis, p_param, p_value); } -real_t PhysicsServer3DSW::generic_6dof_joint_get_param(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisParam p_param) { +real_t PhysicsServer3DSW::generic_6dof_joint_get_param(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisParam p_param) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, 0); - ERR_FAIL_COND_V(joint->get_type() != JOINT_6DOF, 0); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_6DOF, 0); Generic6DOFJoint3DSW *generic_6dof_joint = static_cast<Generic6DOFJoint3DSW *>(joint); return generic_6dof_joint->get_param(p_axis, p_param); } @@ -1192,15 +1502,15 @@ real_t PhysicsServer3DSW::generic_6dof_joint_get_param(RID p_joint, Vector3::Axi void PhysicsServer3DSW::generic_6dof_joint_set_flag(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisFlag p_flag, bool p_enable) { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND(!joint); - ERR_FAIL_COND(joint->get_type() != JOINT_6DOF); + ERR_FAIL_COND(joint->get_type() != JOINT_TYPE_6DOF); Generic6DOFJoint3DSW *generic_6dof_joint = static_cast<Generic6DOFJoint3DSW *>(joint); generic_6dof_joint->set_flag(p_axis, p_flag, p_enable); } -bool PhysicsServer3DSW::generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisFlag p_flag) { +bool PhysicsServer3DSW::generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis p_axis, G6DOFJointAxisFlag p_flag) const { Joint3DSW *joint = joint_owner.getornull(p_joint); ERR_FAIL_COND_V(!joint, false); - ERR_FAIL_COND_V(joint->get_type() != JOINT_6DOF, false); + ERR_FAIL_COND_V(joint->get_type() != JOINT_TYPE_6DOF, false); Generic6DOFJoint3DSW *generic_6dof_joint = static_cast<Generic6DOFJoint3DSW *>(joint); return generic_6dof_joint->get_flag(p_axis, p_flag); } @@ -1237,7 +1547,13 @@ void PhysicsServer3DSW::free(RID p_rid) { body_owner.free(p_rid); memdelete(body); + } else if (soft_body_owner.owns(p_rid)) { + SoftBody3DSW *soft_body = soft_body_owner.getornull(p_rid); + + soft_body->set_space(nullptr); + soft_body_owner.free(p_rid); + memdelete(soft_body); } else if (area_owner.owns(p_rid)) { Area3DSW *area = area_owner.getornull(p_rid); @@ -1271,9 +1587,6 @@ void PhysicsServer3DSW::free(RID p_rid) { } else if (joint_owner.owns(p_rid)) { Joint3DSW *joint = joint_owner.getornull(p_rid); - for (int i = 0; i < joint->get_body_count(); i++) { - joint->get_body_ptr()[i]->remove_constraint(joint); - } joint_owner.free(p_rid); memdelete(joint); @@ -1317,6 +1630,10 @@ void PhysicsServer3DSW::step(real_t p_step) { #endif } +void PhysicsServer3DSW::sync() { + doing_sync = true; +}; + void PhysicsServer3DSW::flush_queries() { #ifndef _3D_DISABLED @@ -1370,6 +1687,10 @@ void PhysicsServer3DSW::flush_queries() { #endif }; +void PhysicsServer3DSW::end_sync() { + doing_sync = false; +}; + void PhysicsServer3DSW::finish() { memdelete(stepper); memdelete(direct_state); @@ -1398,7 +1719,7 @@ void PhysicsServer3DSW::_update_shapes() { } } -void PhysicsServer3DSW::_shape_col_cbk(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata) { +void PhysicsServer3DSW::_shape_col_cbk(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { CollCbkData *cbk = (CollCbkData *)p_userdata; if (cbk->max == 0) { @@ -1431,14 +1752,15 @@ void PhysicsServer3DSW::_shape_col_cbk(const Vector3 &p_point_A, const Vector3 & } } -PhysicsServer3DSW *PhysicsServer3DSW::singleton = nullptr; -PhysicsServer3DSW::PhysicsServer3DSW() { - singleton = this; +PhysicsServer3DSW *PhysicsServer3DSW::singletonsw = nullptr; +PhysicsServer3DSW::PhysicsServer3DSW(bool p_using_threads) { + singletonsw = this; BroadPhase3DSW::create_func = BroadPhaseOctree::_create; island_count = 0; active_objects = 0; collision_pairs = 0; - + using_threads = p_using_threads; active = true; flushing_queries = false; + doing_sync = false; }; diff --git a/servers/physics_3d/physics_server_3d_sw.h b/servers/physics_3d/physics_server_3d_sw.h index 9b6b113677..f92652bfad 100644 --- a/servers/physics_3d/physics_server_3d_sw.h +++ b/servers/physics_3d/physics_server_3d_sw.h @@ -50,6 +50,8 @@ class PhysicsServer3DSW : public PhysicsServer3D { int active_objects; int collision_pairs; + bool using_threads; + bool doing_sync; bool flushing_queries; Step3DSW *stepper; @@ -57,29 +59,40 @@ class PhysicsServer3DSW : public PhysicsServer3D { PhysicsDirectBodyState3DSW *direct_state; - mutable RID_PtrOwner<Shape3DSW> shape_owner; - mutable RID_PtrOwner<Space3DSW> space_owner; - mutable RID_PtrOwner<Area3DSW> area_owner; - mutable RID_PtrOwner<Body3DSW> body_owner; - mutable RID_PtrOwner<Joint3DSW> joint_owner; + mutable RID_PtrOwner<Shape3DSW, true> shape_owner; + mutable RID_PtrOwner<Space3DSW, true> space_owner; + mutable RID_PtrOwner<Area3DSW, true> area_owner; + mutable RID_PtrOwner<Body3DSW, true> body_owner; + mutable RID_PtrOwner<SoftBody3DSW, true> soft_body_owner; + mutable RID_PtrOwner<Joint3DSW, true> joint_owner; //void _clear_query(QuerySW *p_query); friend class CollisionObject3DSW; SelfList<CollisionObject3DSW>::List pending_shape_update_list; void _update_shapes(); -public: - static PhysicsServer3DSW *singleton; + static PhysicsServer3DSW *singletonsw; +public: struct CollCbkData { int max; int amount; Vector3 *ptr; }; - static void _shape_col_cbk(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata); + static void _shape_col_cbk(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata); + + virtual RID plane_shape_create() override; + virtual RID ray_shape_create() override; + virtual RID sphere_shape_create() override; + virtual RID box_shape_create() override; + virtual RID capsule_shape_create() override; + virtual RID cylinder_shape_create() override; + virtual RID convex_polygon_shape_create() override; + virtual RID concave_polygon_shape_create() override; + virtual RID heightmap_shape_create() override; + virtual RID custom_shape_create() override; - virtual RID shape_create(ShapeType p_shape) override; virtual void shape_set_data(RID p_shape, const Variant &p_data) override; virtual void shape_set_custom_solver_bias(RID p_shape, real_t p_bias) override; @@ -140,7 +153,6 @@ public: virtual Transform area_get_transform(RID p_area) const override; virtual void area_set_ray_pickable(RID p_area, bool p_enable) override; - virtual bool area_is_ray_pickable(RID p_area) const override; virtual void area_set_collision_mask(RID p_area, uint32_t p_mask) override; virtual void area_set_collision_layer(RID p_area, uint32_t p_layer) override; @@ -153,7 +165,7 @@ public: /* BODY API */ // create a body of a given type - virtual RID body_create(BodyMode p_mode = BODY_MODE_RIGID, bool p_init_sleeping = false) override; + virtual RID body_create() override; virtual void body_set_space(RID p_body, RID p_space) override; virtual RID body_get_space(RID p_body) const override; @@ -232,83 +244,75 @@ public: virtual void body_set_force_integration_callback(RID p_body, Object *p_receiver, const StringName &p_method, const Variant &p_udata = Variant()) override; virtual void body_set_ray_pickable(RID p_body, bool p_enable) override; - virtual bool body_is_ray_pickable(RID p_body) const override; virtual bool body_test_motion(RID p_body, const Transform &p_from, const Vector3 &p_motion, bool p_infinite_inertia, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) override; - virtual int body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin = 0.001) override; + virtual int body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) override; // this function only works on physics process, errors and returns null otherwise virtual PhysicsDirectBodyState3D *body_get_direct_state(RID p_body) override; /* SOFT BODY */ - virtual RID soft_body_create(bool p_init_sleeping = false) override { return RID(); } + virtual RID soft_body_create() override; - virtual void soft_body_update_rendering_server(RID p_body, class SoftBodyRenderingServerHandler *p_rendering_server_handler) override {} + virtual void soft_body_update_rendering_server(RID p_body, RenderingServerHandler *p_rendering_server_handler) override; - virtual void soft_body_set_space(RID p_body, RID p_space) override {} - virtual RID soft_body_get_space(RID p_body) const override { return RID(); } + virtual void soft_body_set_space(RID p_body, RID p_space) override; + virtual RID soft_body_get_space(RID p_body) const override; - virtual void soft_body_set_collision_layer(RID p_body, uint32_t p_layer) override {} - virtual uint32_t soft_body_get_collision_layer(RID p_body) const override { return 0; } + virtual void soft_body_set_collision_layer(RID p_body, uint32_t p_layer) override; + virtual uint32_t soft_body_get_collision_layer(RID p_body) const override; - virtual void soft_body_set_collision_mask(RID p_body, uint32_t p_mask) override {} - virtual uint32_t soft_body_get_collision_mask(RID p_body) const override { return 0; } + virtual void soft_body_set_collision_mask(RID p_body, uint32_t p_mask) override; + virtual uint32_t soft_body_get_collision_mask(RID p_body) const override; - virtual void soft_body_add_collision_exception(RID p_body, RID p_body_b) override {} - virtual void soft_body_remove_collision_exception(RID p_body, RID p_body_b) override {} - virtual void soft_body_get_collision_exceptions(RID p_body, List<RID> *p_exceptions) override {} + virtual void soft_body_add_collision_exception(RID p_body, RID p_body_b) override; + virtual void soft_body_remove_collision_exception(RID p_body, RID p_body_b) override; + virtual void soft_body_get_collision_exceptions(RID p_body, List<RID> *p_exceptions) override; - virtual void soft_body_set_state(RID p_body, BodyState p_state, const Variant &p_variant) override {} - virtual Variant soft_body_get_state(RID p_body, BodyState p_state) const override { return Variant(); } + virtual void soft_body_set_state(RID p_body, BodyState p_state, const Variant &p_variant) override; + virtual Variant soft_body_get_state(RID p_body, BodyState p_state) const override; - virtual void soft_body_set_transform(RID p_body, const Transform &p_transform) override {} - virtual Vector3 soft_body_get_vertex_position(RID p_body, int vertex_index) const override { return Vector3(); } + virtual void soft_body_set_transform(RID p_body, const Transform &p_transform) override; - virtual void soft_body_set_ray_pickable(RID p_body, bool p_enable) override {} - virtual bool soft_body_is_ray_pickable(RID p_body) const override { return false; } + virtual void soft_body_set_ray_pickable(RID p_body, bool p_enable) override; - virtual void soft_body_set_simulation_precision(RID p_body, int p_simulation_precision) override {} - virtual int soft_body_get_simulation_precision(RID p_body) override { return 0; } + virtual void soft_body_set_simulation_precision(RID p_body, int p_simulation_precision) override; + virtual int soft_body_get_simulation_precision(RID p_body) const override; - virtual void soft_body_set_total_mass(RID p_body, real_t p_total_mass) override {} - virtual real_t soft_body_get_total_mass(RID p_body) override { return 0.; } + virtual void soft_body_set_total_mass(RID p_body, real_t p_total_mass) override; + virtual real_t soft_body_get_total_mass(RID p_body) const override; - virtual void soft_body_set_linear_stiffness(RID p_body, real_t p_stiffness) override {} - virtual real_t soft_body_get_linear_stiffness(RID p_body) override { return 0.; } + virtual void soft_body_set_linear_stiffness(RID p_body, real_t p_stiffness) override; + virtual real_t soft_body_get_linear_stiffness(RID p_body) const override; - virtual void soft_body_set_areaAngular_stiffness(RID p_body, real_t p_stiffness) override {} - virtual real_t soft_body_get_areaAngular_stiffness(RID p_body) override { return 0.; } + virtual void soft_body_set_pressure_coefficient(RID p_body, real_t p_pressure_coefficient) override; + virtual real_t soft_body_get_pressure_coefficient(RID p_body) const override; - virtual void soft_body_set_volume_stiffness(RID p_body, real_t p_stiffness) override {} - virtual real_t soft_body_get_volume_stiffness(RID p_body) override { return 0.; } + virtual void soft_body_set_damping_coefficient(RID p_body, real_t p_damping_coefficient) override; + virtual real_t soft_body_get_damping_coefficient(RID p_body) const override; - virtual void soft_body_set_pressure_coefficient(RID p_body, real_t p_pressure_coefficient) override {} - virtual real_t soft_body_get_pressure_coefficient(RID p_body) override { return 0.; } + virtual void soft_body_set_drag_coefficient(RID p_body, real_t p_drag_coefficient) override; + virtual real_t soft_body_get_drag_coefficient(RID p_body) const override; - virtual void soft_body_set_pose_matching_coefficient(RID p_body, real_t p_pose_matching_coefficient) override {} - virtual real_t soft_body_get_pose_matching_coefficient(RID p_body) override { return 0.; } + virtual void soft_body_set_mesh(RID p_body, const REF &p_mesh) override; - virtual void soft_body_set_damping_coefficient(RID p_body, real_t p_damping_coefficient) override {} - virtual real_t soft_body_get_damping_coefficient(RID p_body) override { return 0.; } + virtual AABB soft_body_get_bounds(RID p_body) const override; - virtual void soft_body_set_drag_coefficient(RID p_body, real_t p_drag_coefficient) override {} - virtual real_t soft_body_get_drag_coefficient(RID p_body) override { return 0.; } + virtual void soft_body_move_point(RID p_body, int p_point_index, const Vector3 &p_global_position) override; + virtual Vector3 soft_body_get_point_global_position(RID p_body, int p_point_index) const override; - virtual void soft_body_set_mesh(RID p_body, const REF &p_mesh) override {} + virtual void soft_body_remove_all_pinned_points(RID p_body) override; + virtual void soft_body_pin_point(RID p_body, int p_point_index, bool p_pin) override; + virtual bool soft_body_is_point_pinned(RID p_body, int p_point_index) const override; - virtual void soft_body_move_point(RID p_body, int p_point_index, const Vector3 &p_global_position) override {} - virtual Vector3 soft_body_get_point_global_position(RID p_body, int p_point_index) override { return Vector3(); } - - virtual Vector3 soft_body_get_point_offset(RID p_body, int p_point_index) const override { return Vector3(); } + /* JOINT API */ - virtual void soft_body_remove_all_pinned_points(RID p_body) override {} - virtual void soft_body_pin_point(RID p_body, int p_point_index, bool p_pin) override {} - virtual bool soft_body_is_point_pinned(RID p_body, int p_point_index) override { return false; } + virtual RID joint_create() override; - /* JOINT API */ + virtual void joint_clear(RID p_joint) override; //resets type - virtual RID joint_create_pin(RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) override; + virtual void joint_make_pin(RID p_joint, RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) override; virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) override; virtual real_t pin_joint_get_param(RID p_joint, PinJointParam p_param) const override; @@ -319,8 +323,8 @@ public: virtual void pin_joint_set_local_b(RID p_joint, const Vector3 &p_B) override; virtual Vector3 pin_joint_get_local_b(RID p_joint) const override; - virtual RID joint_create_hinge(RID p_body_A, const Transform &p_frame_A, RID p_body_B, const Transform &p_frame_B) override; - virtual RID joint_create_hinge_simple(RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) override; + virtual void joint_make_hinge(RID p_joint, RID p_body_A, const Transform &p_frame_A, RID p_body_B, const Transform &p_frame_B) override; + virtual void joint_make_hinge_simple(RID p_joint, RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) override; virtual void hinge_joint_set_param(RID p_joint, HingeJointParam p_param, real_t p_value) override; virtual real_t hinge_joint_get_param(RID p_joint, HingeJointParam p_param) const override; @@ -328,23 +332,23 @@ public: virtual void hinge_joint_set_flag(RID p_joint, HingeJointFlag p_flag, bool p_value) override; virtual bool hinge_joint_get_flag(RID p_joint, HingeJointFlag p_flag) const override; - virtual RID joint_create_slider(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A + virtual void joint_make_slider(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A virtual void slider_joint_set_param(RID p_joint, SliderJointParam p_param, real_t p_value) override; virtual real_t slider_joint_get_param(RID p_joint, SliderJointParam p_param) const override; - virtual RID joint_create_cone_twist(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A + virtual void joint_make_cone_twist(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A virtual void cone_twist_joint_set_param(RID p_joint, ConeTwistJointParam p_param, real_t p_value) override; virtual real_t cone_twist_joint_get_param(RID p_joint, ConeTwistJointParam p_param) const override; - virtual RID joint_create_generic_6dof(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A + virtual void joint_make_generic_6dof(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) override; //reference frame is A virtual void generic_6dof_joint_set_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param, real_t p_value) override; - virtual real_t generic_6dof_joint_get_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param) override; + virtual real_t generic_6dof_joint_get_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param) const override; virtual void generic_6dof_joint_set_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag, bool p_enable) override; - virtual bool generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag) override; + virtual bool generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag) const override; virtual JointType joint_get_type(RID p_joint) const override; @@ -361,14 +365,16 @@ public: virtual void set_active(bool p_active) override; virtual void init() override; virtual void step(real_t p_step) override; + virtual void sync() override; virtual void flush_queries() override; + virtual void end_sync() override; virtual void finish() override; virtual bool is_flushing_queries() const override { return flushing_queries; } int get_process_info(ProcessInfo p_info) override; - PhysicsServer3DSW(); + PhysicsServer3DSW(bool p_using_threads = false); ~PhysicsServer3DSW() {} }; diff --git a/servers/physics_3d/physics_server_3d_wrap_mt.cpp b/servers/physics_3d/physics_server_3d_wrap_mt.cpp new file mode 100644 index 0000000000..f73f67a756 --- /dev/null +++ b/servers/physics_3d/physics_server_3d_wrap_mt.cpp @@ -0,0 +1,140 @@ +/*************************************************************************/ +/* physics_server_3d_wrap_mt.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "physics_server_3d_wrap_mt.h" + +#include "core/os/os.h" + +void PhysicsServer3DWrapMT::thread_exit() { + exit = true; +} + +void PhysicsServer3DWrapMT::thread_step(real_t p_delta) { + physics_3d_server->step(p_delta); + step_sem.post(); +} + +void PhysicsServer3DWrapMT::_thread_callback(void *_instance) { + PhysicsServer3DWrapMT *vsmt = reinterpret_cast<PhysicsServer3DWrapMT *>(_instance); + + vsmt->thread_loop(); +} + +void PhysicsServer3DWrapMT::thread_loop() { + server_thread = Thread::get_caller_id(); + + physics_3d_server->init(); + + exit = false; + step_thread_up = true; + while (!exit) { + // flush commands one by one, until exit is requested + command_queue.wait_and_flush_one(); + } + + command_queue.flush_all(); // flush all + + physics_3d_server->finish(); +} + +/* EVENT QUEUING */ + +void PhysicsServer3DWrapMT::step(real_t p_step) { + if (create_thread) { + command_queue.push(this, &PhysicsServer3DWrapMT::thread_step, p_step); + } else { + command_queue.flush_all(); //flush all pending from other threads + physics_3d_server->step(p_step); + } +} + +void PhysicsServer3DWrapMT::sync() { + if (create_thread) { + if (first_frame) { + first_frame = false; + } else { + step_sem.wait(); //must not wait if a step was not issued + } + } + physics_3d_server->sync(); +} + +void PhysicsServer3DWrapMT::flush_queries() { + physics_3d_server->flush_queries(); +} + +void PhysicsServer3DWrapMT::end_sync() { + physics_3d_server->end_sync(); +} + +void PhysicsServer3DWrapMT::init() { + if (create_thread) { + //OS::get_singleton()->release_rendering_thread(); + thread.start(_thread_callback, this); + while (!step_thread_up) { + OS::get_singleton()->delay_usec(1000); + } + } else { + physics_3d_server->init(); + } +} + +void PhysicsServer3DWrapMT::finish() { + if (thread.is_started()) { + command_queue.push(this, &PhysicsServer3DWrapMT::thread_exit); + thread.wait_to_finish(); + } else { + physics_3d_server->finish(); + } +} + +PhysicsServer3DWrapMT::PhysicsServer3DWrapMT(PhysicsServer3D *p_contained, bool p_create_thread) : + command_queue(p_create_thread) { + physics_3d_server = p_contained; + create_thread = p_create_thread; + step_pending = 0; + step_thread_up = false; + + pool_max_size = GLOBAL_GET("memory/limits/multithreaded_server/rid_pool_prealloc"); + + if (!p_create_thread) { + server_thread = Thread::get_caller_id(); + } else { + server_thread = 0; + } + + main_thread = Thread::get_caller_id(); + first_frame = true; +} + +PhysicsServer3DWrapMT::~PhysicsServer3DWrapMT() { + memdelete(physics_3d_server); + //finish(); +} diff --git a/servers/physics_3d/physics_server_3d_wrap_mt.h b/servers/physics_3d/physics_server_3d_wrap_mt.h new file mode 100644 index 0000000000..49ae60db92 --- /dev/null +++ b/servers/physics_3d/physics_server_3d_wrap_mt.h @@ -0,0 +1,413 @@ +/*************************************************************************/ +/* physics_server_3d_wrap_mt.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef PHYSICS3DSERVERWRAPMT_H +#define PHYSICS3DSERVERWRAPMT_H + +#include "core/config/project_settings.h" +#include "core/os/thread.h" +#include "core/templates/command_queue_mt.h" +#include "servers/physics_server_3d.h" + +#ifdef DEBUG_SYNC +#define SYNC_DEBUG print_line("sync on: " + String(__FUNCTION__)); +#else +#define SYNC_DEBUG +#endif + +class PhysicsServer3DWrapMT : public PhysicsServer3D { + mutable PhysicsServer3D *physics_3d_server; + + mutable CommandQueueMT command_queue; + + static void _thread_callback(void *_instance); + void thread_loop(); + + Thread::ID server_thread; + Thread::ID main_thread; + volatile bool exit = false; + Thread thread; + volatile bool step_thread_up = false; + bool create_thread = false; + + Semaphore step_sem; + int step_pending; + void thread_step(real_t p_delta); + void thread_flush(); + + void thread_exit(); + + bool first_frame = true; + + Mutex alloc_mutex; + int pool_max_size = 0; + +public: +#define ServerName PhysicsServer3D +#define ServerNameWrapMT PhysicsServer3DWrapMT +#define server_name physics_3d_server +#define WRITE_ACTION + +#include "servers/server_wrap_mt_common.h" + + //FUNC1RID(shape,ShapeType); todo fix + FUNCRID(plane_shape) + FUNCRID(ray_shape) + FUNCRID(sphere_shape) + FUNCRID(box_shape) + FUNCRID(capsule_shape) + FUNCRID(cylinder_shape) + FUNCRID(convex_polygon_shape) + FUNCRID(concave_polygon_shape) + FUNCRID(heightmap_shape) + FUNCRID(custom_shape) + + FUNC2(shape_set_data, RID, const Variant &); + FUNC2(shape_set_custom_solver_bias, RID, real_t); + + FUNC2(shape_set_margin, RID, real_t) + FUNC1RC(real_t, shape_get_margin, RID) + + FUNC1RC(ShapeType, shape_get_type, RID); + FUNC1RC(Variant, shape_get_data, RID); + FUNC1RC(real_t, shape_get_custom_solver_bias, RID); +#if 0 + //these work well, but should be used from the main thread only + bool shape_collide(RID p_shape_A, const Transform &p_xform_A, const Vector3 &p_motion_A, RID p_shape_B, const Transform &p_xform_B, const Vector3 &p_motion_B, Vector3 *r_results, int p_result_max, int &r_result_count) { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); + return physics_3d_server->shape_collide(p_shape_A, p_xform_A, p_motion_A, p_shape_B, p_xform_B, p_motion_B, r_results, p_result_max, r_result_count); + } +#endif + /* SPACE API */ + + FUNCRID(space); + FUNC2(space_set_active, RID, bool); + FUNC1RC(bool, space_is_active, RID); + + FUNC3(space_set_param, RID, SpaceParameter, real_t); + FUNC2RC(real_t, space_get_param, RID, SpaceParameter); + + // this function only works on physics process, errors and returns null otherwise + PhysicsDirectSpaceState3D *space_get_direct_state(RID p_space) override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), nullptr); + return physics_3d_server->space_get_direct_state(p_space); + } + + FUNC2(space_set_debug_contacts, RID, int); + virtual Vector<Vector3> space_get_contacts(RID p_space) const override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), Vector<Vector3>()); + return physics_3d_server->space_get_contacts(p_space); + } + + virtual int space_get_contact_count(RID p_space) const override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), 0); + return physics_3d_server->space_get_contact_count(p_space); + } + + /* AREA API */ + + //FUNC0RID(area); + FUNCRID(area); + + FUNC2(area_set_space, RID, RID); + FUNC1RC(RID, area_get_space, RID); + + FUNC2(area_set_space_override_mode, RID, AreaSpaceOverrideMode); + FUNC1RC(AreaSpaceOverrideMode, area_get_space_override_mode, RID); + + FUNC4(area_add_shape, RID, RID, const Transform &, bool); + FUNC3(area_set_shape, RID, int, RID); + FUNC3(area_set_shape_transform, RID, int, const Transform &); + FUNC3(area_set_shape_disabled, RID, int, bool); + + FUNC1RC(int, area_get_shape_count, RID); + FUNC2RC(RID, area_get_shape, RID, int); + FUNC2RC(Transform, area_get_shape_transform, RID, int); + FUNC2(area_remove_shape, RID, int); + FUNC1(area_clear_shapes, RID); + + FUNC2(area_attach_object_instance_id, RID, ObjectID); + FUNC1RC(ObjectID, area_get_object_instance_id, RID); + + FUNC3(area_set_param, RID, AreaParameter, const Variant &); + FUNC2(area_set_transform, RID, const Transform &); + + FUNC2RC(Variant, area_get_param, RID, AreaParameter); + FUNC1RC(Transform, area_get_transform, RID); + + FUNC2(area_set_collision_mask, RID, uint32_t); + FUNC2(area_set_collision_layer, RID, uint32_t); + + FUNC2(area_set_monitorable, RID, bool); + FUNC2(area_set_ray_pickable, RID, bool); + + FUNC3(area_set_monitor_callback, RID, Object *, const StringName &); + FUNC3(area_set_area_monitor_callback, RID, Object *, const StringName &); + + /* BODY API */ + + //FUNC2RID(body,BodyMode,bool); + FUNCRID(body) + + FUNC2(body_set_space, RID, RID); + FUNC1RC(RID, body_get_space, RID); + + FUNC2(body_set_mode, RID, BodyMode); + FUNC1RC(BodyMode, body_get_mode, RID); + + FUNC4(body_add_shape, RID, RID, const Transform &, bool); + FUNC3(body_set_shape, RID, int, RID); + FUNC3(body_set_shape_transform, RID, int, const Transform &); + + FUNC1RC(int, body_get_shape_count, RID); + FUNC2RC(Transform, body_get_shape_transform, RID, int); + FUNC2RC(RID, body_get_shape, RID, int); + + FUNC3(body_set_shape_disabled, RID, int, bool); + + FUNC2(body_remove_shape, RID, int); + FUNC1(body_clear_shapes, RID); + + FUNC2(body_attach_object_instance_id, RID, ObjectID); + FUNC1RC(ObjectID, body_get_object_instance_id, RID); + + FUNC2(body_set_enable_continuous_collision_detection, RID, bool); + FUNC1RC(bool, body_is_continuous_collision_detection_enabled, RID); + + FUNC2(body_set_collision_layer, RID, uint32_t); + FUNC1RC(uint32_t, body_get_collision_layer, RID); + + FUNC2(body_set_collision_mask, RID, uint32_t); + FUNC1RC(uint32_t, body_get_collision_mask, RID); + + FUNC2(body_set_user_flags, RID, uint32_t); + FUNC1RC(uint32_t, body_get_user_flags, RID); + + FUNC3(body_set_param, RID, BodyParameter, real_t); + FUNC2RC(real_t, body_get_param, RID, BodyParameter); + + FUNC2(body_set_kinematic_safe_margin, RID, real_t); + FUNC1RC(real_t, body_get_kinematic_safe_margin, RID); + + FUNC3(body_set_state, RID, BodyState, const Variant &); + FUNC2RC(Variant, body_get_state, RID, BodyState); + + FUNC2(body_set_applied_force, RID, const Vector3 &); + FUNC1RC(Vector3, body_get_applied_force, RID); + + FUNC2(body_set_applied_torque, RID, const Vector3 &); + FUNC1RC(Vector3, body_get_applied_torque, RID); + + FUNC2(body_add_central_force, RID, const Vector3 &); + FUNC3(body_add_force, RID, const Vector3 &, const Vector3 &); + FUNC2(body_add_torque, RID, const Vector3 &); + FUNC2(body_apply_torque_impulse, RID, const Vector3 &); + FUNC2(body_apply_central_impulse, RID, const Vector3 &); + FUNC3(body_apply_impulse, RID, const Vector3 &, const Vector3 &); + FUNC2(body_set_axis_velocity, RID, const Vector3 &); + + FUNC3(body_set_axis_lock, RID, BodyAxis, bool); + FUNC2RC(bool, body_is_axis_locked, RID, BodyAxis); + + FUNC2(body_add_collision_exception, RID, RID); + FUNC2(body_remove_collision_exception, RID, RID); + FUNC2S(body_get_collision_exceptions, RID, List<RID> *); + + FUNC2(body_set_max_contacts_reported, RID, int); + FUNC1RC(int, body_get_max_contacts_reported, RID); + + FUNC2(body_set_contacts_reported_depth_threshold, RID, real_t); + FUNC1RC(real_t, body_get_contacts_reported_depth_threshold, RID); + + FUNC2(body_set_omit_force_integration, RID, bool); + FUNC1RC(bool, body_is_omitting_force_integration, RID); + + FUNC4(body_set_force_integration_callback, RID, Object *, const StringName &, const Variant &); + + FUNC2(body_set_ray_pickable, RID, bool); + + bool body_test_motion(RID p_body, const Transform &p_from, const Vector3 &p_motion, bool p_infinite_inertia, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); + return physics_3d_server->body_test_motion(p_body, p_from, p_motion, p_infinite_inertia, r_result, p_exclude_raycast_shapes); + } + + int body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), false); + return physics_3d_server->body_test_ray_separation(p_body, p_transform, p_infinite_inertia, r_recover_motion, r_results, p_result_max, p_margin); + } + + // this function only works on physics process, errors and returns null otherwise + PhysicsDirectBodyState3D *body_get_direct_state(RID p_body) override { + ERR_FAIL_COND_V(main_thread != Thread::get_caller_id(), nullptr); + return physics_3d_server->body_get_direct_state(p_body); + } + + /* SOFT BODY API */ + + FUNCRID(soft_body) + + FUNC2(soft_body_update_rendering_server, RID, class RenderingServerHandler *) + + FUNC2(soft_body_set_space, RID, RID) + FUNC1RC(RID, soft_body_get_space, RID) + + FUNC2(soft_body_set_ray_pickable, RID, bool); + + FUNC2(soft_body_set_collision_layer, RID, uint32_t) + FUNC1RC(uint32_t, soft_body_get_collision_layer, RID) + + FUNC2(soft_body_set_collision_mask, RID, uint32_t) + FUNC1RC(uint32_t, soft_body_get_collision_mask, RID) + + FUNC2(soft_body_add_collision_exception, RID, RID) + FUNC2(soft_body_remove_collision_exception, RID, RID) + FUNC2S(soft_body_get_collision_exceptions, RID, List<RID> *) + + FUNC3(soft_body_set_state, RID, BodyState, const Variant &); + FUNC2RC(Variant, soft_body_get_state, RID, BodyState); + + FUNC2(soft_body_set_transform, RID, const Transform &); + + FUNC2(soft_body_set_simulation_precision, RID, int); + FUNC1RC(int, soft_body_get_simulation_precision, RID); + + FUNC2(soft_body_set_total_mass, RID, real_t); + FUNC1RC(real_t, soft_body_get_total_mass, RID); + + FUNC2(soft_body_set_linear_stiffness, RID, real_t); + FUNC1RC(real_t, soft_body_get_linear_stiffness, RID); + + FUNC2(soft_body_set_pressure_coefficient, RID, real_t); + FUNC1RC(real_t, soft_body_get_pressure_coefficient, RID); + + FUNC2(soft_body_set_damping_coefficient, RID, real_t); + FUNC1RC(real_t, soft_body_get_damping_coefficient, RID); + + FUNC2(soft_body_set_drag_coefficient, RID, real_t); + FUNC1RC(real_t, soft_body_get_drag_coefficient, RID); + + FUNC2(soft_body_set_mesh, RID, const REF &); + + FUNC1RC(AABB, soft_body_get_bounds, RID); + + FUNC3(soft_body_move_point, RID, int, const Vector3 &); + FUNC2RC(Vector3, soft_body_get_point_global_position, RID, int); + + FUNC1(soft_body_remove_all_pinned_points, RID); + FUNC3(soft_body_pin_point, RID, int, bool); + FUNC2RC(bool, soft_body_is_point_pinned, RID, int); + + /* JOINT API */ + + FUNCRID(joint) + + FUNC1(joint_clear, RID) + + FUNC5(joint_make_pin, RID, RID, const Vector3 &, RID, const Vector3 &) + + FUNC3(pin_joint_set_param, RID, PinJointParam, real_t) + FUNC2RC(real_t, pin_joint_get_param, RID, PinJointParam) + + FUNC2(pin_joint_set_local_a, RID, const Vector3 &) + FUNC1RC(Vector3, pin_joint_get_local_a, RID) + + FUNC2(pin_joint_set_local_b, RID, const Vector3 &) + FUNC1RC(Vector3, pin_joint_get_local_b, RID) + + FUNC5(joint_make_hinge, RID, RID, const Transform &, RID, const Transform &) + FUNC7(joint_make_hinge_simple, RID, RID, const Vector3 &, const Vector3 &, RID, const Vector3 &, const Vector3 &) + + FUNC3(hinge_joint_set_param, RID, HingeJointParam, real_t) + FUNC2RC(real_t, hinge_joint_get_param, RID, HingeJointParam) + + FUNC3(hinge_joint_set_flag, RID, HingeJointFlag, bool) + FUNC2RC(bool, hinge_joint_get_flag, RID, HingeJointFlag) + + FUNC5(joint_make_slider, RID, RID, const Transform &, RID, const Transform &) + + FUNC3(slider_joint_set_param, RID, SliderJointParam, real_t) + FUNC2RC(real_t, slider_joint_get_param, RID, SliderJointParam) + + FUNC5(joint_make_cone_twist, RID, RID, const Transform &, RID, const Transform &) + + FUNC3(cone_twist_joint_set_param, RID, ConeTwistJointParam, real_t) + FUNC2RC(real_t, cone_twist_joint_get_param, RID, ConeTwistJointParam) + + FUNC5(joint_make_generic_6dof, RID, RID, const Transform &, RID, const Transform &) + + FUNC4(generic_6dof_joint_set_param, RID, Vector3::Axis, G6DOFJointAxisParam, real_t) + FUNC3RC(real_t, generic_6dof_joint_get_param, RID, Vector3::Axis, G6DOFJointAxisParam) + + FUNC4(generic_6dof_joint_set_flag, RID, Vector3::Axis, G6DOFJointAxisFlag, bool) + FUNC3RC(bool, generic_6dof_joint_get_flag, RID, Vector3::Axis, G6DOFJointAxisFlag) + + FUNC1RC(JointType, joint_get_type, RID); + + FUNC2(joint_set_solver_priority, RID, int); + FUNC1RC(int, joint_get_solver_priority, RID); + + FUNC2(joint_disable_collisions_between_bodies, RID, const bool); + FUNC1RC(bool, joint_is_disabled_collisions_between_bodies, RID); + + /* MISC */ + + FUNC1(free, RID); + FUNC1(set_active, bool); + + virtual void init() override; + virtual void step(real_t p_step) override; + virtual void sync() override; + virtual void end_sync() override; + virtual void flush_queries() override; + virtual void finish() override; + + virtual bool is_flushing_queries() const override { + return physics_3d_server->is_flushing_queries(); + } + + int get_process_info(ProcessInfo p_info) override { + return physics_3d_server->get_process_info(p_info); + } + + PhysicsServer3DWrapMT(PhysicsServer3D *p_contained, bool p_create_thread); + ~PhysicsServer3DWrapMT(); + +#undef ServerNameWrapMT +#undef ServerName +#undef server_name +#undef WRITE_ACTION +}; + +#ifdef DEBUG_SYNC +#undef DEBUG_SYNC +#endif +#undef SYNC_DEBUG + +#endif // PHYSICS3DSERVERWRAPMT_H diff --git a/servers/physics_3d/shape_3d_sw.cpp b/servers/physics_3d/shape_3d_sw.cpp index f2adcc1072..4c14cb3162 100644 --- a/servers/physics_3d/shape_3d_sw.cpp +++ b/servers/physics_3d/shape_3d_sw.cpp @@ -34,10 +34,12 @@ #include "core/math/quick_hull.h" #include "core/templates/sort_array.h" -#define _POINT_SNAP 0.001953125 #define _EDGE_IS_VALID_SUPPORT_THRESHOLD 0.0002 #define _FACE_IS_VALID_SUPPORT_THRESHOLD 0.9998 +#define _CYLINDER_EDGE_IS_VALID_SUPPORT_THRESHOLD 0.002 +#define _CYLINDER_FACE_IS_VALID_SUPPORT_THRESHOLD 0.999 + void Shape3DSW::configure(const AABB &p_aabb) { aabb = p_aabb; configured = true; @@ -50,7 +52,8 @@ void Shape3DSW::configure(const AABB &p_aabb) { Vector3 Shape3DSW::get_support(const Vector3 &p_normal) const { Vector3 res; int amnt; - get_supports(p_normal, 1, &res, amnt); + FeatureType type; + get_supports(p_normal, 1, &res, amnt, type); return res; } @@ -167,16 +170,19 @@ Vector3 RayShape3DSW::get_support(const Vector3 &p_normal) const { } } -void RayShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void RayShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { if (Math::abs(p_normal.z) < _EDGE_IS_VALID_SUPPORT_THRESHOLD) { r_amount = 2; + r_type = FEATURE_EDGE; r_supports[0] = Vector3(0, 0, 0); r_supports[1] = Vector3(0, 0, length); } else if (p_normal.z > 0) { r_amount = 1; + r_type = FEATURE_POINT; *r_supports = Vector3(0, 0, length); } else { r_amount = 1; + r_type = FEATURE_POINT; *r_supports = Vector3(0, 0, 0); } } @@ -246,9 +252,10 @@ Vector3 SphereShape3DSW::get_support(const Vector3 &p_normal) const { return p_normal * radius; } -void SphereShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void SphereShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { *r_supports = p_normal * radius; r_amount = 1; + r_type = FEATURE_POINT; } bool SphereShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const { @@ -261,7 +268,7 @@ bool SphereShape3DSW::intersect_point(const Vector3 &p_point) const { Vector3 SphereShape3DSW::get_closest_point_to(const Vector3 &p_point) const { Vector3 p = p_point; - float l = p.length(); + real_t l = p.length(); if (l < radius) { return p_point; } @@ -312,7 +319,7 @@ Vector3 BoxShape3DSW::get_support(const Vector3 &p_normal) const { return point; } -void BoxShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void BoxShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { static const int next[3] = { 1, 2, 0 }; static const int next2[3] = { 2, 0, 1 }; @@ -325,6 +332,7 @@ void BoxShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_s bool neg = dot < 0; r_amount = 4; + r_type = FEATURE_FACE; Vector3 point; point[i] = half_extents[i]; @@ -362,6 +370,7 @@ void BoxShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_s if (Math::abs(p_normal.dot(axis)) < _EDGE_IS_VALID_SUPPORT_THRESHOLD) { r_amount = 2; + r_type = FEATURE_EDGE; int i_n = next[i]; int i_n2 = next2[i]; @@ -389,6 +398,7 @@ void BoxShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_s (p_normal.z < 0) ? -half_extents.z : half_extents.z); r_amount = 1; + r_type = FEATURE_POINT; r_supports[0] = point; } @@ -429,7 +439,7 @@ Vector3 BoxShape3DSW::get_closest_point_to(const Vector3 &p_point) const { } //check segments - float min_distance = 1e20; + real_t min_distance = 1e20; Vector3 closest_vertex = half_extents * p_point.sign(); Vector3 s[2] = { closest_vertex, @@ -442,7 +452,7 @@ Vector3 BoxShape3DSW::get_closest_point_to(const Vector3 &p_point) const { Vector3 closest_edge = Geometry3D::get_closest_point_to_segment(p_point, s); - float d = p_point.distance_to(closest_edge); + real_t d = p_point.distance_to(closest_edge); if (d < min_distance) { min_point = closest_edge; min_distance = d; @@ -481,10 +491,10 @@ BoxShape3DSW::BoxShape3DSW() { void CapsuleShape3DSW::project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const { Vector3 n = p_transform.basis.xform_inv(p_normal).normalized(); - real_t h = (n.z > 0) ? height : -height; + real_t h = (n.y > 0) ? height : -height; n *= radius; - n.z += h * 0.5; + n.y += h * 0.5; r_max = p_normal.dot(p_transform.xform(n)); r_min = p_normal.dot(p_transform.xform(-n)); @@ -493,36 +503,38 @@ void CapsuleShape3DSW::project_range(const Vector3 &p_normal, const Transform &p Vector3 CapsuleShape3DSW::get_support(const Vector3 &p_normal) const { Vector3 n = p_normal; - real_t h = (n.z > 0) ? height : -height; + real_t h = (n.y > 0) ? height : -height; n *= radius; - n.z += h * 0.5; + n.y += h * 0.5; return n; } -void CapsuleShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void CapsuleShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { Vector3 n = p_normal; - real_t d = n.z; + real_t d = n.y; if (Math::abs(d) < _EDGE_IS_VALID_SUPPORT_THRESHOLD) { // make it flat - n.z = 0.0; + n.y = 0.0; n.normalize(); n *= radius; r_amount = 2; + r_type = FEATURE_EDGE; r_supports[0] = n; - r_supports[0].z += height * 0.5; + r_supports[0].y += height * 0.5; r_supports[1] = n; - r_supports[1].z -= height * 0.5; + r_supports[1].y -= height * 0.5; } else { real_t h = (d > 0) ? height : -height; n *= radius; - n.z += h * 0.5; + n.y += h * 0.5; r_amount = 1; + r_type = FEATURE_POINT; *r_supports = n; } } @@ -539,7 +551,7 @@ bool CapsuleShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 & // test against cylinder and spheres :-| - collided = Geometry3D::segment_intersects_cylinder(p_begin, p_end, height, radius, &auxres, &auxn); + collided = Geometry3D::segment_intersects_cylinder(p_begin, p_end, height, radius, &auxres, &auxn, 1); if (collided) { real_t d = norm.dot(auxres); @@ -551,7 +563,7 @@ bool CapsuleShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 & } } - collided = Geometry3D::segment_intersects_sphere(p_begin, p_end, Vector3(0, 0, height * 0.5), radius, &auxres, &auxn); + collided = Geometry3D::segment_intersects_sphere(p_begin, p_end, Vector3(0, height * 0.5, 0), radius, &auxres, &auxn); if (collided) { real_t d = norm.dot(auxres); @@ -563,7 +575,7 @@ bool CapsuleShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 & } } - collided = Geometry3D::segment_intersects_sphere(p_begin, p_end, Vector3(0, 0, height * -0.5), radius, &auxres, &auxn); + collided = Geometry3D::segment_intersects_sphere(p_begin, p_end, Vector3(0, height * -0.5, 0), radius, &auxres, &auxn); if (collided) { real_t d = norm.dot(auxres); @@ -584,19 +596,19 @@ bool CapsuleShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 & } bool CapsuleShape3DSW::intersect_point(const Vector3 &p_point) const { - if (Math::abs(p_point.z) < height * 0.5) { - return Vector3(p_point.x, p_point.y, 0).length() < radius; + if (Math::abs(p_point.y) < height * 0.5) { + return Vector3(p_point.x, 0, p_point.z).length() < radius; } else { Vector3 p = p_point; - p.z = Math::abs(p.z) - height * 0.5; + p.y = Math::abs(p.y) - height * 0.5; return p.length() < radius; } } Vector3 CapsuleShape3DSW::get_closest_point_to(const Vector3 &p_point) const { Vector3 s[2] = { - Vector3(0, 0, -height * 0.5), - Vector3(0, 0, height * 0.5), + Vector3(0, -height * 0.5, 0), + Vector3(0, height * 0.5, 0), }; Vector3 p = Geometry3D::get_closest_point_to_segment(p_point, s); @@ -615,13 +627,13 @@ Vector3 CapsuleShape3DSW::get_moment_of_inertia(real_t p_mass) const { return Vector3( (p_mass / 3.0) * (extents.y * extents.y + extents.z * extents.z), (p_mass / 3.0) * (extents.x * extents.x + extents.z * extents.z), - (p_mass / 3.0) * (extents.y * extents.y + extents.y * extents.y)); + (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); } void CapsuleShape3DSW::_setup(real_t p_height, real_t p_radius) { height = p_height; radius = p_radius; - configure(AABB(Vector3(-radius, -radius, -height * 0.5 - radius), Vector3(radius * 2, radius * 2, height + radius * 2.0))); + configure(AABB(Vector3(-radius, -height * 0.5 - radius, -radius), Vector3(radius * 2, height + radius * 2.0, radius * 2))); } void CapsuleShape3DSW::set_data(const Variant &p_data) { @@ -642,6 +654,186 @@ CapsuleShape3DSW::CapsuleShape3DSW() { height = radius = 0; } +/********** CYLINDER *************/ + +void CylinderShape3DSW::project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const { + Vector3 cylinder_axis = p_transform.basis.get_axis(1).normalized(); + real_t axis_dot = cylinder_axis.dot(p_normal); + + Vector3 local_normal = p_transform.basis.xform_inv(p_normal); + real_t scale = local_normal.length(); + real_t scaled_radius = radius * scale; + real_t scaled_height = height * scale; + + real_t length; + if (Math::abs(axis_dot) > 1.0) { + length = scaled_height * 0.5; + } else { + length = Math::abs(axis_dot * scaled_height * 0.5) + scaled_radius * Math::sqrt(1.0 - axis_dot * axis_dot); + } + + real_t distance = p_normal.dot(p_transform.origin); + + r_min = distance - length; + r_max = distance + length; +} + +Vector3 CylinderShape3DSW::get_support(const Vector3 &p_normal) const { + Vector3 n = p_normal; + real_t h = (n.y > 0) ? height : -height; + real_t s = Math::sqrt(n.x * n.x + n.z * n.z); + if (Math::is_zero_approx(s)) { + n.x = radius; + n.y = h * 0.5; + n.z = 0.0; + } else { + real_t d = radius / s; + n.x = n.x * d; + n.y = h * 0.5; + n.z = n.z * d; + } + + return n; +} + +void CylinderShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { + real_t d = p_normal.y; + if (Math::abs(d) > _CYLINDER_FACE_IS_VALID_SUPPORT_THRESHOLD) { + real_t h = (d > 0) ? height : -height; + + Vector3 n = p_normal; + n.x = 0.0; + n.z = 0.0; + n.y = h * 0.5; + + r_amount = 3; + r_type = FEATURE_CIRCLE; + r_supports[0] = n; + r_supports[1] = n; + r_supports[1].x += radius; + r_supports[2] = n; + r_supports[2].z += radius; + } else if (Math::abs(d) < _CYLINDER_EDGE_IS_VALID_SUPPORT_THRESHOLD) { + // make it flat + Vector3 n = p_normal; + n.y = 0.0; + n.normalize(); + n *= radius; + + r_amount = 2; + r_type = FEATURE_EDGE; + r_supports[0] = n; + r_supports[0].y += height * 0.5; + r_supports[1] = n; + r_supports[1].y -= height * 0.5; + } else { + r_amount = 1; + r_type = FEATURE_POINT; + r_supports[0] = get_support(p_normal); + return; + + Vector3 n = p_normal; + real_t h = n.y * Math::sqrt(0.25 * height * height + radius * radius); + if (Math::abs(h) > 1.0) { + // Top or bottom surface. + n.y = (n.y > 0.0) ? height * 0.5 : -height * 0.5; + } else { + // Lateral surface. + n.y = height * 0.5 * h; + } + + real_t s = Math::sqrt(n.x * n.x + n.z * n.z); + if (Math::is_zero_approx(s)) { + n.x = 0.0; + n.z = 0.0; + } else { + real_t scaled_radius = radius / s; + n.x = n.x * scaled_radius; + n.z = n.z * scaled_radius; + } + + r_supports[0] = n; + } +} + +bool CylinderShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const { + return Geometry3D::segment_intersects_cylinder(p_begin, p_end, height, radius, &r_result, &r_normal, 1); +} + +bool CylinderShape3DSW::intersect_point(const Vector3 &p_point) const { + if (Math::abs(p_point.y) < height * 0.5) { + return Vector3(p_point.x, 0, p_point.z).length() < radius; + } + return false; +} + +Vector3 CylinderShape3DSW::get_closest_point_to(const Vector3 &p_point) const { + if (Math::absf(p_point.y) > height * 0.5) { + // Project point to top disk. + real_t dir = p_point.y > 0.0 ? 1.0 : -1.0; + Vector3 circle_pos(0.0, dir * height * 0.5, 0.0); + Plane circle_plane(circle_pos, Vector3(0.0, dir, 0.0)); + Vector3 proj_point = circle_plane.project(p_point); + + // Clip position. + Vector3 delta_point_1 = proj_point - circle_pos; + real_t dist_point_1 = delta_point_1.length_squared(); + if (!Math::is_zero_approx(dist_point_1)) { + dist_point_1 = Math::sqrt(dist_point_1); + proj_point = circle_pos + delta_point_1 * MIN(dist_point_1, radius) / dist_point_1; + } + + return proj_point; + } else { + Vector3 s[2] = { + Vector3(0, -height * 0.5, 0), + Vector3(0, height * 0.5, 0), + }; + + Vector3 p = Geometry3D::get_closest_point_to_segment(p_point, s); + + if (p.distance_to(p_point) < radius) { + return p_point; + } + + return p + (p_point - p).normalized() * radius; + } +} + +Vector3 CylinderShape3DSW::get_moment_of_inertia(real_t p_mass) const { + // use bad AABB approximation + Vector3 extents = get_aabb().size * 0.5; + + return Vector3( + (p_mass / 3.0) * (extents.y * extents.y + extents.z * extents.z), + (p_mass / 3.0) * (extents.x * extents.x + extents.z * extents.z), + (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); +} + +void CylinderShape3DSW::_setup(real_t p_height, real_t p_radius) { + height = p_height; + radius = p_radius; + configure(AABB(Vector3(-radius, -height * 0.5, -radius), Vector3(radius * 2.0, height, radius * 2.0))); +} + +void CylinderShape3DSW::set_data(const Variant &p_data) { + Dictionary d = p_data; + ERR_FAIL_COND(!d.has("radius")); + ERR_FAIL_COND(!d.has("height")); + _setup(d["height"], d["radius"]); +} + +Variant CylinderShape3DSW::get_data() const { + Dictionary d; + d["radius"] = radius; + d["height"] = height; + return d; +} + +CylinderShape3DSW::CylinderShape3DSW() { + height = radius = 0; +} + /********** CONVEX POLYGON *************/ void ConvexPolygonShape3DSW::project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const { @@ -689,7 +881,7 @@ Vector3 ConvexPolygonShape3DSW::get_support(const Vector3 &p_normal) const { return vrts[vert_support_idx]; } -void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { const Geometry3D::MeshData::Face *faces = mesh.faces.ptr(); int fc = mesh.faces.size(); @@ -699,6 +891,9 @@ void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Ve const Vector3 *vertices = mesh.vertices.ptr(); int vc = mesh.vertices.size(); + r_amount = 0; + ERR_FAIL_COND_MSG(vc == 0, "Convex polygon shape has no vertices."); + //find vertex first real_t max = 0; int vtx = 0; @@ -734,6 +929,7 @@ void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Ve r_supports[j] = vertices[ind[j]]; } r_amount = m; + r_type = FEATURE_FACE; return; } } @@ -743,6 +939,7 @@ void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Ve dot = ABS(dot); if (dot < _EDGE_IS_VALID_SUPPORT_THRESHOLD && (edges[i].a == vtx || edges[i].b == vtx)) { r_amount = 2; + r_type = FEATURE_EDGE; r_supports[0] = vertices[edges[i].a]; r_supports[1] = vertices[edges[i].b]; return; @@ -751,6 +948,7 @@ void ConvexPolygonShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Ve r_supports[0] = vertices[vtx]; r_amount = 1; + r_type = FEATURE_POINT; } bool ConvexPolygonShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const { @@ -839,7 +1037,7 @@ Vector3 ConvexPolygonShape3DSW::get_closest_point_to(const Vector3 &p_point) con return p_point; } - float min_distance = 1e20; + real_t min_distance = 1e20; Vector3 min_point; //check edges @@ -852,7 +1050,7 @@ Vector3 ConvexPolygonShape3DSW::get_closest_point_to(const Vector3 &p_point) con }; Vector3 closest = Geometry3D::get_closest_point_to_segment(p_point, s); - float d = closest.distance_to(p_point); + real_t d = closest.distance_to(p_point); if (d < min_distance) { min_distance = d; min_point = closest; @@ -869,7 +1067,7 @@ Vector3 ConvexPolygonShape3DSW::get_moment_of_inertia(real_t p_mass) const { return Vector3( (p_mass / 3.0) * (extents.y * extents.y + extents.z * extents.z), (p_mass / 3.0) * (extents.x * extents.x + extents.z * extents.z), - (p_mass / 3.0) * (extents.y * extents.y + extents.y * extents.y)); + (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); } void ConvexPolygonShape3DSW::_setup(const Vector<Vector3> &p_vertices) { @@ -935,12 +1133,13 @@ Vector3 FaceShape3DSW::get_support(const Vector3 &p_normal) const { return vertex[vert_support_idx]; } -void FaceShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { +void FaceShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { Vector3 n = p_normal; /** TEST FACE AS SUPPORT **/ - if (normal.dot(n) > _FACE_IS_VALID_SUPPORT_THRESHOLD) { + if (Math::abs(normal.dot(n)) > _FACE_IS_VALID_SUPPORT_THRESHOLD) { r_amount = 3; + r_type = FEATURE_FACE; for (int i = 0; i < 3; i++) { r_supports[i] = vertex[i]; } @@ -974,6 +1173,7 @@ void FaceShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_ dot = ABS(dot); if (dot < _EDGE_IS_VALID_SUPPORT_THRESHOLD) { r_amount = 2; + r_type = FEATURE_EDGE; r_supports[0] = vertex[i]; r_supports[1] = vertex[nx]; return; @@ -981,6 +1181,7 @@ void FaceShape3DSW::get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_ } r_amount = 1; + r_type = FEATURE_POINT; r_supports[0] = vertex[vert_support_idx]; } @@ -989,7 +1190,11 @@ bool FaceShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 &p_e if (c) { r_normal = Plane(vertex[0], vertex[1], vertex[2]).normal; if (r_normal.dot(p_end - p_begin) > 0) { - r_normal = -r_normal; + if (backface_collision) { + r_normal = -r_normal; + } else { + c = false; + } } } @@ -1087,30 +1292,24 @@ void ConcavePolygonShape3DSW::_cull_segment(int p_idx, _SegmentCullParams *p_par } if (bvh->face_index >= 0) { - Vector3 res; - Vector3 vertices[3] = { - p_params->vertices[p_params->faces[bvh->face_index].indices[0]], - p_params->vertices[p_params->faces[bvh->face_index].indices[1]], - p_params->vertices[p_params->faces[bvh->face_index].indices[2]] - }; + const Face *f = &p_params->faces[bvh->face_index]; + FaceShape3DSW *face = p_params->face; + face->normal = f->normal; + face->vertex[0] = p_params->vertices[f->indices[0]]; + face->vertex[1] = p_params->vertices[f->indices[1]]; + face->vertex[2] = p_params->vertices[f->indices[2]]; - if (Geometry3D::segment_intersects_triangle( - p_params->from, - p_params->to, - vertices[0], - vertices[1], - vertices[2], - &res)) { + Vector3 res; + Vector3 normal; + if (face->intersect_segment(p_params->from, p_params->to, res, normal)) { real_t d = p_params->dir.dot(res) - p_params->dir.dot(p_params->from); - //TODO, seems segmen/triangle intersection is broken :( - if (d > 0 && d < p_params->min_d) { + if ((d > 0) && (d < p_params->min_d)) { p_params->min_d = d; p_params->result = res; - p_params->normal = Plane(vertices[0], vertices[1], vertices[2]).normal; + p_params->normal = normal; p_params->collisions++; } } - } else { if (bvh->left >= 0) { _cull_segment(bvh->left, p_params); @@ -1131,17 +1330,20 @@ bool ConcavePolygonShape3DSW::intersect_segment(const Vector3 &p_begin, const Ve const Vector3 *vr = vertices.ptr(); const BVH *br = bvh.ptr(); + FaceShape3DSW face; + face.backface_collision = backface_collision; + _SegmentCullParams params; params.from = p_begin; params.to = p_end; - params.collisions = 0; params.dir = (p_end - p_begin).normalized(); params.faces = fr; params.vertices = vr; params.bvh = br; - params.min_d = 1e20; + params.face = &face; + // cull _cull_segment(0, ¶ms); @@ -1203,6 +1405,7 @@ void ConcavePolygonShape3DSW::cull(const AABB &p_local_aabb, Callback p_callback const BVH *br = bvh.ptr(); FaceShape3DSW face; // use this to send in the callback + face.backface_collision = backface_collision; _CullParams params; params.aabb = local_aabb; @@ -1224,7 +1427,7 @@ Vector3 ConcavePolygonShape3DSW::get_moment_of_inertia(real_t p_mass) const { return Vector3( (p_mass / 3.0) * (extents.y * extents.y + extents.z * extents.z), (p_mass / 3.0) * (extents.x * extents.x + extents.z * extents.z), - (p_mass / 3.0) * (extents.y * extents.y + extents.y * extents.y)); + (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); } struct _VolumeSW_BVH_Element { @@ -1334,7 +1537,7 @@ void ConcavePolygonShape3DSW::_fill_bvh(_VolumeSW_BVH *p_bvh_tree, BVH *p_bvh_ar memdelete(p_bvh_tree); } -void ConcavePolygonShape3DSW::_setup(Vector<Vector3> p_faces) { +void ConcavePolygonShape3DSW::_setup(const Vector<Vector3> &p_faces, bool p_backface_collision) { int src_face_count = p_faces.size(); if (src_face_count == 0) { configure(AABB()); @@ -1389,15 +1592,24 @@ void ConcavePolygonShape3DSW::_setup(Vector<Vector3> p_faces) { int idx = 0; _fill_bvh(bvh_tree, bvh_arrayw2, idx); + backface_collision = p_backface_collision; + configure(_aabb); // this type of shape has no margin } void ConcavePolygonShape3DSW::set_data(const Variant &p_data) { - _setup(p_data); + Dictionary d = p_data; + ERR_FAIL_COND(!d.has("faces")); + + _setup(d["faces"], d["backface_collision"]); } Variant ConcavePolygonShape3DSW::get_data() const { - return get_faces(); + Dictionary d; + d["faces"] = get_faces(); + d["backface_collision"] = backface_collision; + + return d; } ConcavePolygonShape3DSW::ConcavePolygonShape3DSW() { @@ -1453,7 +1665,7 @@ Vector3 HeightMapShape3DSW::get_moment_of_inertia(real_t p_mass) const { return Vector3( (p_mass / 3.0) * (extents.y * extents.y + extents.z * extents.z), (p_mass / 3.0) * (extents.x * extents.x + extents.z * extents.z), - (p_mass / 3.0) * (extents.y * extents.y + extents.y * extents.y)); + (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); } void HeightMapShape3DSW::_setup(Vector<real_t> p_heights, int p_width, int p_depth, real_t p_cell_size) { diff --git a/servers/physics_3d/shape_3d_sw.h b/servers/physics_3d/shape_3d_sw.h index 851c0d9443..988e76c699 100644 --- a/servers/physics_3d/shape_3d_sw.h +++ b/servers/physics_3d/shape_3d_sw.h @@ -67,8 +67,11 @@ protected: void configure(const AABB &p_aabb); public: - enum { - MAX_SUPPORTS = 8 + enum FeatureType { + FEATURE_POINT, + FEATURE_EDGE, + FEATURE_FACE, + FEATURE_CIRCLE, }; virtual real_t get_area() const { return aabb.get_area(); } @@ -85,7 +88,7 @@ public: virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const = 0; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const = 0; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const = 0; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const = 0; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_point, Vector3 &r_normal) const = 0; virtual bool intersect_point(const Vector3 &p_point) const = 0; @@ -110,7 +113,7 @@ class ConcaveShape3DSW : public Shape3DSW { public: virtual bool is_concave() const { return true; } typedef void (*Callback)(void *p_userdata, Shape3DSW *p_convex); - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { r_amount = 0; } + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { r_amount = 0; } virtual void cull(const AABB &p_local_aabb, Callback p_callback, void *p_userdata) const = 0; @@ -129,7 +132,7 @@ public: virtual PhysicsServer3D::ShapeType get_type() const { return PhysicsServer3D::SHAPE_PLANE; } virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { r_amount = 0; } + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { r_amount = 0; } virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; @@ -156,7 +159,7 @@ public: virtual PhysicsServer3D::ShapeType get_type() const { return PhysicsServer3D::SHAPE_RAY; } virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; @@ -184,7 +187,7 @@ public: virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; @@ -209,7 +212,7 @@ public: virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; @@ -238,7 +241,7 @@ public: virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; @@ -251,6 +254,35 @@ public: CapsuleShape3DSW(); }; +class CylinderShape3DSW : public Shape3DSW { + real_t height; + real_t radius; + + void _setup(real_t p_height, real_t p_radius); + +public: + _FORCE_INLINE_ real_t get_height() const { return height; } + _FORCE_INLINE_ real_t get_radius() const { return radius; } + + virtual real_t get_area() const { return 4.0 / 3.0 * Math_PI * radius * radius * radius + height * Math_PI * radius * radius; } + + virtual PhysicsServer3D::ShapeType get_type() const { return PhysicsServer3D::SHAPE_CYLINDER; } + + virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; + virtual Vector3 get_support(const Vector3 &p_normal) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; + virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; + virtual bool intersect_point(const Vector3 &p_point) const; + virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; + + virtual Vector3 get_moment_of_inertia(real_t p_mass) const; + + virtual void set_data(const Variant &p_data); + virtual Variant get_data() const; + + CylinderShape3DSW(); +}; + struct ConvexPolygonShape3DSW : public Shape3DSW { Geometry3D::MeshData mesh; @@ -263,7 +295,7 @@ public: virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; virtual Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; @@ -302,34 +334,37 @@ struct ConcavePolygonShape3DSW : public ConcaveShape3DSW { struct _CullParams { AABB aabb; - Callback callback; - void *userdata; - const Face *faces; - const Vector3 *vertices; - const BVH *bvh; - FaceShape3DSW *face; + Callback callback = nullptr; + void *userdata = nullptr; + const Face *faces = nullptr; + const Vector3 *vertices = nullptr; + const BVH *bvh = nullptr; + FaceShape3DSW *face = nullptr; }; struct _SegmentCullParams { Vector3 from; Vector3 to; - const Face *faces; - const Vector3 *vertices; - const BVH *bvh; Vector3 dir; + const Face *faces = nullptr; + const Vector3 *vertices = nullptr; + const BVH *bvh = nullptr; + FaceShape3DSW *face = nullptr; Vector3 result; Vector3 normal; - real_t min_d; - int collisions; + real_t min_d = 1e20; + int collisions = 0; }; + bool backface_collision = false; + void _cull_segment(int p_idx, _SegmentCullParams *p_params) const; void _cull(int p_idx, _CullParams *p_params) const; void _fill_bvh(_VolumeSW_BVH *p_bvh_tree, BVH *p_bvh_array, int &p_idx); - void _setup(Vector<Vector3> p_faces); + void _setup(const Vector<Vector3> &p_faces, bool p_backface_collision); public: Vector<Vector3> get_faces() const; @@ -392,6 +427,7 @@ public: struct FaceShape3DSW : public Shape3DSW { Vector3 normal; //cache Vector3 vertex[3]; + bool backface_collision = false; virtual PhysicsServer3D::ShapeType get_type() const { return PhysicsServer3D::SHAPE_CONCAVE_POLYGON; } @@ -399,7 +435,7 @@ struct FaceShape3DSW : public Shape3DSW { void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const; Vector3 get_support(const Vector3 &p_normal) const; - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const; + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const; bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; virtual bool intersect_point(const Vector3 &p_point) const; virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; @@ -437,7 +473,7 @@ struct MotionShape3DSW : public Shape3DSW { } return support; } - virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount) const { r_amount = 0; } + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { r_amount = 0; } bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const { return false; } virtual bool intersect_point(const Vector3 &p_point) const { return false; } virtual Vector3 get_closest_point_to(const Vector3 &p_point) const { return p_point; } diff --git a/servers/physics_3d/soft_body_3d_sw.cpp b/servers/physics_3d/soft_body_3d_sw.cpp new file mode 100644 index 0000000000..f63a470cbe --- /dev/null +++ b/servers/physics_3d/soft_body_3d_sw.cpp @@ -0,0 +1,1221 @@ +/*************************************************************************/ +/* soft_body_3d_sw.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "soft_body_3d_sw.h" +#include "space_3d_sw.h" + +#include "core/math/geometry_3d.h" +#include "core/templates/map.h" + +// Based on Bullet soft body. + +/* +Bullet Continuous Collision Detection and Physics Library +Copyright (c) 2003-2006 Erwin Coumans http://continuousphysics.com/Bullet/ + +This software is provided 'as-is', without any express or implied warranty. +In no event will the authors be held liable for any damages arising from the use of this software. +Permission is granted to anyone to use this software for any purpose, +including commercial applications, and to alter it and redistribute it freely, +subject to the following restrictions: + +1. The origin of this software must not be misrepresented; you must not claim that you wrote the original software. If you use this software in a product, an acknowledgment in the product documentation would be appreciated but is not required. +2. Altered source versions must be plainly marked as such, and must not be misrepresented as being the original software. +3. This notice may not be removed or altered from any source distribution. +*/ +///btSoftBody implementation by Nathanael Presson + +SoftBody3DSW::SoftBody3DSW() : + CollisionObject3DSW(TYPE_SOFT_BODY), + active_list(this) { + _set_static(false); +} + +void SoftBody3DSW::_shapes_changed() { +} + +void SoftBody3DSW::set_state(PhysicsServer3D::BodyState p_state, const Variant &p_variant) { + switch (p_state) { + case PhysicsServer3D::BODY_STATE_TRANSFORM: { + _set_transform(p_variant); + _set_inv_transform(get_transform().inverse()); + + apply_nodes_transform(get_transform()); + + } break; + case PhysicsServer3D::BODY_STATE_LINEAR_VELOCITY: { + // Not supported. + ERR_FAIL_MSG("Linear velocity is not supported for Soft bodies."); + } break; + case PhysicsServer3D::BODY_STATE_ANGULAR_VELOCITY: { + ERR_FAIL_MSG("Angular velocity is not supported for Soft bodies."); + } break; + case PhysicsServer3D::BODY_STATE_SLEEPING: { + ERR_FAIL_MSG("Sleeping state is not supported for Soft bodies."); + } break; + case PhysicsServer3D::BODY_STATE_CAN_SLEEP: { + ERR_FAIL_MSG("Sleeping state is not supported for Soft bodies."); + } break; + } +} + +Variant SoftBody3DSW::get_state(PhysicsServer3D::BodyState p_state) const { + switch (p_state) { + case PhysicsServer3D::BODY_STATE_TRANSFORM: { + return get_transform(); + } break; + case PhysicsServer3D::BODY_STATE_LINEAR_VELOCITY: { + ERR_FAIL_V_MSG(Vector3(), "Linear velocity is not supported for Soft bodies."); + } break; + case PhysicsServer3D::BODY_STATE_ANGULAR_VELOCITY: { + ERR_FAIL_V_MSG(Vector3(), "Angular velocity is not supported for Soft bodies."); + return Vector3(); + } break; + case PhysicsServer3D::BODY_STATE_SLEEPING: { + ERR_FAIL_V_MSG(false, "Sleeping state is not supported for Soft bodies."); + } break; + case PhysicsServer3D::BODY_STATE_CAN_SLEEP: { + ERR_FAIL_V_MSG(false, "Sleeping state is not supported for Soft bodies."); + } break; + } + + return Variant(); +} + +void SoftBody3DSW::set_space(Space3DSW *p_space) { + if (get_space()) { + get_space()->soft_body_remove_from_active_list(&active_list); + + deinitialize_shape(); + } + + _set_space(p_space); + + if (get_space()) { + get_space()->soft_body_add_to_active_list(&active_list); + + if (bounds != AABB()) { + initialize_shape(true); + } + } +} + +void SoftBody3DSW::set_mesh(const Ref<Mesh> &p_mesh) { + destroy(); + + soft_mesh = p_mesh; + + if (soft_mesh.is_null()) { + return; + } + + Array arrays = soft_mesh->surface_get_arrays(0); + ERR_FAIL_COND(!(soft_mesh->surface_get_format(0) & RS::ARRAY_FORMAT_INDEX)); + + bool success = create_from_trimesh(arrays[RS::ARRAY_INDEX], arrays[RS::ARRAY_VERTEX]); + if (!success) { + destroy(); + soft_mesh = Ref<Mesh>(); + } +} + +void SoftBody3DSW::update_rendering_server(RenderingServerHandler *p_rendering_server_handler) { + if (soft_mesh.is_null()) { + return; + } + + const uint32_t vertex_count = map_visual_to_physics.size(); + for (uint32_t i = 0; i < vertex_count; ++i) { + const uint32_t node_index = map_visual_to_physics[i]; + const Node &node = nodes[node_index]; + const Vector3 &vertex_position = node.x; + const Vector3 &vertex_normal = node.n; + + p_rendering_server_handler->set_vertex(i, &vertex_position); + p_rendering_server_handler->set_normal(i, &vertex_normal); + } + + p_rendering_server_handler->set_aabb(bounds); +} + +void SoftBody3DSW::update_normals() { + uint32_t i, ni; + + for (i = 0, ni = nodes.size(); i < ni; ++i) { + nodes[i].n = Vector3(); + } + + for (i = 0, ni = faces.size(); i < ni; ++i) { + Face &face = faces[i]; + const Vector3 n = vec3_cross(face.n[0]->x - face.n[2]->x, face.n[0]->x - face.n[1]->x); + face.n[0]->n += n; + face.n[1]->n += n; + face.n[2]->n += n; + face.normal = n; + face.normal.normalize(); + } + + for (i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + real_t len = node.n.length(); + if (len > CMP_EPSILON) { + node.n /= len; + } + } +} + +void SoftBody3DSW::update_bounds() { + AABB prev_bounds = bounds; + prev_bounds.grow_by(collision_margin); + + bounds = AABB(); + + const uint32_t nodes_count = nodes.size(); + if (nodes_count == 0) { + deinitialize_shape(); + return; + } + + bool first = true; + bool moved = false; + for (uint32_t node_index = 0; node_index < nodes_count; ++node_index) { + const Node &node = nodes[node_index]; + if (!prev_bounds.has_point(node.x)) { + moved = true; + } + if (first) { + bounds.position = node.x; + first = false; + } else { + bounds.expand_to(node.x); + } + } + + if (get_space()) { + initialize_shape(moved); + } +} + +void SoftBody3DSW::update_constants() { + reset_link_rest_lengths(); + update_link_constants(); + update_area(); +} + +void SoftBody3DSW::update_area() { + int i, ni; + + // Face area. + for (i = 0, ni = faces.size(); i < ni; ++i) { + Face &face = faces[i]; + + const Vector3 &x0 = face.n[0]->x; + const Vector3 &x1 = face.n[1]->x; + const Vector3 &x2 = face.n[2]->x; + + const Vector3 a = x1 - x0; + const Vector3 b = x2 - x0; + const Vector3 cr = vec3_cross(a, b); + face.ra = cr.length(); + } + + // Node area. + LocalVector<int> counts; + counts.resize(nodes.size()); + memset(counts.ptr(), 0, counts.size() * sizeof(int)); + + for (i = 0, ni = nodes.size(); i < ni; ++i) { + nodes[i].area = 0.0; + } + + for (i = 0, ni = faces.size(); i < ni; ++i) { + const Face &face = faces[i]; + for (int j = 0; j < 3; ++j) { + const int index = (int)(face.n[j] - &nodes[0]); + counts[index]++; + face.n[j]->area += Math::abs(face.ra); + } + } + + for (i = 0, ni = nodes.size(); i < ni; ++i) { + if (counts[i] > 0) { + nodes[i].area /= (real_t)counts[i]; + } else { + nodes[i].area = 0.0; + } + } +} + +void SoftBody3DSW::reset_link_rest_lengths() { + for (uint32_t i = 0, ni = links.size(); i < ni; ++i) { + Link &link = links[i]; + link.rl = (link.n[0]->x - link.n[1]->x).length(); + link.c1 = link.rl * link.rl; + } +} + +void SoftBody3DSW::update_link_constants() { + real_t inv_linear_stiffness = 1.0 / linear_stiffness; + for (uint32_t i = 0, ni = links.size(); i < ni; ++i) { + Link &link = links[i]; + link.c0 = (link.n[0]->im + link.n[1]->im) * inv_linear_stiffness; + } +} + +void SoftBody3DSW::apply_nodes_transform(const Transform &p_transform) { + if (soft_mesh.is_null()) { + return; + } + + uint32_t node_count = nodes.size(); + Vector3 leaf_size = Vector3(collision_margin, collision_margin, collision_margin) * 2.0; + for (uint32_t node_index = 0; node_index < node_count; ++node_index) { + Node &node = nodes[node_index]; + + node.x = p_transform.xform(node.x); + node.q = node.x; + node.v = Vector3(); + node.bv = Vector3(); + + AABB node_aabb(node.x, leaf_size); + node_tree.update(node.leaf, node_aabb); + } + + face_tree.clear(); + + update_normals(); + update_bounds(); + update_constants(); +} + +Vector3 SoftBody3DSW::get_vertex_position(int p_index) const { + if (soft_mesh.is_null()) { + return Vector3(); + } + + ERR_FAIL_INDEX_V(p_index, (int)map_visual_to_physics.size(), Vector3()); + uint32_t node_index = map_visual_to_physics[p_index]; + + ERR_FAIL_COND_V(node_index >= nodes.size(), Vector3()); + return nodes[node_index].x; +} + +void SoftBody3DSW::set_vertex_position(int p_index, const Vector3 &p_position) { + if (soft_mesh.is_null()) { + return; + } + + ERR_FAIL_INDEX(p_index, (int)map_visual_to_physics.size()); + uint32_t node_index = map_visual_to_physics[p_index]; + + ERR_FAIL_COND(node_index >= nodes.size()); + Node &node = nodes[node_index]; + node.q = node.x; + node.x = p_position; +} + +void SoftBody3DSW::pin_vertex(int p_index) { + if (is_vertex_pinned(p_index)) { + return; + } + + pinned_vertices.push_back(p_index); + + if (!soft_mesh.is_null()) { + ERR_FAIL_INDEX(p_index, (int)map_visual_to_physics.size()); + uint32_t node_index = map_visual_to_physics[p_index]; + + ERR_FAIL_COND(node_index >= nodes.size()); + Node &node = nodes[node_index]; + node.im = 0.0; + } +} + +void SoftBody3DSW::unpin_vertex(int p_index) { + uint32_t pinned_count = pinned_vertices.size(); + for (uint32_t i = 0; i < pinned_count; ++i) { + if (p_index == pinned_vertices[i]) { + pinned_vertices.remove(i); + + if (!soft_mesh.is_null()) { + ERR_FAIL_INDEX(p_index, (int)map_visual_to_physics.size()); + uint32_t node_index = map_visual_to_physics[p_index]; + + ERR_FAIL_COND(node_index >= nodes.size()); + real_t inv_node_mass = nodes.size() * inv_total_mass; + + Node &node = nodes[node_index]; + node.im = inv_node_mass; + } + + return; + } + } +} + +void SoftBody3DSW::unpin_all_vertices() { + if (!soft_mesh.is_null()) { + real_t inv_node_mass = nodes.size() * inv_total_mass; + uint32_t pinned_count = pinned_vertices.size(); + for (uint32_t i = 0; i < pinned_count; ++i) { + uint32_t vertex_index = pinned_vertices[i]; + + ERR_CONTINUE(vertex_index >= map_visual_to_physics.size()); + uint32_t node_index = map_visual_to_physics[vertex_index]; + + ERR_CONTINUE(node_index >= nodes.size()); + Node &node = nodes[node_index]; + node.im = inv_node_mass; + } + } + + pinned_vertices.clear(); +} + +bool SoftBody3DSW::is_vertex_pinned(int p_index) const { + uint32_t pinned_count = pinned_vertices.size(); + for (uint32_t i = 0; i < pinned_count; ++i) { + if (p_index == pinned_vertices[i]) { + return true; + } + } + + return false; +} + +uint32_t SoftBody3DSW::get_node_count() const { + return nodes.size(); +} + +real_t SoftBody3DSW::get_node_inv_mass(uint32_t p_node_index) const { + ERR_FAIL_COND_V(p_node_index >= nodes.size(), 0.0); + return nodes[p_node_index].im; +} + +Vector3 SoftBody3DSW::get_node_position(uint32_t p_node_index) const { + ERR_FAIL_COND_V(p_node_index >= nodes.size(), Vector3()); + return nodes[p_node_index].x; +} + +Vector3 SoftBody3DSW::get_node_velocity(uint32_t p_node_index) const { + ERR_FAIL_COND_V(p_node_index >= nodes.size(), Vector3()); + return nodes[p_node_index].v; +} + +Vector3 SoftBody3DSW::get_node_biased_velocity(uint32_t p_node_index) const { + ERR_FAIL_COND_V(p_node_index >= nodes.size(), Vector3()); + return nodes[p_node_index].bv; +} + +void SoftBody3DSW::apply_node_impulse(uint32_t p_node_index, const Vector3 &p_impulse) { + ERR_FAIL_COND(p_node_index >= nodes.size()); + Node &node = nodes[p_node_index]; + node.v += p_impulse * node.im; +} + +void SoftBody3DSW::apply_node_bias_impulse(uint32_t p_node_index, const Vector3 &p_impulse) { + ERR_FAIL_COND(p_node_index >= nodes.size()); + Node &node = nodes[p_node_index]; + node.bv += p_impulse * node.im; +} + +uint32_t SoftBody3DSW::get_face_count() const { + return faces.size(); +} + +void SoftBody3DSW::get_face_points(uint32_t p_face_index, Vector3 &r_point_1, Vector3 &r_point_2, Vector3 &r_point_3) const { + ERR_FAIL_COND(p_face_index >= faces.size()); + const Face &face = faces[p_face_index]; + r_point_1 = face.n[0]->x; + r_point_2 = face.n[1]->x; + r_point_3 = face.n[2]->x; +} + +Vector3 SoftBody3DSW::get_face_normal(uint32_t p_face_index) const { + ERR_FAIL_COND_V(p_face_index >= faces.size(), Vector3()); + return faces[p_face_index].normal; +} + +bool SoftBody3DSW::create_from_trimesh(const Vector<int> &p_indices, const Vector<Vector3> &p_vertices) { + uint32_t node_count = 0; + LocalVector<Vector3> vertices; + const int visual_vertex_count(p_vertices.size()); + + LocalVector<int> triangles; + const uint32_t triangle_count(p_indices.size() / 3); + triangles.resize(triangle_count * 3); + + // Merge all overlapping vertices and create a map of physical vertices to visual vertices. + { + // Process vertices. + { + uint32_t vertex_count = 0; + Map<Vector3, uint32_t> unique_vertices; + + vertices.resize(visual_vertex_count); + map_visual_to_physics.resize(visual_vertex_count); + + for (int visual_vertex_index = 0; visual_vertex_index < visual_vertex_count; ++visual_vertex_index) { + const Vector3 &vertex = p_vertices[visual_vertex_index]; + + Map<Vector3, uint32_t>::Element *e = unique_vertices.find(vertex); + uint32_t vertex_id; + if (e) { + // Already existing. + vertex_id = e->value(); + } else { + // Create new one. + vertex_id = vertex_count++; + unique_vertices[vertex] = vertex_id; + vertices[vertex_id] = vertex; + } + + map_visual_to_physics[visual_vertex_index] = vertex_id; + } + + vertices.resize(vertex_count); + } + + // Process triangles. + { + for (uint32_t triangle_index = 0; triangle_index < triangle_count; ++triangle_index) { + for (int i = 0; i < 3; ++i) { + int visual_index = 3 * triangle_index + i; + int physics_index = map_visual_to_physics[p_indices[visual_index]]; + triangles[visual_index] = physics_index; + node_count = MAX((int)node_count, physics_index); + } + } + } + } + + ++node_count; + + // Create nodes from vertices. + nodes.resize(node_count); + real_t inv_node_mass = node_count * inv_total_mass; + Vector3 leaf_size = Vector3(collision_margin, collision_margin, collision_margin) * 2.0; + for (uint32_t i = 0; i < node_count; ++i) { + Node &node = nodes[i]; + node.s = vertices[i]; + node.x = node.s; + node.q = node.s; + node.im = inv_node_mass; + + AABB node_aabb(node.x, leaf_size); + node.leaf = node_tree.insert(node_aabb, &node); + + node.index = i; + } + + // Create links and faces from triangles. + LocalVector<bool> chks; + chks.resize(node_count * node_count); + memset(chks.ptr(), 0, chks.size() * sizeof(bool)); + + for (uint32_t i = 0; i < triangle_count * 3; i += 3) { + const int idx[] = { triangles[i], triangles[i + 1], triangles[i + 2] }; + + for (int j = 2, k = 0; k < 3; j = k++) { + int chk = idx[k] * node_count + idx[j]; + if (!chks[chk]) { + chks[chk] = true; + int inv_chk = idx[j] * node_count + idx[k]; + chks[inv_chk] = true; + + append_link(idx[j], idx[k]); + } + } + + append_face(idx[0], idx[1], idx[2]); + } + + // Set pinned nodes. + uint32_t pinned_count = pinned_vertices.size(); + for (uint32_t i = 0; i < pinned_count; ++i) { + int pinned_vertex = pinned_vertices[i]; + + ERR_CONTINUE(pinned_vertex >= visual_vertex_count); + uint32_t node_index = map_visual_to_physics[pinned_vertex]; + + ERR_CONTINUE(node_index >= node_count); + Node &node = nodes[node_index]; + node.im = 0.0; + } + + generate_bending_constraints(2); + reoptimize_link_order(); + + update_constants(); + update_normals(); + update_bounds(); + + return true; +} + +void SoftBody3DSW::generate_bending_constraints(int p_distance) { + uint32_t i, j; + + if (p_distance > 1) { + // Build graph. + const uint32_t n = nodes.size(); + const unsigned inf = (~(unsigned)0) >> 1; + const uint32_t adj_size = n * n; + unsigned *adj = memnew_arr(unsigned, adj_size); + +#define IDX(_x_, _y_) ((_y_)*n + (_x_)) + for (j = 0; j < n; ++j) { + for (i = 0; i < n; ++i) { + int idx_ij = j * n + i; + int idx_ji = i * n + j; + if (i != j) { + adj[idx_ij] = adj[idx_ji] = inf; + } else { + adj[idx_ij] = adj[idx_ji] = 0; + } + } + } + for (i = 0; i < links.size(); ++i) { + const int ia = (int)(links[i].n[0] - &nodes[0]); + const int ib = (int)(links[i].n[1] - &nodes[0]); + int idx = ib * n + ia; + int idx_inv = ia * n + ib; + adj[idx] = 1; + adj[idx_inv] = 1; + } + + // Special optimized case for distance == 2. + if (p_distance == 2) { + LocalVector<LocalVector<int>> node_links; + + // Build node links. + node_links.resize(nodes.size()); + + for (i = 0; i < links.size(); ++i) { + const int ia = (int)(links[i].n[0] - &nodes[0]); + const int ib = (int)(links[i].n[1] - &nodes[0]); + if (node_links[ia].find(ib) == -1) { + node_links[ia].push_back(ib); + } + + if (node_links[ib].find(ia) == -1) { + node_links[ib].push_back(ia); + } + } + for (uint32_t ii = 0; ii < node_links.size(); ii++) { + for (uint32_t jj = 0; jj < node_links[ii].size(); jj++) { + int k = node_links[ii][jj]; + for (uint32_t kk = 0; kk < node_links[k].size(); kk++) { + int l = node_links[k][kk]; + if ((int)ii != l) { + int idx_ik = k * n + ii; + int idx_kj = l * n + k; + const unsigned sum = adj[idx_ik] + adj[idx_kj]; + ERR_FAIL_COND(sum != 2); + int idx_ij = l * n + ii; + if (adj[idx_ij] > sum) { + int idx_ji = l * n + ii; + adj[idx_ij] = adj[idx_ji] = sum; + } + } + } + } + } + } else { + // Generic Floyd's algorithm. + for (uint32_t k = 0; k < n; ++k) { + for (j = 0; j < n; ++j) { + for (i = j + 1; i < n; ++i) { + int idx_ik = k * n + i; + int idx_kj = j * n + k; + const unsigned sum = adj[idx_ik] + adj[idx_kj]; + int idx_ij = j * n + i; + if (adj[idx_ij] > sum) { + int idx_ji = j * n + i; + adj[idx_ij] = adj[idx_ji] = sum; + } + } + } + } + } + + // Build links. + for (j = 0; j < n; ++j) { + for (i = j + 1; i < n; ++i) { + int idx_ij = j * n + i; + if (adj[idx_ij] == (unsigned)p_distance) { + append_link(i, j); + } + } + } + memdelete_arr(adj); + } +} + +//=================================================================== +// +// +// This function takes in a list of interdependent Links and tries +// to maximize the distance between calculation +// of dependent links. This increases the amount of parallelism that can +// be exploited by out-of-order instruction processors with large but +// (inevitably) finite instruction windows. +// +//=================================================================== + +// A small structure to track lists of dependent link calculations. +class LinkDeps { +public: + int value; // A link calculation that is dependent on this one + // Positive values = "input A" while negative values = "input B" + LinkDeps *next; // Next dependence in the list +}; +typedef LinkDeps *LinkDepsPtr; + +void SoftBody3DSW::reoptimize_link_order() { + const int reop_not_dependent = -1; + const int reop_node_complete = -2; + + uint32_t i, link_count = links.size(), node_count = nodes.size(); + Link *lr; + int ar, br; + Node *node0 = &(nodes[0]); + Node *node1 = &(nodes[1]); + LinkDepsPtr link_dep; + int ready_list_head, ready_list_tail, link_num, link_dep_frees, dep_link; + + // Allocate temporary buffers. + int *node_written_at = memnew_arr(int, node_count + 1); // What link calculation produced this node's current values? + int *link_dep_A = memnew_arr(int, link_count); // Link calculation input is dependent upon prior calculation #N + int *link_dep_B = memnew_arr(int, link_count); + int *ready_list = memnew_arr(int, link_count); // List of ready-to-process link calculations (# of links, maximum) + LinkDeps *link_dep_free_list = memnew_arr(LinkDeps, 2 * link_count); // Dependent-on-me list elements (2x# of links, maximum) + LinkDepsPtr *link_dep_list_starts = memnew_arr(LinkDepsPtr, link_count); // Start nodes of dependent-on-me lists, one for each link + + // Copy the original, unsorted links to a side buffer. + Link *link_buffer = memnew_arr(Link, link_count); + memcpy(link_buffer, &(links[0]), sizeof(Link) * link_count); + + // Clear out the node setup and ready list. + for (i = 0; i < node_count + 1; i++) { + node_written_at[i] = reop_not_dependent; + } + for (i = 0; i < link_count; i++) { + link_dep_list_starts[i] = nullptr; + } + ready_list_head = ready_list_tail = link_dep_frees = 0; + + // Initial link analysis to set up data structures. + for (i = 0; i < link_count; i++) { + // Note which prior link calculations we are dependent upon & build up dependence lists. + lr = &(links[i]); + ar = (lr->n[0] - node0) / (node1 - node0); + br = (lr->n[1] - node0) / (node1 - node0); + if (node_written_at[ar] > reop_not_dependent) { + link_dep_A[i] = node_written_at[ar]; + link_dep = &link_dep_free_list[link_dep_frees++]; + link_dep->value = i; + link_dep->next = link_dep_list_starts[node_written_at[ar]]; + link_dep_list_starts[node_written_at[ar]] = link_dep; + } else { + link_dep_A[i] = reop_not_dependent; + } + if (node_written_at[br] > reop_not_dependent) { + link_dep_B[i] = node_written_at[br]; + link_dep = &link_dep_free_list[link_dep_frees++]; + link_dep->value = -(int)(i + 1); + link_dep->next = link_dep_list_starts[node_written_at[br]]; + link_dep_list_starts[node_written_at[br]] = link_dep; + } else { + link_dep_B[i] = reop_not_dependent; + } + + // Add this link to the initial ready list, if it is not dependent on any other links. + if ((link_dep_A[i] == reop_not_dependent) && (link_dep_B[i] == reop_not_dependent)) { + ready_list[ready_list_tail++] = i; + link_dep_A[i] = link_dep_B[i] = reop_node_complete; // Probably not needed now. + } + + // Update the nodes to mark which ones are calculated by this link. + node_written_at[ar] = node_written_at[br] = i; + } + + // Process the ready list and create the sorted list of links: + // -- By treating the ready list as a queue, we maximize the distance between any + // inter-dependent node calculations. + // -- All other (non-related) nodes in the ready list will automatically be inserted + // in between each set of inter-dependent link calculations by this loop. + i = 0; + while (ready_list_head != ready_list_tail) { + // Use ready list to select the next link to process. + link_num = ready_list[ready_list_head++]; + // Copy the next-to-calculate link back into the original link array. + links[i++] = link_buffer[link_num]; + + // Free up any link inputs that are dependent on this one. + link_dep = link_dep_list_starts[link_num]; + while (link_dep) { + dep_link = link_dep->value; + if (dep_link >= 0) { + link_dep_A[dep_link] = reop_not_dependent; + } else { + dep_link = -dep_link - 1; + link_dep_B[dep_link] = reop_not_dependent; + } + // Add this dependent link calculation to the ready list if *both* inputs are clear. + if ((link_dep_A[dep_link] == reop_not_dependent) && (link_dep_B[dep_link] == reop_not_dependent)) { + ready_list[ready_list_tail++] = dep_link; + link_dep_A[dep_link] = link_dep_B[dep_link] = reop_node_complete; // Probably not needed now. + } + link_dep = link_dep->next; + } + } + + // Delete the temporary buffers. + memdelete_arr(node_written_at); + memdelete_arr(link_dep_A); + memdelete_arr(link_dep_B); + memdelete_arr(ready_list); + memdelete_arr(link_dep_free_list); + memdelete_arr(link_dep_list_starts); + memdelete_arr(link_buffer); +} + +void SoftBody3DSW::append_link(uint32_t p_node1, uint32_t p_node2) { + if (p_node1 == p_node2) { + return; + } + + Node *node1 = &nodes[p_node1]; + Node *node2 = &nodes[p_node2]; + + Link link; + link.n[0] = node1; + link.n[1] = node2; + link.rl = (node1->x - node2->x).length(); + + links.push_back(link); +} + +void SoftBody3DSW::append_face(uint32_t p_node1, uint32_t p_node2, uint32_t p_node3) { + if (p_node1 == p_node2) { + return; + } + if (p_node1 == p_node3) { + return; + } + if (p_node2 == p_node3) { + return; + } + + Node *node1 = &nodes[p_node1]; + Node *node2 = &nodes[p_node2]; + Node *node3 = &nodes[p_node3]; + + Face face; + face.n[0] = node1; + face.n[1] = node2; + face.n[2] = node3; + + face.index = faces.size(); + + faces.push_back(face); +} + +void SoftBody3DSW::set_iteration_count(int p_val) { + iteration_count = p_val; +} + +void SoftBody3DSW::set_total_mass(real_t p_val) { + ERR_FAIL_COND(p_val < 0.0); + + inv_total_mass = 1.0 / p_val; + real_t mass_factor = total_mass * inv_total_mass; + total_mass = p_val; + + uint32_t node_count = nodes.size(); + for (uint32_t node_index = 0; node_index < node_count; ++node_index) { + Node &node = nodes[node_index]; + node.im *= mass_factor; + } + + update_constants(); +} + +void SoftBody3DSW::set_collision_margin(real_t p_val) { + collision_margin = p_val; +} + +void SoftBody3DSW::set_linear_stiffness(real_t p_val) { + linear_stiffness = p_val; +} + +void SoftBody3DSW::set_pressure_coefficient(real_t p_val) { + pressure_coefficient = p_val; +} + +void SoftBody3DSW::set_damping_coefficient(real_t p_val) { + damping_coefficient = p_val; +} + +void SoftBody3DSW::set_drag_coefficient(real_t p_val) { + drag_coefficient = p_val; +} + +void SoftBody3DSW::add_velocity(const Vector3 &p_velocity) { + for (uint32_t i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + if (node.im > 0) { + node.v += p_velocity; + } + } +} + +void SoftBody3DSW::apply_forces() { + if (pressure_coefficient < CMP_EPSILON) { + return; + } + + if (nodes.is_empty()) { + return; + } + + uint32_t i, ni; + + // Calculate volume. + real_t volume = 0.0; + const Vector3 &org = nodes[0].x; + for (i = 0, ni = faces.size(); i < ni; ++i) { + const Face &face = faces[i]; + volume += vec3_dot(face.n[0]->x - org, vec3_cross(face.n[1]->x - org, face.n[2]->x - org)); + } + volume /= 6.0; + + // Apply per node forces. + real_t ivolumetp = 1.0 / Math::abs(volume) * pressure_coefficient; + for (i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + if (node.im > 0) { + node.f += node.n * (node.area * ivolumetp); + } + } +} + +void SoftBody3DSW::predict_motion(real_t p_delta) { + const real_t inv_delta = 1.0 / p_delta; + + ERR_FAIL_COND(!get_space()); + + Area3DSW *def_area = get_space()->get_default_area(); + ERR_FAIL_COND(!def_area); + + // Apply forces. + Vector3 gravity = def_area->get_gravity_vector() * def_area->get_gravity(); + add_velocity(gravity * p_delta); + apply_forces(); + + // Avoid soft body from 'exploding' so use some upper threshold of maximum motion + // that a node can travel per frame. + const real_t max_displacement = 1000.0; + real_t clamp_delta_v = max_displacement * inv_delta; + + // Integrate. + uint32_t i, ni; + for (i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + node.q = node.x; + Vector3 delta_v = node.f * node.im * p_delta; + for (int c = 0; c < 3; c++) { + delta_v[c] = CLAMP(delta_v[c], -clamp_delta_v, clamp_delta_v); + } + node.v += delta_v; + node.x += node.v * p_delta; + node.f = Vector3(); + } + + // Bounds and tree update. + update_bounds(); + + // Node tree update. + for (i = 0, ni = nodes.size(); i < ni; ++i) { + const Node &node = nodes[i]; + + AABB node_aabb(node.x, Vector3()); + node_aabb.expand_to(node.x + node.v * p_delta); + node_aabb.grow_by(collision_margin); + + node_tree.update(node.leaf, node_aabb); + } + + // Face tree update. + if (!face_tree.is_empty()) { + update_face_tree(p_delta); + } + + // Optimize node tree. + node_tree.optimize_incremental(1); + face_tree.optimize_incremental(1); +} + +void SoftBody3DSW::solve_constraints(real_t p_delta) { + const real_t inv_delta = 1.0 / p_delta; + + uint32_t i, ni; + + for (i = 0, ni = links.size(); i < ni; ++i) { + Link &link = links[i]; + link.c3 = link.n[1]->q - link.n[0]->q; + link.c2 = 1 / (link.c3.length_squared() * link.c0); + } + + // Solve velocities. + for (i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + node.x = node.q + node.v * p_delta; + } + + // Solve positions. + for (int isolve = 0; isolve < iteration_count; ++isolve) { + const real_t ti = isolve / (real_t)iteration_count; + solve_links(1.0, ti); + } + const real_t vc = (1.0 - damping_coefficient) * inv_delta; + for (i = 0, ni = nodes.size(); i < ni; ++i) { + Node &node = nodes[i]; + + node.x += node.bv * p_delta; + node.bv = Vector3(); + + node.v = (node.x - node.q) * vc; + + node.q = node.x; + } + + update_normals(); +} + +void SoftBody3DSW::solve_links(real_t kst, real_t ti) { + for (uint32_t i = 0, ni = links.size(); i < ni; ++i) { + Link &link = links[i]; + if (link.c0 > 0) { + Node &node_a = *link.n[0]; + Node &node_b = *link.n[1]; + const Vector3 del = node_b.x - node_a.x; + const real_t len = del.length_squared(); + if (link.c1 + len > CMP_EPSILON) { + const real_t k = ((link.c1 - len) / (link.c0 * (link.c1 + len))) * kst; + node_a.x -= del * (k * node_a.im); + node_b.x += del * (k * node_b.im); + } + } + } +} + +struct AABBQueryResult { + const SoftBody3DSW *soft_body = nullptr; + void *userdata = nullptr; + SoftBody3DSW::QueryResultCallback result_callback = nullptr; + + _FORCE_INLINE_ bool operator()(void *p_data) { + return result_callback(soft_body->get_node_index(p_data), userdata); + }; +}; + +void SoftBody3DSW::query_aabb(const AABB &p_aabb, SoftBody3DSW::QueryResultCallback p_result_callback, void *p_userdata) { + AABBQueryResult query_result; + query_result.soft_body = this; + query_result.result_callback = p_result_callback; + query_result.userdata = p_userdata; + + node_tree.aabb_query(p_aabb, query_result); +} + +struct RayQueryResult { + const SoftBody3DSW *soft_body = nullptr; + void *userdata = nullptr; + SoftBody3DSW::QueryResultCallback result_callback = nullptr; + + _FORCE_INLINE_ bool operator()(void *p_data) { + return result_callback(soft_body->get_face_index(p_data), userdata); + }; +}; + +void SoftBody3DSW::query_ray(const Vector3 &p_from, const Vector3 &p_to, SoftBody3DSW::QueryResultCallback p_result_callback, void *p_userdata) { + if (face_tree.is_empty()) { + initialize_face_tree(); + } + + RayQueryResult query_result; + query_result.soft_body = this; + query_result.result_callback = p_result_callback; + query_result.userdata = p_userdata; + + face_tree.ray_query(p_from, p_to, query_result); +} + +void SoftBody3DSW::initialize_face_tree() { + face_tree.clear(); + for (uint32_t i = 0; i < faces.size(); ++i) { + Face &face = faces[i]; + + AABB face_aabb; + + face_aabb.position = face.n[0]->x; + face_aabb.expand_to(face.n[1]->x); + face_aabb.expand_to(face.n[2]->x); + + face_aabb.grow_by(collision_margin); + + face.leaf = face_tree.insert(face_aabb, &face); + } +} + +void SoftBody3DSW::update_face_tree(real_t p_delta) { + for (uint32_t i = 0; i < faces.size(); ++i) { + const Face &face = faces[i]; + + AABB face_aabb; + + const Node *node0 = face.n[0]; + face_aabb.position = node0->x; + face_aabb.expand_to(node0->x + node0->v * p_delta); + + const Node *node1 = face.n[1]; + face_aabb.expand_to(node1->x); + face_aabb.expand_to(node1->x + node1->v * p_delta); + + const Node *node2 = face.n[2]; + face_aabb.expand_to(node2->x); + face_aabb.expand_to(node2->x + node2->v * p_delta); + + face_aabb.grow_by(collision_margin); + + face_tree.update(face.leaf, face_aabb); + } +} + +void SoftBody3DSW::initialize_shape(bool p_force_move) { + if (get_shape_count() == 0) { + SoftBodyShape3DSW *soft_body_shape = memnew(SoftBodyShape3DSW(this)); + add_shape(soft_body_shape); + } else if (p_force_move) { + SoftBodyShape3DSW *soft_body_shape = static_cast<SoftBodyShape3DSW *>(get_shape(0)); + soft_body_shape->update_bounds(); + } +} + +void SoftBody3DSW::deinitialize_shape() { + if (get_shape_count() > 0) { + Shape3DSW *shape = get_shape(0); + remove_shape(shape); + memdelete(shape); + } +} + +void SoftBody3DSW::destroy() { + map_visual_to_physics.clear(); + + node_tree.clear(); + face_tree.clear(); + + nodes.clear(); + links.clear(); + faces.clear(); + + bounds = AABB(); + deinitialize_shape(); +} + +void SoftBodyShape3DSW::update_bounds() { + ERR_FAIL_COND(!soft_body); + + AABB collision_aabb = soft_body->get_bounds(); + collision_aabb.grow_by(soft_body->get_collision_margin()); + configure(collision_aabb); +} + +SoftBodyShape3DSW::SoftBodyShape3DSW(SoftBody3DSW *p_soft_body) { + soft_body = p_soft_body; + update_bounds(); +} + +struct _SoftBodyIntersectSegmentInfo { + const SoftBody3DSW *soft_body = nullptr; + Vector3 from; + Vector3 dir; + Vector3 hit_position; + uint32_t hit_face_index = -1; + real_t hit_dist_sq = Math_INF; + + static bool process_hit(uint32_t p_face_index, void *p_userdata) { + _SoftBodyIntersectSegmentInfo &query_info = *(_SoftBodyIntersectSegmentInfo *)(p_userdata); + + Vector3 points[3]; + query_info.soft_body->get_face_points(p_face_index, points[0], points[1], points[2]); + + Vector3 result; + if (Geometry3D::ray_intersects_triangle(query_info.from, query_info.dir, points[0], points[1], points[2], &result)) { + real_t dist_sq = query_info.from.distance_squared_to(result); + if (dist_sq < query_info.hit_dist_sq) { + query_info.hit_dist_sq = dist_sq; + query_info.hit_position = result; + query_info.hit_face_index = p_face_index; + } + } + + // Continue with the query. + return false; + } +}; + +bool SoftBodyShape3DSW::intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const { + _SoftBodyIntersectSegmentInfo query_info; + query_info.soft_body = soft_body; + query_info.from = p_begin; + query_info.dir = (p_end - p_begin).normalized(); + + soft_body->query_ray(p_begin, p_end, _SoftBodyIntersectSegmentInfo::process_hit, &query_info); + + if (query_info.hit_dist_sq != Math_INF) { + r_result = query_info.hit_position; + r_normal = soft_body->get_face_normal(query_info.hit_face_index); + return true; + } + + return false; +} + +bool SoftBodyShape3DSW::intersect_point(const Vector3 &p_point) const { + return false; +} + +Vector3 SoftBodyShape3DSW::get_closest_point_to(const Vector3 &p_point) const { + ERR_FAIL_V_MSG(Vector3(), "Get closest point is not supported for soft bodies."); +} diff --git a/servers/physics_3d/soft_body_3d_sw.h b/servers/physics_3d/soft_body_3d_sw.h new file mode 100644 index 0000000000..6c0451ff45 --- /dev/null +++ b/servers/physics_3d/soft_body_3d_sw.h @@ -0,0 +1,247 @@ +/*************************************************************************/ +/* soft_body_3d_sw.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef SOFT_BODY_3D_SW_H +#define SOFT_BODY_3D_SW_H + +#include "collision_object_3d_sw.h" + +#include "core/math/aabb.h" +#include "core/math/dynamic_bvh.h" +#include "core/math/vector3.h" +#include "core/templates/local_vector.h" +#include "core/templates/set.h" +#include "core/templates/vset.h" +#include "scene/resources/mesh.h" + +class Constraint3DSW; + +class SoftBody3DSW : public CollisionObject3DSW { + Ref<Mesh> soft_mesh; + + struct Node { + Vector3 s; // Source position + Vector3 x; // Position + Vector3 q; // Previous step position/Test position + Vector3 f; // Force accumulator + Vector3 v; // Velocity + Vector3 bv; // Biased Velocity + Vector3 n; // Normal + real_t area = 0.0; // Area + real_t im = 0.0; // 1/mass + DynamicBVH::ID leaf; // Leaf data + uint32_t index = 0; + }; + + struct Link { + Vector3 c3; // gradient + Node *n[2] = { nullptr, nullptr }; // Node pointers + real_t rl = 0.0; // Rest length + real_t c0 = 0.0; // (ima+imb)*kLST + real_t c1 = 0.0; // rl^2 + real_t c2 = 0.0; // |gradient|^2/c0 + }; + + struct Face { + Node *n[3] = { nullptr, nullptr, nullptr }; // Node pointers + Vector3 normal; // Normal + real_t ra = 0.0; // Rest area + DynamicBVH::ID leaf; // Leaf data + uint32_t index = 0; + }; + + LocalVector<Node> nodes; + LocalVector<Link> links; + LocalVector<Face> faces; + + DynamicBVH node_tree; + DynamicBVH face_tree; + + LocalVector<uint32_t> map_visual_to_physics; + + AABB bounds; + + real_t collision_margin = 0.05; + + real_t total_mass = 1.0; + real_t inv_total_mass = 1.0; + + int iteration_count = 5; + real_t linear_stiffness = 0.5; // [0,1] + real_t pressure_coefficient = 0.0; // [-inf,+inf] + real_t damping_coefficient = 0.01; // [0,1] + real_t drag_coefficient = 0.0; // [0,1] + LocalVector<int> pinned_vertices; + + SelfList<SoftBody3DSW> active_list; + + Set<Constraint3DSW *> constraints; + + VSet<RID> exceptions; + +public: + SoftBody3DSW(); + + const AABB &get_bounds() const { return bounds; } + + void set_state(PhysicsServer3D::BodyState p_state, const Variant &p_variant); + Variant get_state(PhysicsServer3D::BodyState p_state) const; + + _FORCE_INLINE_ void add_constraint(Constraint3DSW *p_constraint) { constraints.insert(p_constraint); } + _FORCE_INLINE_ void remove_constraint(Constraint3DSW *p_constraint) { constraints.erase(p_constraint); } + _FORCE_INLINE_ const Set<Constraint3DSW *> &get_constraints() const { return constraints; } + _FORCE_INLINE_ void clear_constraints() { constraints.clear(); } + + _FORCE_INLINE_ void add_exception(const RID &p_exception) { exceptions.insert(p_exception); } + _FORCE_INLINE_ void remove_exception(const RID &p_exception) { exceptions.erase(p_exception); } + _FORCE_INLINE_ bool has_exception(const RID &p_exception) const { return exceptions.has(p_exception); } + _FORCE_INLINE_ const VSet<RID> &get_exceptions() const { return exceptions; } + + virtual void set_space(Space3DSW *p_space); + + void set_mesh(const Ref<Mesh> &p_mesh); + + void update_rendering_server(RenderingServerHandler *p_rendering_server_handler); + + Vector3 get_vertex_position(int p_index) const; + void set_vertex_position(int p_index, const Vector3 &p_position); + + void pin_vertex(int p_index); + void unpin_vertex(int p_index); + void unpin_all_vertices(); + bool is_vertex_pinned(int p_index) const; + + uint32_t get_node_count() const; + real_t get_node_inv_mass(uint32_t p_node_index) const; + Vector3 get_node_position(uint32_t p_node_index) const; + Vector3 get_node_velocity(uint32_t p_node_index) const; + Vector3 get_node_biased_velocity(uint32_t p_node_index) const; + void apply_node_impulse(uint32_t p_node_index, const Vector3 &p_impulse); + void apply_node_bias_impulse(uint32_t p_node_index, const Vector3 &p_impulse); + + uint32_t get_face_count() const; + void get_face_points(uint32_t p_face_index, Vector3 &r_point_1, Vector3 &r_point_2, Vector3 &r_point_3) const; + Vector3 get_face_normal(uint32_t p_face_index) const; + + void set_iteration_count(int p_val); + _FORCE_INLINE_ real_t get_iteration_count() const { return iteration_count; } + + void set_total_mass(real_t p_val); + _FORCE_INLINE_ real_t get_total_mass() const { return total_mass; } + _FORCE_INLINE_ real_t get_total_inv_mass() const { return inv_total_mass; } + + void set_collision_margin(real_t p_val); + _FORCE_INLINE_ real_t get_collision_margin() const { return collision_margin; } + + void set_linear_stiffness(real_t p_val); + _FORCE_INLINE_ real_t get_linear_stiffness() const { return linear_stiffness; } + + void set_pressure_coefficient(real_t p_val); + _FORCE_INLINE_ real_t get_pressure_coefficient() const { return pressure_coefficient; } + + void set_damping_coefficient(real_t p_val); + _FORCE_INLINE_ real_t get_damping_coefficient() const { return damping_coefficient; } + + void set_drag_coefficient(real_t p_val); + _FORCE_INLINE_ real_t get_drag_coefficient() const { return drag_coefficient; } + + void predict_motion(real_t p_delta); + void solve_constraints(real_t p_delta); + + _FORCE_INLINE_ uint32_t get_node_index(void *p_node) const { return ((Node *)p_node)->index; } + _FORCE_INLINE_ uint32_t get_face_index(void *p_face) const { return ((Face *)p_face)->index; } + + // Return true to stop the query. + // p_index is the node index for AABB query, face index for Ray query. + typedef bool (*QueryResultCallback)(uint32_t p_index, void *p_userdata); + + void query_aabb(const AABB &p_aabb, QueryResultCallback p_result_callback, void *p_userdata); + void query_ray(const Vector3 &p_from, const Vector3 &p_to, QueryResultCallback p_result_callback, void *p_userdata); + +protected: + virtual void _shapes_changed(); + +private: + void update_normals(); + void update_bounds(); + void update_constants(); + void update_area(); + void reset_link_rest_lengths(); + void update_link_constants(); + + void apply_nodes_transform(const Transform &p_transform); + + void add_velocity(const Vector3 &p_velocity); + + void apply_forces(); + + bool create_from_trimesh(const Vector<int> &p_indices, const Vector<Vector3> &p_vertices); + void generate_bending_constraints(int p_distance); + void reoptimize_link_order(); + void append_link(uint32_t p_node1, uint32_t p_node2); + void append_face(uint32_t p_node1, uint32_t p_node2, uint32_t p_node3); + + void solve_links(real_t kst, real_t ti); + + void initialize_face_tree(); + void update_face_tree(real_t p_delta); + + void initialize_shape(bool p_force_move = true); + void deinitialize_shape(); + + void destroy(); +}; + +class SoftBodyShape3DSW : public Shape3DSW { + SoftBody3DSW *soft_body = nullptr; + +public: + SoftBody3DSW *get_soft_body() const { return soft_body; } + + virtual PhysicsServer3D::ShapeType get_type() const { return PhysicsServer3D::SHAPE_SOFT_BODY; } + virtual void project_range(const Vector3 &p_normal, const Transform &p_transform, real_t &r_min, real_t &r_max) const { r_min = r_max = 0.0; } + virtual Vector3 get_support(const Vector3 &p_normal) const { return Vector3(); } + virtual void get_supports(const Vector3 &p_normal, int p_max, Vector3 *r_supports, int &r_amount, FeatureType &r_type) const { r_amount = 0; } + + virtual bool intersect_segment(const Vector3 &p_begin, const Vector3 &p_end, Vector3 &r_result, Vector3 &r_normal) const; + virtual bool intersect_point(const Vector3 &p_point) const; + virtual Vector3 get_closest_point_to(const Vector3 &p_point) const; + virtual Vector3 get_moment_of_inertia(real_t p_mass) const { return Vector3(); } + + virtual void set_data(const Variant &p_data) {} + virtual Variant get_data() const { return Variant(); } + + void update_bounds(); + + SoftBodyShape3DSW(SoftBody3DSW *p_soft_body); + ~SoftBodyShape3DSW() {} +}; + +#endif // SOFT_BODY_3D_SW_H diff --git a/servers/physics_3d/space_3d_sw.cpp b/servers/physics_3d/space_3d_sw.cpp index 2b2b5122da..2df824b320 100644 --- a/servers/physics_3d/space_3d_sw.cpp +++ b/servers/physics_3d/space_3d_sw.cpp @@ -47,6 +47,10 @@ _FORCE_INLINE_ static bool _can_collide_with(CollisionObject3DSW *p_object, uint return false; } + if (p_object->get_type() == CollisionObject3DSW::TYPE_SOFT_BODY && !p_collide_with_bodies) { + return false; + } + return true; } @@ -181,7 +185,7 @@ int PhysicsDirectSpaceState3DSW::intersect_shape(const RID &p_shape, const Trans return 0; } - Shape3DSW *shape = static_cast<PhysicsServer3DSW *>(PhysicsServer3D::get_singleton())->shape_owner.getornull(p_shape); + Shape3DSW *shape = PhysicsServer3DSW::singletonsw->shape_owner.getornull(p_shape); ERR_FAIL_COND_V(!shape, 0); AABB aabb = p_xform.xform(shape->get_aabb()); @@ -210,6 +214,10 @@ int PhysicsDirectSpaceState3DSW::intersect_shape(const RID &p_shape, const Trans const CollisionObject3DSW *col_obj = space->intersection_query_results[i]; int shape_idx = space->intersection_query_subindex_results[i]; + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + if (!CollisionSolver3DSW::solve_static(shape, p_xform, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), nullptr, nullptr, nullptr, p_margin, 0)) { continue; } @@ -232,7 +240,7 @@ int PhysicsDirectSpaceState3DSW::intersect_shape(const RID &p_shape, const Trans } bool PhysicsDirectSpaceState3DSW::cast_motion(const RID &p_shape, const Transform &p_xform, const Vector3 &p_motion, real_t p_margin, real_t &p_closest_safe, real_t &p_closest_unsafe, const Set<RID> &p_exclude, uint32_t p_collision_mask, bool p_collide_with_bodies, bool p_collide_with_areas, ShapeRestInfo *r_info) { - Shape3DSW *shape = static_cast<PhysicsServer3DSW *>(PhysicsServer3D::get_singleton())->shape_owner.getornull(p_shape); + Shape3DSW *shape = PhysicsServer3DSW::singletonsw->shape_owner.getornull(p_shape); ERR_FAIL_COND_V(!shape, false); AABB aabb = p_xform.xform(shape->get_aabb()); @@ -265,6 +273,10 @@ bool PhysicsDirectSpaceState3DSW::cast_motion(const RID &p_shape, const Transfor const CollisionObject3DSW *col_obj = space->intersection_query_results[i]; int shape_idx = space->intersection_query_subindex_results[i]; + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + Vector3 point_A, point_B; Vector3 sep_axis = p_motion.normalized(); @@ -274,11 +286,11 @@ bool PhysicsDirectSpaceState3DSW::cast_motion(const RID &p_shape, const Transfor continue; } - //test initial overlap + //test initial overlap, ignore objects it's inside of. sep_axis = p_motion.normalized(); if (!CollisionSolver3DSW::solve_distance(shape, p_xform, col_obj->get_shape(shape_idx), col_obj_xform, point_A, point_B, aabb, &sep_axis)) { - return false; + continue; } //just do kinematic solving @@ -324,7 +336,8 @@ bool PhysicsDirectSpaceState3DSW::cast_motion(const RID &p_shape, const Transfor best_first = false; if (col_obj->get_type() == CollisionObject3DSW::TYPE_BODY) { const Body3DSW *body = static_cast<const Body3DSW *>(col_obj); - r_info->linear_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(body->get_transform().origin - closest_B); + Vector3 rel_vec = closest_B - (body->get_transform().origin + body->get_center_of_mass()); + r_info->linear_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(rel_vec); } } } @@ -340,7 +353,7 @@ bool PhysicsDirectSpaceState3DSW::collide_shape(RID p_shape, const Transform &p_ return false; } - Shape3DSW *shape = static_cast<PhysicsServer3DSW *>(PhysicsServer3D::get_singleton())->shape_owner.getornull(p_shape); + Shape3DSW *shape = PhysicsServer3DSW::singletonsw->shape_owner.getornull(p_shape); ERR_FAIL_COND_V(!shape, 0); AABB aabb = p_shape_xform.xform(shape->get_aabb()); @@ -365,12 +378,17 @@ bool PhysicsDirectSpaceState3DSW::collide_shape(RID p_shape, const Transform &p_ } const CollisionObject3DSW *col_obj = space->intersection_query_results[i]; - int shape_idx = space->intersection_query_subindex_results[i]; if (p_exclude.has(col_obj->get_self())) { continue; } + int shape_idx = space->intersection_query_subindex_results[i]; + + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + if (CollisionSolver3DSW::solve_static(shape, p_shape_xform, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), cbkres, cbkptr, nullptr, p_margin)) { collided = true; } @@ -384,6 +402,8 @@ bool PhysicsDirectSpaceState3DSW::collide_shape(RID p_shape, const Transform &p_ struct _RestCallbackData { const CollisionObject3DSW *object; const CollisionObject3DSW *best_object; + int local_shape; + int best_local_shape; int shape; int best_shape; Vector3 best_contact; @@ -392,7 +412,7 @@ struct _RestCallbackData { real_t min_allowed_depth; }; -static void _rest_cbk_result(const Vector3 &p_point_A, const Vector3 &p_point_B, void *p_userdata) { +static void _rest_cbk_result(const Vector3 &p_point_A, int p_index_A, const Vector3 &p_point_B, int p_index_B, void *p_userdata) { _RestCallbackData *rd = (_RestCallbackData *)p_userdata; Vector3 contact_rel = p_point_B - p_point_A; @@ -409,10 +429,11 @@ static void _rest_cbk_result(const Vector3 &p_point_A, const Vector3 &p_point_B, rd->best_normal = contact_rel / len; rd->best_object = rd->object; rd->best_shape = rd->shape; + rd->best_local_shape = rd->local_shape; } bool PhysicsDirectSpaceState3DSW::rest_info(RID p_shape, const Transform &p_shape_xform, real_t p_margin, ShapeRestInfo *r_info, const Set<RID> &p_exclude, uint32_t p_collision_mask, bool p_collide_with_bodies, bool p_collide_with_areas) { - Shape3DSW *shape = static_cast<PhysicsServer3DSW *>(PhysicsServer3D::get_singleton())->shape_owner.getornull(p_shape); + Shape3DSW *shape = PhysicsServer3DSW::singletonsw->shape_owner.getornull(p_shape); ERR_FAIL_COND_V(!shape, 0); AABB aabb = p_shape_xform.xform(shape->get_aabb()); @@ -432,12 +453,17 @@ bool PhysicsDirectSpaceState3DSW::rest_info(RID p_shape, const Transform &p_shap } const CollisionObject3DSW *col_obj = space->intersection_query_results[i]; - int shape_idx = space->intersection_query_subindex_results[i]; if (p_exclude.has(col_obj->get_self())) { continue; } + int shape_idx = space->intersection_query_subindex_results[i]; + + if (col_obj->is_shape_set_as_disabled(shape_idx)) { + continue; + } + rcd.object = col_obj; rcd.shape = shape_idx; bool sc = CollisionSolver3DSW::solve_static(shape, p_shape_xform, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), _rest_cbk_result, &rcd, nullptr, p_margin); @@ -457,8 +483,8 @@ bool PhysicsDirectSpaceState3DSW::rest_info(RID p_shape, const Transform &p_shap r_info->rid = rcd.best_object->get_self(); if (rcd.best_object->get_type() == CollisionObject3DSW::TYPE_BODY) { const Body3DSW *body = static_cast<const Body3DSW *>(rcd.best_object); - r_info->linear_velocity = body->get_linear_velocity() + - (body->get_angular_velocity()).cross(body->get_transform().origin - rcd.best_contact); // * mPos); + Vector3 rel_vec = rcd.best_contact - (body->get_transform().origin + body->get_center_of_mass()); + r_info->linear_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(rel_vec); } else { r_info->linear_velocity = Vector3(); @@ -468,15 +494,15 @@ bool PhysicsDirectSpaceState3DSW::rest_info(RID p_shape, const Transform &p_shap } Vector3 PhysicsDirectSpaceState3DSW::get_closest_point_to_object_volume(RID p_object, const Vector3 p_point) const { - CollisionObject3DSW *obj = PhysicsServer3DSW::singleton->area_owner.getornull(p_object); + CollisionObject3DSW *obj = PhysicsServer3DSW::singletonsw->area_owner.getornull(p_object); if (!obj) { - obj = PhysicsServer3DSW::singleton->body_owner.getornull(p_object); + obj = PhysicsServer3DSW::singletonsw->body_owner.getornull(p_object); } ERR_FAIL_COND_V(!obj, Vector3()); ERR_FAIL_COND_V(obj->get_space() != space, Vector3()); - float min_distance = 1e20; + real_t min_distance = 1e20; Vector3 min_point; bool shapes_found = false; @@ -492,7 +518,7 @@ Vector3 PhysicsDirectSpaceState3DSW::get_closest_point_to_object_volume(RID p_ob Vector3 point = shape->get_closest_point_to(shape_xform.affine_inverse().xform(p_point)); point = shape_xform.xform(point); - float dist = point.distance_to(p_point); + real_t dist = point.distance_to(p_point); if (dist < min_distance) { min_distance = dist; min_point = point; @@ -523,6 +549,8 @@ int Space3DSW::_cull_aabb_for_body(Body3DSW *p_body, const AABB &p_aabb) { keep = false; } else if (intersection_query_results[i]->get_type() == CollisionObject3DSW::TYPE_AREA) { keep = false; + } else if (intersection_query_results[i]->get_type() == CollisionObject3DSW::TYPE_SOFT_BODY) { + keep = false; } else if ((static_cast<Body3DSW *>(intersection_query_results[i])->test_collision_mask(p_body)) == 0) { keep = false; } else if (static_cast<Body3DSW *>(intersection_query_results[i])->has_exception(p_body->get_self()) || p_body->has_exception(intersection_query_results[i]->get_self())) { @@ -649,9 +677,9 @@ int Space3DSW::test_body_ray_separation(Body3DSW *p_body, const Transform &p_tra Vector3 a = sr[k * 2 + 0]; Vector3 b = sr[k * 2 + 1]; - recover_motion += (b - a) * 0.4; + recover_motion += (b - a) / cbk.amount; - float depth = a.distance_to(b); + real_t depth = a.distance_to(b); if (depth > result.collision_depth) { result.collision_depth = depth; result.collision_point = b; @@ -663,10 +691,8 @@ int Space3DSW::test_body_ray_separation(Body3DSW *p_body, const Transform &p_tra //result.collider_metadata = col_obj->get_shape_metadata(shape_idx); if (col_obj->get_type() == CollisionObject3DSW::TYPE_BODY) { Body3DSW *body = (Body3DSW *)col_obj; - - Vector3 rel_vec = b - body->get_transform().get_origin(); - //result.collider_velocity = Vector3(-body->get_angular_velocity() * rel_vec.y, body->get_angular_velocity() * rel_vec.x) + body->get_linear_velocity(); - result.collider_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(body->get_transform().origin - rel_vec); // * mPos); + Vector3 rel_vec = b - (body->get_transform().origin + body->get_center_of_mass()); + result.collider_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(rel_vec); } } } @@ -739,8 +765,13 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons body_aabb = p_from.xform(p_body->get_inv_transform().xform(body_aabb)); body_aabb = body_aabb.grow(p_margin); + real_t motion_length = p_motion.length(); + Vector3 motion_normal = p_motion / motion_length; + Transform body_transform = p_from; + bool recovered = false; + { //STEP 1, FREE BODY IF STUCK @@ -791,7 +822,17 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons for (int i = 0; i < cbk.amount; i++) { Vector3 a = sr[i * 2 + 0]; Vector3 b = sr[i * 2 + 1]; - recover_motion += (b - a) * 0.4; + + // Compute plane on b towards a. + Vector3 n = (a - b).normalized(); + real_t d = n.dot(b); + + // Compute depth on recovered motion. + real_t depth = n.dot(a + recover_motion) - d; + if (depth > 0.0) { + // Only recover if there is penetration. + recover_motion -= n * depth * 0.4; + } } if (recover_motion == Vector3()) { @@ -799,6 +840,8 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons break; } + recovered = true; + body_transform.origin += recover_motion; body_aabb.position += recover_motion; @@ -848,14 +891,14 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons //test initial overlap, does it collide if going all the way? Vector3 point_A, point_B; - Vector3 sep_axis = p_motion.normalized(); + Vector3 sep_axis = motion_normal; Transform col_obj_xform = col_obj->get_transform() * col_obj->get_shape_transform(shape_idx); //test initial overlap, does it collide if going all the way? if (CollisionSolver3DSW::solve_distance(&mshape, body_shape_xform, col_obj->get_shape(shape_idx), col_obj_xform, point_A, point_B, motion_aabb, &sep_axis)) { continue; } - sep_axis = p_motion.normalized(); + sep_axis = motion_normal; if (!CollisionSolver3DSW::solve_distance(body_shape, body_shape_xform, col_obj->get_shape(shape_idx), col_obj_xform, point_A, point_B, motion_aabb, &sep_axis)) { stuck = true; @@ -865,13 +908,12 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons //just do kinematic solving real_t low = 0; real_t hi = 1; - Vector3 mnormal = p_motion.normalized(); for (int k = 0; k < 8; k++) { //steps should be customizable.. real_t ofs = (low + hi) * 0.5; - Vector3 sep = mnormal; //important optimization for this to work fast enough + Vector3 sep = motion_normal; //important optimization for this to work fast enough mshape.motion = body_shape_xform_inv.basis.xform(p_motion * ofs); @@ -912,16 +954,11 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons } bool collided = false; - if (safe >= 1) { - //not collided - collided = false; - if (r_result) { - r_result->motion = p_motion; - r_result->remainder = Vector3(); - r_result->motion += (body_transform.get_origin() - p_from.get_origin()); + if (recovered || (safe < 1)) { + if (safe >= 1) { + best_shape = -1; //no best shape with cast, reset to -1 } - } else { //it collided, let's get the rest info in unsafe advance Transform ugt = body_transform; ugt.origin += p_motion * unsafe; @@ -930,25 +967,40 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons rcd.best_len = 0; rcd.best_object = nullptr; rcd.best_shape = 0; - rcd.min_allowed_depth = test_motion_min_contact_depth; - Transform body_shape_xform = ugt * p_body->get_shape_transform(best_shape); - Shape3DSW *body_shape = p_body->get_shape(best_shape); + // Allowed depth can't be lower than motion length, in order to handle contacts at low speed. + rcd.min_allowed_depth = MIN(motion_length, test_motion_min_contact_depth); - body_aabb.position += p_motion * unsafe; + int from_shape = best_shape != -1 ? best_shape : 0; + int to_shape = best_shape != -1 ? best_shape + 1 : p_body->get_shape_count(); - int amount = _cull_aabb_for_body(p_body, body_aabb); + for (int j = from_shape; j < to_shape; j++) { + if (p_body->is_shape_set_as_disabled(j)) { + continue; + } - for (int i = 0; i < amount; i++) { - const CollisionObject3DSW *col_obj = intersection_query_results[i]; - int shape_idx = intersection_query_subindex_results[i]; + Transform body_shape_xform = ugt * p_body->get_shape_transform(j); + Shape3DSW *body_shape = p_body->get_shape(j); - rcd.object = col_obj; - rcd.shape = shape_idx; - bool sc = CollisionSolver3DSW::solve_static(body_shape, body_shape_xform, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), _rest_cbk_result, &rcd, nullptr, p_margin); - if (!sc) { + if (p_exclude_raycast_shapes && body_shape->get_type() == PhysicsServer3D::SHAPE_RAY) { continue; } + + body_aabb.position += p_motion * unsafe; + + int amount = _cull_aabb_for_body(p_body, body_aabb); + + for (int i = 0; i < amount; i++) { + const CollisionObject3DSW *col_obj = intersection_query_results[i]; + int shape_idx = intersection_query_subindex_results[i]; + + rcd.object = col_obj; + rcd.shape = shape_idx; + bool sc = CollisionSolver3DSW::solve_static(body_shape, body_shape_xform, col_obj->get_shape(shape_idx), col_obj->get_transform() * col_obj->get_shape_transform(shape_idx), _rest_cbk_result, &rcd, nullptr, p_margin); + if (!sc) { + continue; + } + } } if (rcd.best_len != 0) { @@ -956,15 +1008,15 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons r_result->collider = rcd.best_object->get_self(); r_result->collider_id = rcd.best_object->get_instance_id(); r_result->collider_shape = rcd.best_shape; - r_result->collision_local_shape = best_shape; + r_result->collision_local_shape = rcd.best_local_shape; r_result->collision_normal = rcd.best_normal; r_result->collision_point = rcd.best_contact; //r_result->collider_metadata = rcd.best_object->get_shape_metadata(rcd.best_shape); const Body3DSW *body = static_cast<const Body3DSW *>(rcd.best_object); - //Vector3 rel_vec = r_result->collision_point - body->get_transform().get_origin(); - // r_result->collider_velocity = Vector3(-body->get_angular_velocity() * rel_vec.y, body->get_angular_velocity() * rel_vec.x) + body->get_linear_velocity(); - r_result->collider_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(body->get_transform().origin - rcd.best_contact); // * mPos); + + Vector3 rel_vec = rcd.best_contact - (body->get_transform().origin + body->get_center_of_mass()); + r_result->collider_velocity = body->get_linear_velocity() + (body->get_angular_velocity()).cross(rel_vec); r_result->motion = safe * p_motion; r_result->remainder = p_motion - safe * p_motion; @@ -972,17 +1024,15 @@ bool Space3DSW::test_body_motion(Body3DSW *p_body, const Transform &p_from, cons } collided = true; - } else { - if (r_result) { - r_result->motion = p_motion; - r_result->remainder = Vector3(); - r_result->motion += (body_transform.get_origin() - p_from.get_origin()); - } - - collided = false; } } + if (!collided && r_result) { + r_result->motion = p_motion; + r_result->remainder = Vector3(); + r_result->motion += (body_transform.get_origin() - p_from.get_origin()); + } + return collided; } @@ -1009,14 +1059,23 @@ void *Space3DSW::_broadphase_pair(CollisionObject3DSW *A, int p_subindex_A, Coll Area3DSW *area_b = static_cast<Area3DSW *>(B); Area2Pair3DSW *area2_pair = memnew(Area2Pair3DSW(area_b, p_subindex_B, area, p_subindex_A)); return area2_pair; + } else if (type_B == CollisionObject3DSW::TYPE_SOFT_BODY) { + // Area/Soft Body, not supported. } else { Body3DSW *body = static_cast<Body3DSW *>(B); AreaPair3DSW *area_pair = memnew(AreaPair3DSW(body, p_subindex_B, area, p_subindex_A)); return area_pair; } + } else if (type_A == CollisionObject3DSW::TYPE_BODY) { + if (type_B == CollisionObject3DSW::TYPE_SOFT_BODY) { + BodySoftBodyPair3DSW *soft_pair = memnew(BodySoftBodyPair3DSW((Body3DSW *)A, p_subindex_A, (SoftBody3DSW *)B)); + return soft_pair; + } else { + BodyPair3DSW *b = memnew(BodyPair3DSW((Body3DSW *)A, p_subindex_A, (Body3DSW *)B, p_subindex_B)); + return b; + } } else { - BodyPair3DSW *b = memnew(BodyPair3DSW((Body3DSW *)A, p_subindex_A, (Body3DSW *)B, p_subindex_B)); - return b; + // Soft Body/Soft Body, not supported. } return nullptr; @@ -1099,6 +1158,18 @@ const SelfList<Area3DSW>::List &Space3DSW::get_moved_area_list() const { return area_moved_list; } +const SelfList<SoftBody3DSW>::List &Space3DSW::get_active_soft_body_list() const { + return active_soft_body_list; +} + +void Space3DSW::soft_body_add_to_active_list(SelfList<SoftBody3DSW> *p_soft_body) { + active_soft_body_list.add(p_soft_body); +} + +void Space3DSW::soft_body_remove_from_active_list(SelfList<SoftBody3DSW> *p_soft_body) { + active_soft_body_list.remove(p_soft_body); +} + void Space3DSW::call_queries() { while (state_query_list.first()) { Body3DSW *b = state_query_list.first()->self(); @@ -1211,7 +1282,7 @@ Space3DSW::Space3DSW() { constraint_bias = 0.01; body_linear_velocity_sleep_threshold = GLOBAL_DEF("physics/3d/sleep_threshold_linear", 0.1); - body_angular_velocity_sleep_threshold = GLOBAL_DEF("physics/3d/sleep_threshold_angular", (8.0 / 180.0 * Math_PI)); + body_angular_velocity_sleep_threshold = GLOBAL_DEF("physics/3d/sleep_threshold_angular", Math::deg2rad(8.0)); body_time_to_sleep = GLOBAL_DEF("physics/3d/time_before_sleep", 0.5); ProjectSettings::get_singleton()->set_custom_property_info("physics/3d/time_before_sleep", PropertyInfo(Variant::FLOAT, "physics/3d/time_before_sleep", PROPERTY_HINT_RANGE, "0,5,0.01,or_greater")); body_angular_velocity_damp_ratio = 10; diff --git a/servers/physics_3d/space_3d_sw.h b/servers/physics_3d/space_3d_sw.h index eed3d86a72..3a8f452e54 100644 --- a/servers/physics_3d/space_3d_sw.h +++ b/servers/physics_3d/space_3d_sw.h @@ -40,6 +40,7 @@ #include "core/config/project_settings.h" #include "core/templates/hash_map.h" #include "core/typedefs.h" +#include "soft_body_3d_sw.h" class PhysicsDirectSpaceState3DSW : public PhysicsDirectSpaceState3D { GDCLASS(PhysicsDirectSpaceState3DSW, PhysicsDirectSpaceState3D); @@ -82,6 +83,7 @@ private: SelfList<Body3DSW>::List state_query_list; SelfList<Area3DSW>::List monitor_query_list; SelfList<Area3DSW>::List area_moved_list; + SelfList<SoftBody3DSW>::List active_soft_body_list; static void *_broadphase_pair(CollisionObject3DSW *A, int p_subindex_A, CollisionObject3DSW *B, int p_subindex_B, void *p_self); static void _broadphase_unpair(CollisionObject3DSW *A, int p_subindex_A, CollisionObject3DSW *B, int p_subindex_B, void *p_data, void *p_self); @@ -145,6 +147,10 @@ public: void area_remove_from_moved_list(SelfList<Area3DSW> *p_area); const SelfList<Area3DSW>::List &get_moved_area_list() const; + const SelfList<SoftBody3DSW>::List &get_active_soft_body_list() const; + void soft_body_add_to_active_list(SelfList<SoftBody3DSW> *p_soft_body); + void soft_body_remove_from_active_list(SelfList<SoftBody3DSW> *p_soft_body); + BroadPhase3DSW *get_broadphase(); void add_object(CollisionObject3DSW *p_object); diff --git a/servers/physics_3d/step_3d_sw.cpp b/servers/physics_3d/step_3d_sw.cpp index d9370de6a3..2133a38670 100644 --- a/servers/physics_3d/step_3d_sw.cpp +++ b/servers/physics_3d/step_3d_sw.cpp @@ -146,6 +146,8 @@ void Step3DSW::step(Space3DSW *p_space, real_t p_delta, int p_iterations) { const SelfList<Body3DSW>::List *body_list = &p_space->get_active_body_list(); + const SelfList<SoftBody3DSW>::List *soft_body_list = &p_space->get_active_soft_body_list(); + /* INTEGRATE FORCES */ uint64_t profile_begtime = OS::get_singleton()->get_ticks_usec(); @@ -160,6 +162,15 @@ void Step3DSW::step(Space3DSW *p_space, real_t p_delta, int p_iterations) { active_count++; } + /* UPDATE SOFT BODY MOTION */ + + const SelfList<SoftBody3DSW> *sb = soft_body_list->first(); + while (sb) { + sb->self()->predict_motion(p_delta); + sb = sb->next(); + active_count++; + } + p_space->set_active_objects(active_count); { //profile @@ -214,6 +225,21 @@ void Step3DSW::step(Space3DSW *p_space, real_t p_delta, int p_iterations) { p_space->area_remove_from_moved_list((SelfList<Area3DSW> *)aml.first()); //faster to remove here } + sb = soft_body_list->first(); + while (sb) { + for (const Set<Constraint3DSW *>::Element *E = sb->self()->get_constraints().front(); E; E = E->next()) { + Constraint3DSW *c = E->get(); + if (c->get_island_step() == _step) { + continue; + } + c->set_island_step(_step); + c->set_island_next(nullptr); + c->set_island_list_next(constraint_island_list); + constraint_island_list = c; + } + sb = sb->next(); + } + { //profile profile_endtime = OS::get_singleton()->get_ticks_usec(); p_space->set_elapsed_time(Space3DSW::ELAPSED_TIME_GENERATE_ISLANDS, profile_endtime - profile_begtime); @@ -272,6 +298,14 @@ void Step3DSW::step(Space3DSW *p_space, real_t p_delta, int p_iterations) { } } + /* UPDATE SOFT BODY CONSTRAINTS */ + + sb = soft_body_list->first(); + while (sb) { + sb->self()->solve_constraints(p_delta); + sb = sb->next(); + } + { //profile profile_endtime = OS::get_singleton()->get_ticks_usec(); p_space->set_elapsed_time(Space3DSW::ELAPSED_TIME_INTEGRATE_VELOCITIES, profile_endtime - profile_begtime); diff --git a/servers/physics_server_2d.cpp b/servers/physics_server_2d.cpp index a6f64f5848..83ebc0c55b 100644 --- a/servers/physics_server_2d.cpp +++ b/servers/physics_server_2d.cpp @@ -42,7 +42,7 @@ void PhysicsDirectBodyState2D::integrate_forces() { real_t av = get_angular_velocity(); - float damp = 1.0 - step * get_total_linear_damp(); + real_t damp = 1.0 - step * get_total_linear_damp(); if (damp < 0) { // reached zero in the given time damp = 0; @@ -168,11 +168,11 @@ Vector2 PhysicsShapeQueryParameters2D::get_motion() const { return motion; } -void PhysicsShapeQueryParameters2D::set_margin(float p_margin) { +void PhysicsShapeQueryParameters2D::set_margin(real_t p_margin) { margin = p_margin; } -float PhysicsShapeQueryParameters2D::get_margin() const { +real_t PhysicsShapeQueryParameters2D::get_margin() const { return margin; } @@ -311,7 +311,7 @@ Array PhysicsDirectSpaceState2D::_intersect_shape(const Ref<PhysicsShapeQueryPar Array PhysicsDirectSpaceState2D::_cast_motion(const Ref<PhysicsShapeQueryParameters2D> &p_shape_query) { ERR_FAIL_COND_V(!p_shape_query.is_valid(), Array()); - float closest_safe, closest_unsafe; + real_t closest_safe, closest_unsafe; bool res = cast_motion(p_shape_query->shape, p_shape_query->transform, p_shape_query->motion, p_shape_query->margin, closest_safe, closest_unsafe, p_shape_query->exclude, p_shape_query->collision_mask, p_shape_query->collide_with_bodies, p_shape_query->collide_with_areas); if (!res) { return Array(); @@ -517,7 +517,7 @@ PhysicsTestMotionResult2D::PhysicsTestMotionResult2D() { /////////////////////////////////////// -bool PhysicsServer2D::_body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, float p_margin, const Ref<PhysicsTestMotionResult2D> &p_result) { +bool PhysicsServer2D::_body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin, const Ref<PhysicsTestMotionResult2D> &p_result) { MotionResult *r = nullptr; if (p_result.is_valid()) { r = p_result->get_result_ptr(); @@ -655,12 +655,16 @@ void PhysicsServer2D::_bind_methods() { /* JOINT API */ + ClassDB::bind_method(D_METHOD("joint_create"), &PhysicsServer2D::joint_create); + + ClassDB::bind_method(D_METHOD("joint_clear", "joint"), &PhysicsServer2D::joint_clear); + ClassDB::bind_method(D_METHOD("joint_set_param", "joint", "param", "value"), &PhysicsServer2D::joint_set_param); ClassDB::bind_method(D_METHOD("joint_get_param", "joint", "param"), &PhysicsServer2D::joint_get_param); - ClassDB::bind_method(D_METHOD("pin_joint_create", "anchor", "body_a", "body_b"), &PhysicsServer2D::pin_joint_create, DEFVAL(RID())); - ClassDB::bind_method(D_METHOD("groove_joint_create", "groove1_a", "groove2_a", "anchor_b", "body_a", "body_b"), &PhysicsServer2D::groove_joint_create, DEFVAL(RID()), DEFVAL(RID())); - ClassDB::bind_method(D_METHOD("damped_spring_joint_create", "anchor_a", "anchor_b", "body_a", "body_b"), &PhysicsServer2D::damped_spring_joint_create, DEFVAL(RID())); + ClassDB::bind_method(D_METHOD("joint_make_pin", "joint", "anchor", "body_a", "body_b"), &PhysicsServer2D::joint_make_pin, DEFVAL(RID())); + ClassDB::bind_method(D_METHOD("joint_make_groove", "joint", "groove1_a", "groove2_a", "anchor_b", "body_a", "body_b"), &PhysicsServer2D::joint_make_groove, DEFVAL(RID()), DEFVAL(RID())); + ClassDB::bind_method(D_METHOD("joint_make_damped_spring", "joint", "anchor_a", "anchor_b", "body_a", "body_b"), &PhysicsServer2D::joint_make_damped_spring, DEFVAL(RID())); ClassDB::bind_method(D_METHOD("damped_spring_joint_set_param", "joint", "param", "value"), &PhysicsServer2D::damped_spring_joint_set_param); ClassDB::bind_method(D_METHOD("damped_spring_joint_get_param", "joint", "param"), &PhysicsServer2D::damped_spring_joint_get_param); @@ -727,9 +731,10 @@ void PhysicsServer2D::_bind_methods() { BIND_ENUM_CONSTANT(BODY_STATE_SLEEPING); BIND_ENUM_CONSTANT(BODY_STATE_CAN_SLEEP); - BIND_ENUM_CONSTANT(JOINT_PIN); - BIND_ENUM_CONSTANT(JOINT_GROOVE); - BIND_ENUM_CONSTANT(JOINT_DAMPED_SPRING); + BIND_ENUM_CONSTANT(JOINT_TYPE_PIN); + BIND_ENUM_CONSTANT(JOINT_TYPE_GROOVE); + BIND_ENUM_CONSTANT(JOINT_TYPE_DAMPED_SPRING); + BIND_ENUM_CONSTANT(JOINT_TYPE_MAX); BIND_ENUM_CONSTANT(JOINT_PARAM_BIAS); BIND_ENUM_CONSTANT(JOINT_PARAM_MAX_BIAS); diff --git a/servers/physics_server_2d.h b/servers/physics_server_2d.h index dd38855199..28f22ce06b 100644 --- a/servers/physics_server_2d.h +++ b/servers/physics_server_2d.h @@ -45,10 +45,10 @@ protected: public: virtual Vector2 get_total_gravity() const = 0; // get gravity vector working on this body space/area - virtual float get_total_linear_damp() const = 0; // get density of this body space/area - virtual float get_total_angular_damp() const = 0; // get density of this body space/area + virtual real_t get_total_linear_damp() const = 0; // get density of this body space/area + virtual real_t get_total_angular_damp() const = 0; // get density of this body space/area - virtual float get_inverse_mass() const = 0; // get the mass + virtual real_t get_inverse_mass() const = 0; // get the mass virtual real_t get_inverse_inertia() const = 0; // get density of this body space virtual void set_linear_velocity(const Vector2 &p_velocity) = 0; @@ -103,7 +103,7 @@ class PhysicsShapeQueryParameters2D : public Reference { RID shape; Transform2D transform; Vector2 motion; - float margin; + real_t margin; Set<RID> exclude; uint32_t collision_mask; @@ -125,8 +125,8 @@ public: void set_motion(const Vector2 &p_motion); Vector2 get_motion() const; - void set_margin(float p_margin); - float get_margin() const; + void set_margin(real_t p_margin); + real_t get_margin() const; void set_collision_mask(int p_collision_mask); int get_collision_mask() const; @@ -182,11 +182,11 @@ public: virtual int intersect_point(const Vector2 &p_point, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false, bool p_pick_point = false) = 0; virtual int intersect_point_on_canvas(const Vector2 &p_point, ObjectID p_canvas_instance_id, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false, bool p_pick_point = false) = 0; - virtual int intersect_shape(const RID &p_shape, const Transform2D &p_xform, const Vector2 &p_motion, float p_margin, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual int intersect_shape(const RID &p_shape, const Transform2D &p_xform, const Vector2 &p_motion, real_t p_margin, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; - virtual bool cast_motion(const RID &p_shape, const Transform2D &p_xform, const Vector2 &p_motion, float p_margin, float &p_closest_safe, float &p_closest_unsafe, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual bool cast_motion(const RID &p_shape, const Transform2D &p_xform, const Vector2 &p_motion, real_t p_margin, real_t &p_closest_safe, real_t &p_closest_unsafe, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; - virtual bool collide_shape(RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, float p_margin, Vector2 *r_results, int p_result_max, int &r_result_count, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual bool collide_shape(RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, real_t p_margin, Vector2 *r_results, int p_result_max, int &r_result_count, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; struct ShapeRestInfo { Vector2 point; @@ -198,7 +198,7 @@ public: Variant metadata; }; - virtual bool rest_info(RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, float p_margin, ShapeRestInfo *r_info, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual bool rest_info(RID p_shape, const Transform2D &p_shape_xform, const Vector2 &p_motion, real_t p_margin, ShapeRestInfo *r_info, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_layer = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; PhysicsDirectSpaceState2D(); }; @@ -230,7 +230,7 @@ class PhysicsServer2D : public Object { static PhysicsServer2D *singleton; - virtual bool _body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, float p_margin = 0.08, const Ref<PhysicsTestMotionResult2D> &p_result = Ref<PhysicsTestMotionResult2D>()); + virtual bool _body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin = 0.08, const Ref<PhysicsTestMotionResult2D> &p_result = Ref<PhysicsTestMotionResult2D>()); protected: static void _bind_methods(); @@ -393,7 +393,7 @@ public: virtual Variant body_get_shape_metadata(RID p_body, int p_shape_idx) const = 0; virtual void body_set_shape_disabled(RID p_body, int p_shape, bool p_disabled) = 0; - virtual void body_set_shape_as_one_way_collision(RID p_body, int p_shape, bool p_enabled, float p_margin = 0) = 0; + virtual void body_set_shape_as_one_way_collision(RID p_body, int p_shape, bool p_enabled, real_t p_margin = 0) = 0; virtual void body_remove_shape(RID p_body, int p_shape_idx) = 0; virtual void body_clear_shapes(RID p_body) = 0; @@ -431,8 +431,8 @@ public: BODY_PARAM_MAX, }; - virtual void body_set_param(RID p_body, BodyParameter p_param, float p_value) = 0; - virtual float body_get_param(RID p_body, BodyParameter p_param) const = 0; + virtual void body_set_param(RID p_body, BodyParameter p_param, real_t p_value) = 0; + virtual real_t body_get_param(RID p_body, BodyParameter p_param) const = 0; //state enum BodyState { @@ -450,15 +450,15 @@ public: virtual void body_set_applied_force(RID p_body, const Vector2 &p_force) = 0; virtual Vector2 body_get_applied_force(RID p_body) const = 0; - virtual void body_set_applied_torque(RID p_body, float p_torque) = 0; - virtual float body_get_applied_torque(RID p_body) const = 0; + virtual void body_set_applied_torque(RID p_body, real_t p_torque) = 0; + virtual real_t body_get_applied_torque(RID p_body) const = 0; virtual void body_add_central_force(RID p_body, const Vector2 &p_force) = 0; virtual void body_add_force(RID p_body, const Vector2 &p_force, const Vector2 &p_position = Vector2()) = 0; - virtual void body_add_torque(RID p_body, float p_torque) = 0; + virtual void body_add_torque(RID p_body, real_t p_torque) = 0; virtual void body_apply_central_impulse(RID p_body, const Vector2 &p_impulse) = 0; - virtual void body_apply_torque_impulse(RID p_body, float p_torque) = 0; + virtual void body_apply_torque_impulse(RID p_body, real_t p_torque) = 0; virtual void body_apply_impulse(RID p_body, const Vector2 &p_impulse, const Vector2 &p_position = Vector2()) = 0; virtual void body_set_axis_velocity(RID p_body, const Vector2 &p_axis_velocity) = 0; @@ -471,8 +471,8 @@ public: virtual int body_get_max_contacts_reported(RID p_body) const = 0; //missing remove - virtual void body_set_contacts_reported_depth_threshold(RID p_body, float p_threshold) = 0; - virtual float body_get_contacts_reported_depth_threshold(RID p_body) const = 0; + virtual void body_set_contacts_reported_depth_threshold(RID p_body, real_t p_threshold) = 0; + virtual real_t body_get_contacts_reported_depth_threshold(RID p_body) const = 0; virtual void body_set_omit_force_integration(RID p_body, bool p_omit) = 0; virtual bool body_is_omitting_force_integration(RID p_body) const = 0; @@ -506,10 +506,10 @@ public: } }; - virtual bool body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, float p_margin = 0.001, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) = 0; + virtual bool body_test_motion(RID p_body, const Transform2D &p_from, const Vector2 &p_motion, bool p_infinite_inertia, real_t p_margin = 0.001, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) = 0; struct SeparationResult { - float collision_depth; + real_t collision_depth; Vector2 collision_point; Vector2 collision_normal; Vector2 collider_velocity; @@ -520,14 +520,19 @@ public: Variant collider_metadata; }; - virtual int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin = 0.001) = 0; + virtual int body_test_ray_separation(RID p_body, const Transform2D &p_transform, bool p_infinite_inertia, Vector2 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) = 0; /* JOINT API */ + virtual RID joint_create() = 0; + + virtual void joint_clear(RID p_joint) = 0; + enum JointType { - JOINT_PIN, - JOINT_GROOVE, - JOINT_DAMPED_SPRING + JOINT_TYPE_PIN, + JOINT_TYPE_GROOVE, + JOINT_TYPE_DAMPED_SPRING, + JOINT_TYPE_MAX }; enum JointParam { @@ -542,9 +547,9 @@ public: virtual void joint_disable_collisions_between_bodies(RID p_joint, const bool p_disable) = 0; virtual bool joint_is_disabled_collisions_between_bodies(RID p_joint) const = 0; - virtual RID pin_joint_create(const Vector2 &p_anchor, RID p_body_a, RID p_body_b = RID()) = 0; - virtual RID groove_joint_create(const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) = 0; - virtual RID damped_spring_joint_create(const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b = RID()) = 0; + virtual void joint_make_pin(RID p_joint, const Vector2 &p_anchor, RID p_body_a, RID p_body_b = RID()) = 0; + virtual void joint_make_groove(RID p_joint, const Vector2 &p_a_groove1, const Vector2 &p_a_groove2, const Vector2 &p_b_anchor, RID p_body_a, RID p_body_b) = 0; + virtual void joint_make_damped_spring(RID p_joint, const Vector2 &p_anchor_a, const Vector2 &p_anchor_b, RID p_body_a, RID p_body_b = RID()) = 0; enum PinJointParam { PIN_JOINT_SOFTNESS @@ -576,7 +581,7 @@ public: virtual void set_active(bool p_active) = 0; virtual void init() = 0; - virtual void step(float p_step) = 0; + virtual void step(real_t p_step) = 0; virtual void sync() = 0; virtual void flush_queries() = 0; virtual void end_sync() = 0; diff --git a/servers/physics_server_3d.cpp b/servers/physics_server_3d.cpp index 27ebe32e50..586845de99 100644 --- a/servers/physics_server_3d.cpp +++ b/servers/physics_server_3d.cpp @@ -42,13 +42,13 @@ void PhysicsDirectBodyState3D::integrate_forces() { Vector3 av = get_angular_velocity(); - float linear_damp = 1.0 - step * get_total_linear_damp(); + real_t linear_damp = 1.0 - step * get_total_linear_damp(); if (linear_damp < 0) { // reached zero in the given time linear_damp = 0; } - float angular_damp = 1.0 - step * get_total_angular_damp(); + real_t angular_damp = 1.0 - step * get_total_angular_damp(); if (angular_damp < 0) { // reached zero in the given time angular_damp = 0; @@ -128,7 +128,7 @@ void PhysicsDirectBodyState3D::_bind_methods() { ADD_PROPERTY(PropertyInfo(Variant::VECTOR3, "angular_velocity"), "set_angular_velocity", "get_angular_velocity"); ADD_PROPERTY(PropertyInfo(Variant::VECTOR3, "linear_velocity"), "set_linear_velocity", "get_linear_velocity"); ADD_PROPERTY(PropertyInfo(Variant::BOOL, "sleeping"), "set_sleep_state", "is_sleeping"); - ADD_PROPERTY(PropertyInfo(Variant::TRANSFORM2D, "transform"), "set_transform", "get_transform"); + ADD_PROPERTY(PropertyInfo(Variant::TRANSFORM, "transform"), "set_transform", "get_transform"); } PhysicsDirectBodyState3D::PhysicsDirectBodyState3D() {} @@ -164,11 +164,11 @@ Transform PhysicsShapeQueryParameters3D::get_transform() const { return transform; } -void PhysicsShapeQueryParameters3D::set_margin(float p_margin) { +void PhysicsShapeQueryParameters3D::set_margin(real_t p_margin) { margin = p_margin; } -float PhysicsShapeQueryParameters3D::get_margin() const { +real_t PhysicsShapeQueryParameters3D::get_margin() const { return margin; } @@ -303,7 +303,7 @@ Array PhysicsDirectSpaceState3D::_intersect_shape(const Ref<PhysicsShapeQueryPar Array PhysicsDirectSpaceState3D::_cast_motion(const Ref<PhysicsShapeQueryParameters3D> &p_shape_query, const Vector3 &p_motion) { ERR_FAIL_COND_V(!p_shape_query.is_valid(), Array()); - float closest_safe = 1.0f, closest_unsafe = 1.0f; + real_t closest_safe = 1.0f, closest_unsafe = 1.0f; bool res = cast_motion(p_shape_query->shape, p_shape_query->transform, p_motion, p_shape_query->margin, closest_safe, closest_unsafe, p_shape_query->exclude, p_shape_query->collision_mask, p_shape_query->collide_with_bodies, p_shape_query->collide_with_areas); if (!res) { return Array(); @@ -398,10 +398,47 @@ void PhysicsShapeQueryResult3D::_bind_methods() { /////////////////////////////////////// +RID PhysicsServer3D::shape_create(ShapeType p_shape) { + switch (p_shape) { + case SHAPE_PLANE: + return plane_shape_create(); + case SHAPE_RAY: + return ray_shape_create(); + case SHAPE_SPHERE: + return sphere_shape_create(); + case SHAPE_BOX: + return box_shape_create(); + case SHAPE_CAPSULE: + return capsule_shape_create(); + case SHAPE_CYLINDER: + return cylinder_shape_create(); + case SHAPE_CONVEX_POLYGON: + return convex_polygon_shape_create(); + case SHAPE_CONCAVE_POLYGON: + return concave_polygon_shape_create(); + case SHAPE_HEIGHTMAP: + return heightmap_shape_create(); + case SHAPE_CUSTOM: + return custom_shape_create(); + default: + return RID(); + } +} + void PhysicsServer3D::_bind_methods() { #ifndef _3D_DISABLED - ClassDB::bind_method(D_METHOD("shape_create", "type"), &PhysicsServer3D::shape_create); + ClassDB::bind_method(D_METHOD("plane_shape_create"), &PhysicsServer3D::plane_shape_create); + ClassDB::bind_method(D_METHOD("ray_shape_create"), &PhysicsServer3D::ray_shape_create); + ClassDB::bind_method(D_METHOD("sphere_shape_create"), &PhysicsServer3D::sphere_shape_create); + ClassDB::bind_method(D_METHOD("box_shape_create"), &PhysicsServer3D::box_shape_create); + ClassDB::bind_method(D_METHOD("capsule_shape_create"), &PhysicsServer3D::capsule_shape_create); + ClassDB::bind_method(D_METHOD("cylinder_shape_create"), &PhysicsServer3D::cylinder_shape_create); + ClassDB::bind_method(D_METHOD("convex_polygon_shape_create"), &PhysicsServer3D::convex_polygon_shape_create); + ClassDB::bind_method(D_METHOD("concave_polygon_shape_create"), &PhysicsServer3D::concave_polygon_shape_create); + ClassDB::bind_method(D_METHOD("heightmap_shape_create"), &PhysicsServer3D::heightmap_shape_create); + ClassDB::bind_method(D_METHOD("custom_shape_create"), &PhysicsServer3D::custom_shape_create); + ClassDB::bind_method(D_METHOD("shape_set_data", "shape", "data"), &PhysicsServer3D::shape_set_data); ClassDB::bind_method(D_METHOD("shape_get_type", "shape"), &PhysicsServer3D::shape_get_type); @@ -450,9 +487,8 @@ void PhysicsServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("area_set_monitorable", "area", "monitorable"), &PhysicsServer3D::area_set_monitorable); ClassDB::bind_method(D_METHOD("area_set_ray_pickable", "area", "enable"), &PhysicsServer3D::area_set_ray_pickable); - ClassDB::bind_method(D_METHOD("area_is_ray_pickable", "area"), &PhysicsServer3D::area_is_ray_pickable); - ClassDB::bind_method(D_METHOD("body_create", "mode", "init_sleeping"), &PhysicsServer3D::body_create, DEFVAL(BODY_MODE_RIGID), DEFVAL(false)); + ClassDB::bind_method(D_METHOD("body_create"), &PhysicsServer3D::body_create); ClassDB::bind_method(D_METHOD("body_set_space", "body", "space"), &PhysicsServer3D::body_set_space); ClassDB::bind_method(D_METHOD("body_get_space", "body"), &PhysicsServer3D::body_get_space); @@ -517,19 +553,26 @@ void PhysicsServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("body_set_force_integration_callback", "body", "receiver", "method", "userdata"), &PhysicsServer3D::body_set_force_integration_callback, DEFVAL(Variant())); ClassDB::bind_method(D_METHOD("body_set_ray_pickable", "body", "enable"), &PhysicsServer3D::body_set_ray_pickable); - ClassDB::bind_method(D_METHOD("body_is_ray_pickable", "body"), &PhysicsServer3D::body_is_ray_pickable); ClassDB::bind_method(D_METHOD("body_get_direct_state", "body"), &PhysicsServer3D::body_get_direct_state); + /* SOFT BODY API */ + + ClassDB::bind_method(D_METHOD("soft_body_get_bounds", "body"), &PhysicsServer3D::soft_body_get_bounds); + /* JOINT API */ - BIND_ENUM_CONSTANT(JOINT_PIN); - BIND_ENUM_CONSTANT(JOINT_HINGE); - BIND_ENUM_CONSTANT(JOINT_SLIDER); - BIND_ENUM_CONSTANT(JOINT_CONE_TWIST); - BIND_ENUM_CONSTANT(JOINT_6DOF); + ClassDB::bind_method(D_METHOD("joint_create"), &PhysicsServer3D::joint_create); + ClassDB::bind_method(D_METHOD("joint_clear", "joint"), &PhysicsServer3D::joint_clear); + + BIND_ENUM_CONSTANT(JOINT_TYPE_PIN); + BIND_ENUM_CONSTANT(JOINT_TYPE_HINGE); + BIND_ENUM_CONSTANT(JOINT_TYPE_SLIDER); + BIND_ENUM_CONSTANT(JOINT_TYPE_CONE_TWIST); + BIND_ENUM_CONSTANT(JOINT_TYPE_6DOF); + BIND_ENUM_CONSTANT(JOINT_TYPE_MAX); - ClassDB::bind_method(D_METHOD("joint_create_pin", "body_A", "local_A", "body_B", "local_B"), &PhysicsServer3D::joint_create_pin); + ClassDB::bind_method(D_METHOD("joint_make_pin", "joint", "body_A", "local_A", "body_B", "local_B"), &PhysicsServer3D::joint_make_pin); ClassDB::bind_method(D_METHOD("pin_joint_set_param", "joint", "param", "value"), &PhysicsServer3D::pin_joint_set_param); ClassDB::bind_method(D_METHOD("pin_joint_get_param", "joint", "param"), &PhysicsServer3D::pin_joint_get_param); @@ -555,7 +598,7 @@ void PhysicsServer3D::_bind_methods() { BIND_ENUM_CONSTANT(HINGE_JOINT_FLAG_USE_LIMIT); BIND_ENUM_CONSTANT(HINGE_JOINT_FLAG_ENABLE_MOTOR); - ClassDB::bind_method(D_METHOD("joint_create_hinge", "body_A", "hinge_A", "body_B", "hinge_B"), &PhysicsServer3D::joint_create_hinge); + ClassDB::bind_method(D_METHOD("joint_make_hinge", "joint", "body_A", "hinge_A", "body_B", "hinge_B"), &PhysicsServer3D::joint_make_hinge); ClassDB::bind_method(D_METHOD("hinge_joint_set_param", "joint", "param", "value"), &PhysicsServer3D::hinge_joint_set_param); ClassDB::bind_method(D_METHOD("hinge_joint_get_param", "joint", "param"), &PhysicsServer3D::hinge_joint_get_param); @@ -563,7 +606,7 @@ void PhysicsServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("hinge_joint_set_flag", "joint", "flag", "enabled"), &PhysicsServer3D::hinge_joint_set_flag); ClassDB::bind_method(D_METHOD("hinge_joint_get_flag", "joint", "flag"), &PhysicsServer3D::hinge_joint_get_flag); - ClassDB::bind_method(D_METHOD("joint_create_slider", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_create_slider); + ClassDB::bind_method(D_METHOD("joint_make_slider", "joint", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_make_slider); ClassDB::bind_method(D_METHOD("slider_joint_set_param", "joint", "param", "value"), &PhysicsServer3D::slider_joint_set_param); ClassDB::bind_method(D_METHOD("slider_joint_get_param", "joint", "param"), &PhysicsServer3D::slider_joint_get_param); @@ -593,7 +636,7 @@ void PhysicsServer3D::_bind_methods() { BIND_ENUM_CONSTANT(SLIDER_JOINT_ANGULAR_ORTHOGONAL_DAMPING); BIND_ENUM_CONSTANT(SLIDER_JOINT_MAX); - ClassDB::bind_method(D_METHOD("joint_create_cone_twist", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_create_cone_twist); + ClassDB::bind_method(D_METHOD("joint_make_cone_twist", "joint", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_make_cone_twist); ClassDB::bind_method(D_METHOD("cone_twist_joint_set_param", "joint", "param", "value"), &PhysicsServer3D::cone_twist_joint_set_param); ClassDB::bind_method(D_METHOD("cone_twist_joint_get_param", "joint", "param"), &PhysicsServer3D::cone_twist_joint_get_param); @@ -631,7 +674,7 @@ void PhysicsServer3D::_bind_methods() { ClassDB::bind_method(D_METHOD("joint_set_solver_priority", "joint", "priority"), &PhysicsServer3D::joint_set_solver_priority); ClassDB::bind_method(D_METHOD("joint_get_solver_priority", "joint"), &PhysicsServer3D::joint_get_solver_priority); - ClassDB::bind_method(D_METHOD("joint_create_generic_6dof", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_create_generic_6dof); + ClassDB::bind_method(D_METHOD("joint_make_generic_6dof", "joint", "body_A", "local_ref_A", "body_B", "local_ref_B"), &PhysicsServer3D::joint_make_generic_6dof); ClassDB::bind_method(D_METHOD("generic_6dof_joint_set_param", "joint", "axis", "param", "value"), &PhysicsServer3D::generic_6dof_joint_set_param); ClassDB::bind_method(D_METHOD("generic_6dof_joint_get_param", "joint", "axis", "param"), &PhysicsServer3D::generic_6dof_joint_get_param); @@ -654,6 +697,7 @@ void PhysicsServer3D::_bind_methods() { BIND_ENUM_CONSTANT(SHAPE_CONVEX_POLYGON); BIND_ENUM_CONSTANT(SHAPE_CONCAVE_POLYGON); BIND_ENUM_CONSTANT(SHAPE_HEIGHTMAP); + BIND_ENUM_CONSTANT(SHAPE_SOFT_BODY); BIND_ENUM_CONSTANT(SHAPE_CUSTOM); BIND_ENUM_CONSTANT(AREA_PARAM_GRAVITY); @@ -718,7 +762,6 @@ void PhysicsServer3D::_bind_methods() { } PhysicsServer3D::PhysicsServer3D() { - ERR_FAIL_COND(singleton != nullptr); singleton = this; } diff --git a/servers/physics_server_3d.h b/servers/physics_server_3d.h index 303825f37c..69f5c1c0ad 100644 --- a/servers/physics_server_3d.h +++ b/servers/physics_server_3d.h @@ -44,12 +44,12 @@ protected: public: virtual Vector3 get_total_gravity() const = 0; - virtual float get_total_angular_damp() const = 0; - virtual float get_total_linear_damp() const = 0; + virtual real_t get_total_angular_damp() const = 0; + virtual real_t get_total_linear_damp() const = 0; virtual Vector3 get_center_of_mass() const = 0; virtual Basis get_principal_inertia_axes() const = 0; - virtual float get_inverse_mass() const = 0; // get the mass + virtual real_t get_inverse_mass() const = 0; // get the mass virtual Vector3 get_inverse_inertia() const = 0; // get density of this body space virtual Basis get_inverse_inertia_tensor() const = 0; // get density of this body space @@ -76,7 +76,7 @@ public: virtual Vector3 get_contact_local_position(int p_contact_idx) const = 0; virtual Vector3 get_contact_local_normal(int p_contact_idx) const = 0; - virtual float get_contact_impulse(int p_contact_idx) const = 0; + virtual real_t get_contact_impulse(int p_contact_idx) const = 0; virtual int get_contact_local_shape(int p_contact_idx) const = 0; virtual RID get_contact_collider(int p_contact_idx) const = 0; @@ -103,7 +103,7 @@ class PhysicsShapeQueryParameters3D : public Reference { RES shape_ref; RID shape; Transform transform; - float margin; + real_t margin; Set<RID> exclude; uint32_t collision_mask; @@ -122,8 +122,8 @@ public: void set_transform(const Transform &p_transform); Transform get_transform() const; - void set_margin(float p_margin); - float get_margin() const; + void set_margin(real_t p_margin); + real_t get_margin() const; void set_collision_mask(int p_collision_mask); int get_collision_mask() const; @@ -174,7 +174,7 @@ public: virtual bool intersect_ray(const Vector3 &p_from, const Vector3 &p_to, RayResult &r_result, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false, bool p_pick_ray = false) = 0; - virtual int intersect_shape(const RID &p_shape, const Transform &p_xform, float p_margin, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual int intersect_shape(const RID &p_shape, const Transform &p_xform, real_t p_margin, ShapeResult *r_results, int p_result_max, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; struct ShapeRestInfo { Vector3 point; @@ -185,11 +185,11 @@ public: Vector3 linear_velocity; //velocity at contact point }; - virtual bool cast_motion(const RID &p_shape, const Transform &p_xform, const Vector3 &p_motion, float p_margin, float &p_closest_safe, float &p_closest_unsafe, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false, ShapeRestInfo *r_info = nullptr) = 0; + virtual bool cast_motion(const RID &p_shape, const Transform &p_xform, const Vector3 &p_motion, real_t p_margin, real_t &p_closest_safe, real_t &p_closest_unsafe, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false, ShapeRestInfo *r_info = nullptr) = 0; - virtual bool collide_shape(RID p_shape, const Transform &p_shape_xform, float p_margin, Vector3 *r_results, int p_result_max, int &r_result_count, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual bool collide_shape(RID p_shape, const Transform &p_shape_xform, real_t p_margin, Vector3 *r_results, int p_result_max, int &r_result_count, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; - virtual bool rest_info(RID p_shape, const Transform &p_shape_xform, float p_margin, ShapeRestInfo *r_info, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; + virtual bool rest_info(RID p_shape, const Transform &p_shape_xform, real_t p_margin, ShapeRestInfo *r_info, const Set<RID> &p_exclude = Set<RID>(), uint32_t p_collision_mask = 0xFFFFFFFF, bool p_collide_with_bodies = true, bool p_collide_with_areas = false) = 0; virtual Vector3 get_closest_point_to_object_volume(RID p_object, const Vector3 p_point) const = 0; @@ -216,6 +216,15 @@ public: PhysicsShapeQueryResult3D(); }; +class RenderingServerHandler { +public: + virtual void set_vertex(int p_vertex_id, const void *p_vector3) = 0; + virtual void set_normal(int p_vertex_id, const void *p_vector3) = 0; + virtual void set_aabb(const AABB &p_aabb) = 0; + + virtual ~RenderingServerHandler() {} +}; + class PhysicsServer3D : public Object { GDCLASS(PhysicsServer3D, Object); @@ -237,10 +246,23 @@ public: SHAPE_CONVEX_POLYGON, ///< array of planes:"planes" SHAPE_CONCAVE_POLYGON, ///< vector3 array:"triangles" , or Dictionary with "indices" (int array) and "triangles" (Vector3 array) SHAPE_HEIGHTMAP, ///< dict( int:"width", int:"depth",float:"cell_size", float_array:"heights" + SHAPE_SOFT_BODY, ///< Used internally, can't be created from the physics server. SHAPE_CUSTOM, ///< Server-Implementation based custom shape, calling shape_create() with this value will result in an error }; - virtual RID shape_create(ShapeType p_shape) = 0; + RID shape_create(ShapeType p_shape); + + virtual RID plane_shape_create() = 0; + virtual RID ray_shape_create() = 0; + virtual RID sphere_shape_create() = 0; + virtual RID box_shape_create() = 0; + virtual RID capsule_shape_create() = 0; + virtual RID cylinder_shape_create() = 0; + virtual RID convex_polygon_shape_create() = 0; + virtual RID concave_polygon_shape_create() = 0; + virtual RID heightmap_shape_create() = 0; + virtual RID custom_shape_create() = 0; + virtual void shape_set_data(RID p_shape, const Variant &p_data) = 0; virtual void shape_set_custom_solver_bias(RID p_shape, real_t p_bias) = 0; @@ -344,7 +366,6 @@ public: virtual void area_set_area_monitor_callback(RID p_area, Object *p_receiver, const StringName &p_method) = 0; virtual void area_set_ray_pickable(RID p_area, bool p_enable) = 0; - virtual bool area_is_ray_pickable(RID p_area) const = 0; /* BODY API */ @@ -357,7 +378,7 @@ public: BODY_MODE_CHARACTER }; - virtual RID body_create(BodyMode p_mode = BODY_MODE_RIGID, bool p_init_sleeping = false) = 0; + virtual RID body_create() = 0; virtual void body_set_space(RID p_body, RID p_space) = 0; virtual RID body_get_space(RID p_body) const = 0; @@ -404,8 +425,8 @@ public: BODY_PARAM_MAX, }; - virtual void body_set_param(RID p_body, BodyParameter p_param, float p_value) = 0; - virtual float body_get_param(RID p_body, BodyParameter p_param) const = 0; + virtual void body_set_param(RID p_body, BodyParameter p_param, real_t p_value) = 0; + virtual real_t body_get_param(RID p_body, BodyParameter p_param) const = 0; virtual void body_set_kinematic_safe_margin(RID p_body, real_t p_margin) = 0; virtual real_t body_get_kinematic_safe_margin(RID p_body) const = 0; @@ -459,8 +480,8 @@ public: virtual int body_get_max_contacts_reported(RID p_body) const = 0; //missing remove - virtual void body_set_contacts_reported_depth_threshold(RID p_body, float p_threshold) = 0; - virtual float body_get_contacts_reported_depth_threshold(RID p_body) const = 0; + virtual void body_set_contacts_reported_depth_threshold(RID p_body, real_t p_threshold) = 0; + virtual real_t body_get_contacts_reported_depth_threshold(RID p_body) const = 0; virtual void body_set_omit_force_integration(RID p_body, bool p_omit) = 0; virtual bool body_is_omitting_force_integration(RID p_body) const = 0; @@ -468,7 +489,6 @@ public: virtual void body_set_force_integration_callback(RID p_body, Object *p_receiver, const StringName &p_method, const Variant &p_udata = Variant()) = 0; virtual void body_set_ray_pickable(RID p_body, bool p_enable) = 0; - virtual bool body_is_ray_pickable(RID p_body) const = 0; // this function only works on physics process, errors and returns null otherwise virtual PhysicsDirectBodyState3D *body_get_direct_state(RID p_body) = 0; @@ -495,7 +515,7 @@ public: virtual bool body_test_motion(RID p_body, const Transform &p_from, const Vector3 &p_motion, bool p_infinite_inertia, MotionResult *r_result = nullptr, bool p_exclude_raycast_shapes = true) = 0; struct SeparationResult { - float collision_depth; + real_t collision_depth; Vector3 collision_point; Vector3 collision_normal; Vector3 collider_velocity; @@ -506,19 +526,21 @@ public: Variant collider_metadata; }; - virtual int body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, float p_margin = 0.001) = 0; + virtual int body_test_ray_separation(RID p_body, const Transform &p_transform, bool p_infinite_inertia, Vector3 &r_recover_motion, SeparationResult *r_results, int p_result_max, real_t p_margin = 0.001) = 0; /* SOFT BODY */ - virtual RID soft_body_create(bool p_init_sleeping = false) = 0; + virtual RID soft_body_create() = 0; - virtual void soft_body_update_rendering_server(RID p_body, class SoftBodyRenderingServerHandler *p_rendering_server_handler) = 0; + virtual void soft_body_update_rendering_server(RID p_body, RenderingServerHandler *p_rendering_server_handler) = 0; virtual void soft_body_set_space(RID p_body, RID p_space) = 0; virtual RID soft_body_get_space(RID p_body) const = 0; virtual void soft_body_set_mesh(RID p_body, const REF &p_mesh) = 0; + virtual AABB soft_body_get_bounds(RID p_body) const = 0; + virtual void soft_body_set_collision_layer(RID p_body, uint32_t p_layer) = 0; virtual uint32_t soft_body_get_collision_layer(RID p_body) const = 0; @@ -533,58 +555,50 @@ public: virtual Variant soft_body_get_state(RID p_body, BodyState p_state) const = 0; virtual void soft_body_set_transform(RID p_body, const Transform &p_transform) = 0; - virtual Vector3 soft_body_get_vertex_position(RID p_body, int vertex_index) const = 0; virtual void soft_body_set_ray_pickable(RID p_body, bool p_enable) = 0; - virtual bool soft_body_is_ray_pickable(RID p_body) const = 0; virtual void soft_body_set_simulation_precision(RID p_body, int p_simulation_precision) = 0; - virtual int soft_body_get_simulation_precision(RID p_body) = 0; + virtual int soft_body_get_simulation_precision(RID p_body) const = 0; virtual void soft_body_set_total_mass(RID p_body, real_t p_total_mass) = 0; - virtual real_t soft_body_get_total_mass(RID p_body) = 0; + virtual real_t soft_body_get_total_mass(RID p_body) const = 0; virtual void soft_body_set_linear_stiffness(RID p_body, real_t p_stiffness) = 0; - virtual real_t soft_body_get_linear_stiffness(RID p_body) = 0; - - virtual void soft_body_set_areaAngular_stiffness(RID p_body, real_t p_stiffness) = 0; - virtual real_t soft_body_get_areaAngular_stiffness(RID p_body) = 0; - - virtual void soft_body_set_volume_stiffness(RID p_body, real_t p_stiffness) = 0; - virtual real_t soft_body_get_volume_stiffness(RID p_body) = 0; + virtual real_t soft_body_get_linear_stiffness(RID p_body) const = 0; virtual void soft_body_set_pressure_coefficient(RID p_body, real_t p_pressure_coefficient) = 0; - virtual real_t soft_body_get_pressure_coefficient(RID p_body) = 0; - - virtual void soft_body_set_pose_matching_coefficient(RID p_body, real_t p_pose_matching_coefficient) = 0; - virtual real_t soft_body_get_pose_matching_coefficient(RID p_body) = 0; + virtual real_t soft_body_get_pressure_coefficient(RID p_body) const = 0; virtual void soft_body_set_damping_coefficient(RID p_body, real_t p_damping_coefficient) = 0; - virtual real_t soft_body_get_damping_coefficient(RID p_body) = 0; + virtual real_t soft_body_get_damping_coefficient(RID p_body) const = 0; virtual void soft_body_set_drag_coefficient(RID p_body, real_t p_drag_coefficient) = 0; - virtual real_t soft_body_get_drag_coefficient(RID p_body) = 0; + virtual real_t soft_body_get_drag_coefficient(RID p_body) const = 0; virtual void soft_body_move_point(RID p_body, int p_point_index, const Vector3 &p_global_position) = 0; - virtual Vector3 soft_body_get_point_global_position(RID p_body, int p_point_index) = 0; - - virtual Vector3 soft_body_get_point_offset(RID p_body, int p_point_index) const = 0; + virtual Vector3 soft_body_get_point_global_position(RID p_body, int p_point_index) const = 0; virtual void soft_body_remove_all_pinned_points(RID p_body) = 0; virtual void soft_body_pin_point(RID p_body, int p_point_index, bool p_pin) = 0; - virtual bool soft_body_is_point_pinned(RID p_body, int p_point_index) = 0; + virtual bool soft_body_is_point_pinned(RID p_body, int p_point_index) const = 0; /* JOINT API */ enum JointType { - JOINT_PIN, - JOINT_HINGE, - JOINT_SLIDER, - JOINT_CONE_TWIST, - JOINT_6DOF + JOINT_TYPE_PIN, + JOINT_TYPE_HINGE, + JOINT_TYPE_SLIDER, + JOINT_TYPE_CONE_TWIST, + JOINT_TYPE_6DOF, + JOINT_TYPE_MAX, }; + virtual RID joint_create() = 0; + + virtual void joint_clear(RID p_joint) = 0; + virtual JointType joint_get_type(RID p_joint) const = 0; virtual void joint_set_solver_priority(RID p_joint, int p_priority) = 0; @@ -593,7 +607,7 @@ public: virtual void joint_disable_collisions_between_bodies(RID p_joint, const bool p_disable) = 0; virtual bool joint_is_disabled_collisions_between_bodies(RID p_joint) const = 0; - virtual RID joint_create_pin(RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) = 0; + virtual void joint_make_pin(RID p_joint, RID p_body_A, const Vector3 &p_local_A, RID p_body_B, const Vector3 &p_local_B) = 0; enum PinJointParam { PIN_JOINT_BIAS, @@ -601,8 +615,8 @@ public: PIN_JOINT_IMPULSE_CLAMP }; - virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, float p_value) = 0; - virtual float pin_joint_get_param(RID p_joint, PinJointParam p_param) const = 0; + virtual void pin_joint_set_param(RID p_joint, PinJointParam p_param, real_t p_value) = 0; + virtual real_t pin_joint_get_param(RID p_joint, PinJointParam p_param) const = 0; virtual void pin_joint_set_local_a(RID p_joint, const Vector3 &p_A) = 0; virtual Vector3 pin_joint_get_local_a(RID p_joint) const = 0; @@ -628,11 +642,11 @@ public: HINGE_JOINT_FLAG_MAX }; - virtual RID joint_create_hinge(RID p_body_A, const Transform &p_hinge_A, RID p_body_B, const Transform &p_hinge_B) = 0; - virtual RID joint_create_hinge_simple(RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) = 0; + virtual void joint_make_hinge(RID p_joint, RID p_body_A, const Transform &p_hinge_A, RID p_body_B, const Transform &p_hinge_B) = 0; + virtual void joint_make_hinge_simple(RID p_joint, RID p_body_A, const Vector3 &p_pivot_A, const Vector3 &p_axis_A, RID p_body_B, const Vector3 &p_pivot_B, const Vector3 &p_axis_B) = 0; - virtual void hinge_joint_set_param(RID p_joint, HingeJointParam p_param, float p_value) = 0; - virtual float hinge_joint_get_param(RID p_joint, HingeJointParam p_param) const = 0; + virtual void hinge_joint_set_param(RID p_joint, HingeJointParam p_param, real_t p_value) = 0; + virtual real_t hinge_joint_get_param(RID p_joint, HingeJointParam p_param) const = 0; virtual void hinge_joint_set_flag(RID p_joint, HingeJointFlag p_flag, bool p_value) = 0; virtual bool hinge_joint_get_flag(RID p_joint, HingeJointFlag p_flag) const = 0; @@ -665,10 +679,10 @@ public: }; - virtual RID joint_create_slider(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A + virtual void joint_make_slider(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A - virtual void slider_joint_set_param(RID p_joint, SliderJointParam p_param, float p_value) = 0; - virtual float slider_joint_get_param(RID p_joint, SliderJointParam p_param) const = 0; + virtual void slider_joint_set_param(RID p_joint, SliderJointParam p_param, real_t p_value) = 0; + virtual real_t slider_joint_get_param(RID p_joint, SliderJointParam p_param) const = 0; enum ConeTwistJointParam { CONE_TWIST_JOINT_SWING_SPAN, @@ -679,10 +693,10 @@ public: CONE_TWIST_MAX }; - virtual RID joint_create_cone_twist(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A + virtual void joint_make_cone_twist(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A - virtual void cone_twist_joint_set_param(RID p_joint, ConeTwistJointParam p_param, float p_value) = 0; - virtual float cone_twist_joint_get_param(RID p_joint, ConeTwistJointParam p_param) const = 0; + virtual void cone_twist_joint_set_param(RID p_joint, ConeTwistJointParam p_param, real_t p_value) = 0; + virtual real_t cone_twist_joint_get_param(RID p_joint, ConeTwistJointParam p_param) const = 0; enum G6DOFJointAxisParam { G6DOF_JOINT_LINEAR_LOWER_LIMIT, @@ -720,13 +734,13 @@ public: G6DOF_JOINT_FLAG_MAX }; - virtual RID joint_create_generic_6dof(RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A + virtual void joint_make_generic_6dof(RID p_joint, RID p_body_A, const Transform &p_local_frame_A, RID p_body_B, const Transform &p_local_frame_B) = 0; //reference frame is A - virtual void generic_6dof_joint_set_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param, float p_value) = 0; - virtual float generic_6dof_joint_get_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param) = 0; + virtual void generic_6dof_joint_set_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param, real_t p_value) = 0; + virtual real_t generic_6dof_joint_get_param(RID p_joint, Vector3::Axis, G6DOFJointAxisParam p_param) const = 0; virtual void generic_6dof_joint_set_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag, bool p_enable) = 0; - virtual bool generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag) = 0; + virtual bool generic_6dof_joint_get_flag(RID p_joint, Vector3::Axis, G6DOFJointAxisFlag p_flag) const = 0; /* QUERY API */ @@ -741,8 +755,10 @@ public: virtual void set_active(bool p_active) = 0; virtual void init() = 0; - virtual void step(float p_step) = 0; + virtual void step(real_t p_step) = 0; + virtual void sync() = 0; virtual void flush_queries() = 0; + virtual void end_sync() = 0; virtual void finish() = 0; virtual bool is_flushing_queries() const = 0; diff --git a/servers/register_server_types.cpp b/servers/register_server_types.cpp index 58bcdf5802..deb230c4fb 100644 --- a/servers/register_server_types.cpp +++ b/servers/register_server_types.cpp @@ -36,6 +36,7 @@ #include "audio/audio_effect.h" #include "audio/audio_stream.h" #include "audio/effects/audio_effect_amplify.h" +#include "audio/effects/audio_effect_capture.h" #include "audio/effects/audio_effect_chorus.h" #include "audio/effects/audio_effect_compressor.h" #include "audio/effects/audio_effect_delay.h" @@ -60,6 +61,7 @@ #include "physics_2d/physics_server_2d_sw.h" #include "physics_2d/physics_server_2d_wrap_mt.h" #include "physics_3d/physics_server_3d_sw.h" +#include "physics_3d/physics_server_3d_wrap_mt.h" #include "physics_server_2d.h" #include "physics_server_3d.h" #include "rendering/renderer_compositor.h" @@ -75,11 +77,19 @@ ShaderTypes *shader_types = nullptr; PhysicsServer3D *_createGodotPhysics3DCallback() { - return memnew(PhysicsServer3DSW); + bool using_threads = GLOBAL_GET("physics/3d/run_on_thread"); + + PhysicsServer3D *physics_server = memnew(PhysicsServer3DSW(using_threads)); + + return memnew(PhysicsServer3DWrapMT(physics_server, using_threads)); } PhysicsServer2D *_createGodotPhysics2DCallback() { - return PhysicsServer2DWrapMT::init_server<PhysicsServer2DSW>(); + bool using_threads = GLOBAL_GET("physics/2d/run_on_thread"); + + PhysicsServer2D *physics_server = memnew(PhysicsServer2DSW(using_threads)); + + return memnew(PhysicsServer2DWrapMT(physics_server, using_threads)); } static bool has_server_feature_callback(const String &p_feature) { @@ -94,6 +104,16 @@ static bool has_server_feature_callback(const String &p_feature) { void preregister_server_types() { shader_types = memnew(ShaderTypes); + + GLOBAL_DEF("internationalization/rendering/text_driver", ""); + String text_driver_options; + for (int i = 0; i < TextServerManager::get_interface_count(); i++) { + if (i > 0) { + text_driver_options += ","; + } + text_driver_options += TextServerManager::get_interface_name(i); + } + ProjectSettings::get_singleton()->set_custom_property_info("internationalization/rendering/text_driver", PropertyInfo(Variant::STRING, "internationalization/rendering/text_driver", PROPERTY_HINT_ENUM, text_driver_options)); } void register_server_types() { @@ -166,6 +186,8 @@ void register_server_types() { ClassDB::register_class<AudioEffectRecord>(); ClassDB::register_class<AudioEffectSpectrumAnalyzer>(); ClassDB::register_virtual_class<AudioEffectSpectrumAnalyzerInstance>(); + + ClassDB::register_class<AudioEffectCapture>(); } ClassDB::register_virtual_class<RenderingDevice>(); diff --git a/servers/rendering/renderer_canvas_cull.cpp b/servers/rendering/renderer_canvas_cull.cpp index 2d2847e6ca..7f3fc2f8f4 100644 --- a/servers/rendering/renderer_canvas_cull.cpp +++ b/servers/rendering/renderer_canvas_cull.cpp @@ -97,7 +97,7 @@ void _collect_ysort_children(RendererCanvasCull::Item *p_canvas_item, Transform2 } } -void _mark_ysort_dirty(RendererCanvasCull::Item *ysort_owner, RID_PtrOwner<RendererCanvasCull::Item> &canvas_item_owner) { +void _mark_ysort_dirty(RendererCanvasCull::Item *ysort_owner, RID_PtrOwner<RendererCanvasCull::Item, true> &canvas_item_owner) { do { ysort_owner->ysort_children_count = -1; ysort_owner = canvas_item_owner.owns(ysort_owner->parent) ? canvas_item_owner.getornull(ysort_owner->parent) : nullptr; @@ -148,6 +148,8 @@ void RendererCanvasCull::_cull_canvas_item(Item *p_canvas_item, const Transform2 } else { ci->final_clip_rect = global_rect; } + ci->final_clip_rect.position = ci->final_clip_rect.position.round(); + ci->final_clip_rect.size = ci->final_clip_rect.size.round(); ci->final_clip_owner = ci; } else { @@ -356,12 +358,12 @@ bool RendererCanvasCull::was_sdf_used() { return sdf_used; } -RID RendererCanvasCull::canvas_create() { +RID RendererCanvasCull::canvas_allocate() { + return canvas_owner.allocate_rid(); +} +void RendererCanvasCull::canvas_initialize(RID p_rid) { Canvas *canvas = memnew(Canvas); - ERR_FAIL_COND_V(!canvas, RID()); - RID rid = canvas_owner.make_rid(canvas); - - return rid; + canvas_owner.initialize_rid(p_rid, canvas); } void RendererCanvasCull::canvas_set_item_mirroring(RID p_canvas, RID p_item, const Point2 &p_mirroring) { @@ -393,11 +395,12 @@ void RendererCanvasCull::canvas_set_parent(RID p_canvas, RID p_parent, float p_s canvas->parent_scale = p_scale; } -RID RendererCanvasCull::canvas_item_create() { +RID RendererCanvasCull::canvas_item_allocate() { + return canvas_item_owner.allocate_rid(); +} +void RendererCanvasCull::canvas_item_initialize(RID p_rid) { Item *canvas_item = memnew(Item); - ERR_FAIL_COND_V(!canvas_item, RID()); - - return canvas_item_owner.make_rid(canvas_item); + canvas_item_owner.initialize_rid(p_rid, canvas_item); } void RendererCanvasCull::canvas_item_set_parent(RID p_item, RID p_parent) { @@ -524,11 +527,11 @@ void RendererCanvasCull::canvas_item_add_line(RID p_item, const Point2 &p_from, Item::CommandPrimitive *line = canvas_item->alloc_command<Item::CommandPrimitive>(); ERR_FAIL_COND(!line); if (p_width > 1.001) { - Vector2 t = (p_from - p_to).orthogonal().normalized(); - line->points[0] = p_from + t * p_width; - line->points[1] = p_from - t * p_width; - line->points[2] = p_to - t * p_width; - line->points[3] = p_to + t * p_width; + Vector2 t = (p_from - p_to).orthogonal().normalized() * p_width * 0.5; + line->points[0] = p_from + t; + line->points[1] = p_from - t; + line->points[2] = p_to - t; + line->points[3] = p_to + t; line->point_count = 4; } else { line->point_count = 2; @@ -721,8 +724,10 @@ void RendererCanvasCull::canvas_item_add_circle(RID p_item, const Point2 &p_pos, static const int circle_points = 64; points.resize(circle_points); + const real_t circle_point_step = Math_TAU / circle_points; + for (int i = 0; i < circle_points; i++) { - float angle = (i / float(circle_points)) * 2 * Math_PI; + float angle = i * circle_point_step; points.write[i].x = Math::cos(angle) * p_radius; points.write[i].y = Math::sin(angle) * p_radius; points.write[i] += p_pos; @@ -1073,10 +1078,13 @@ void RendererCanvasCull::canvas_item_set_canvas_group_mode(RID p_item, RS::Canva } } -RID RendererCanvasCull::canvas_light_create() { +RID RendererCanvasCull::canvas_light_allocate() { + return canvas_light_owner.allocate_rid(); +} +void RendererCanvasCull::canvas_light_initialize(RID p_rid) { RendererCanvasRender::Light *clight = memnew(RendererCanvasRender::Light); clight->light_internal = RSG::canvas_render->light_create(); - return canvas_light_owner.make_rid(clight); + return canvas_light_owner.initialize_rid(p_rid, clight); } void RendererCanvasCull::canvas_light_set_mode(RID p_light, RS::CanvasLightMode p_mode) { @@ -1266,10 +1274,13 @@ void RendererCanvasCull::canvas_light_set_shadow_smooth(RID p_light, float p_smo clight->shadow_smooth = p_smooth; } -RID RendererCanvasCull::canvas_light_occluder_create() { +RID RendererCanvasCull::canvas_light_occluder_allocate() { + return canvas_light_occluder_owner.allocate_rid(); +} +void RendererCanvasCull::canvas_light_occluder_initialize(RID p_rid) { RendererCanvasRender::LightOccluderInstance *occluder = memnew(RendererCanvasRender::LightOccluderInstance); - return canvas_light_occluder_owner.make_rid(occluder); + return canvas_light_occluder_owner.initialize_rid(p_rid, occluder); } void RendererCanvasCull::canvas_light_occluder_attach_to_canvas(RID p_occluder, RID p_canvas) { @@ -1347,10 +1358,13 @@ void RendererCanvasCull::canvas_light_occluder_set_light_mask(RID p_occluder, in occluder->light_mask = p_mask; } -RID RendererCanvasCull::canvas_occluder_polygon_create() { +RID RendererCanvasCull::canvas_occluder_polygon_allocate() { + return canvas_light_occluder_polygon_owner.allocate_rid(); +} +void RendererCanvasCull::canvas_occluder_polygon_initialize(RID p_rid) { LightOccluderPolygon *occluder_poly = memnew(LightOccluderPolygon); occluder_poly->occluder = RSG::canvas_render->occluder_polygon_create(); - return canvas_light_occluder_polygon_owner.make_rid(occluder_poly); + return canvas_light_occluder_polygon_owner.initialize_rid(p_rid, occluder_poly); } void RendererCanvasCull::canvas_occluder_polygon_set_shape(RID p_occluder_polygon, const Vector<Vector2> &p_shape, bool p_closed) { @@ -1391,8 +1405,11 @@ void RendererCanvasCull::canvas_set_shadow_texture_size(int p_size) { RSG::canvas_render->set_shadow_texture_size(p_size); } -RID RendererCanvasCull::canvas_texture_create() { - return RSG::storage->canvas_texture_create(); +RID RendererCanvasCull::canvas_texture_allocate() { + return RSG::storage->canvas_texture_allocate(); +} +void RendererCanvasCull::canvas_texture_initialize(RID p_rid) { + RSG::storage->canvas_texture_initialize(p_rid); } void RendererCanvasCull::canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) { diff --git a/servers/rendering/renderer_canvas_cull.h b/servers/rendering/renderer_canvas_cull.h index 7496a413ee..b71f8e5a9a 100644 --- a/servers/rendering/renderer_canvas_cull.h +++ b/servers/rendering/renderer_canvas_cull.h @@ -101,9 +101,9 @@ public: } }; - RID_PtrOwner<LightOccluderPolygon> canvas_light_occluder_polygon_owner; + RID_PtrOwner<LightOccluderPolygon, true> canvas_light_occluder_polygon_owner; - RID_PtrOwner<RendererCanvasRender::LightOccluderInstance> canvas_light_occluder_owner; + RID_PtrOwner<RendererCanvasRender::LightOccluderInstance, true> canvas_light_occluder_owner; struct Canvas : public RendererViewport::CanvasBase { Set<RID> viewports; @@ -148,9 +148,9 @@ public: } }; - mutable RID_PtrOwner<Canvas> canvas_owner; - RID_PtrOwner<Item> canvas_item_owner; - RID_PtrOwner<RendererCanvasRender::Light> canvas_light_owner; + mutable RID_PtrOwner<Canvas, true> canvas_owner; + RID_PtrOwner<Item, true> canvas_item_owner; + RID_PtrOwner<RendererCanvasRender::Light, true> canvas_light_owner; bool disable_scale; bool sdf_used = false; @@ -168,13 +168,17 @@ public: bool was_sdf_used(); - RID canvas_create(); + RID canvas_allocate(); + void canvas_initialize(RID p_rid); + void canvas_set_item_mirroring(RID p_canvas, RID p_item, const Point2 &p_mirroring); void canvas_set_modulate(RID p_canvas, const Color &p_color); void canvas_set_parent(RID p_canvas, RID p_parent, float p_scale); void canvas_set_disable_scale(bool p_disable); - RID canvas_item_create(); + RID canvas_item_allocate(); + void canvas_item_initialize(RID p_rid); + void canvas_item_set_parent(RID p_item, RID p_parent); void canvas_item_set_visible(RID p_item, bool p_visible); @@ -222,7 +226,9 @@ public: void canvas_item_set_canvas_group_mode(RID p_item, RS::CanvasGroupMode p_mode, float p_clear_margin = 5.0, bool p_fit_empty = false, float p_fit_margin = 0.0, bool p_blur_mipmaps = false); - RID canvas_light_create(); + RID canvas_light_allocate(); + void canvas_light_initialize(RID p_rid); + void canvas_light_set_mode(RID p_light, RS::CanvasLightMode p_mode); void canvas_light_attach_to_canvas(RID p_light, RID p_canvas); void canvas_light_set_enabled(RID p_light, bool p_enabled); @@ -246,7 +252,9 @@ public: void canvas_light_set_shadow_color(RID p_light, const Color &p_color); void canvas_light_set_shadow_smooth(RID p_light, float p_smooth); - RID canvas_light_occluder_create(); + RID canvas_light_occluder_allocate(); + void canvas_light_occluder_initialize(RID p_rid); + void canvas_light_occluder_attach_to_canvas(RID p_occluder, RID p_canvas); void canvas_light_occluder_set_enabled(RID p_occluder, bool p_enabled); void canvas_light_occluder_set_polygon(RID p_occluder, RID p_polygon); @@ -254,14 +262,18 @@ public: void canvas_light_occluder_set_transform(RID p_occluder, const Transform2D &p_xform); void canvas_light_occluder_set_light_mask(RID p_occluder, int p_mask); - RID canvas_occluder_polygon_create(); + RID canvas_occluder_polygon_allocate(); + void canvas_occluder_polygon_initialize(RID p_rid); + void canvas_occluder_polygon_set_shape(RID p_occluder_polygon, const Vector<Vector2> &p_shape, bool p_closed); void canvas_occluder_polygon_set_cull_mode(RID p_occluder_polygon, RS::CanvasOccluderPolygonCullMode p_mode); void canvas_set_shadow_texture_size(int p_size); - RID canvas_texture_create(); + RID canvas_texture_allocate(); + void canvas_texture_initialize(RID p_rid); + void canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture); void canvas_texture_set_shading_parameters(RID p_canvas_texture, const Color &p_base_color, float p_shininess); diff --git a/servers/rendering/renderer_rd/SCsub b/servers/rendering/renderer_rd/SCsub index 6a2e682c67..9c95f538ac 100644 --- a/servers/rendering/renderer_rd/SCsub +++ b/servers/rendering/renderer_rd/SCsub @@ -4,4 +4,5 @@ Import("env") env.add_source_files(env.servers_sources, "*.cpp") +SConscript("forward_clustered/SCsub") SConscript("shaders/SCsub") diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.cpp b/servers/rendering/renderer_rd/cluster_builder_rd.cpp new file mode 100644 index 0000000000..0fdd864d47 --- /dev/null +++ b/servers/rendering/renderer_rd/cluster_builder_rd.cpp @@ -0,0 +1,555 @@ +/*************************************************************************/ +/* cluster_builder_rd.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "cluster_builder_rd.h" +#include "servers/rendering/rendering_device.h" +#include "servers/rendering/rendering_server_globals.h" + +ClusterBuilderSharedDataRD::ClusterBuilderSharedDataRD() { + RD::VertexFormatID vertex_format; + + { + Vector<RD::VertexAttribute> attributes; + { + RD::VertexAttribute va; + va.format = RD::DATA_FORMAT_R32G32B32_SFLOAT; + va.stride = sizeof(float) * 3; + attributes.push_back(va); + } + vertex_format = RD::get_singleton()->vertex_format_create(attributes); + } + + { + Vector<String> versions; + versions.push_back(""); + cluster_render.cluster_render_shader.initialize(versions); + cluster_render.shader_version = cluster_render.cluster_render_shader.version_create(); + cluster_render.shader = cluster_render.cluster_render_shader.version_get_shader(cluster_render.shader_version, 0); + cluster_render.shader_pipelines[ClusterRender::PIPELINE_NORMAL] = RD::get_singleton()->render_pipeline_create(cluster_render.shader, RD::get_singleton()->framebuffer_format_create_empty(), vertex_format, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState(), 0); + RD::PipelineMultisampleState ms; + ms.sample_count = RD::TEXTURE_SAMPLES_4; + cluster_render.shader_pipelines[ClusterRender::PIPELINE_MSAA] = RD::get_singleton()->render_pipeline_create(cluster_render.shader, RD::get_singleton()->framebuffer_format_create_empty(), vertex_format, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), ms, RD::PipelineDepthStencilState(), RD::PipelineColorBlendState(), 0); + } + { + Vector<String> versions; + versions.push_back(""); + cluster_store.cluster_store_shader.initialize(versions); + cluster_store.shader_version = cluster_store.cluster_store_shader.version_create(); + cluster_store.shader = cluster_store.cluster_store_shader.version_get_shader(cluster_store.shader_version, 0); + cluster_store.shader_pipeline = RD::get_singleton()->compute_pipeline_create(cluster_store.shader); + } + { + Vector<String> versions; + versions.push_back(""); + cluster_debug.cluster_debug_shader.initialize(versions); + cluster_debug.shader_version = cluster_debug.cluster_debug_shader.version_create(); + cluster_debug.shader = cluster_debug.cluster_debug_shader.version_get_shader(cluster_debug.shader_version, 0); + cluster_debug.shader_pipeline = RD::get_singleton()->compute_pipeline_create(cluster_debug.shader); + } + + { // SPHERE + static const uint32_t icosphere_vertex_count = 42; + static const float icosphere_vertices[icosphere_vertex_count * 3] = { + 0, 0, -1, 0.7236073, -0.5257253, -0.4472195, -0.276388, -0.8506492, -0.4472199, -0.8944262, 0, -0.4472156, -0.276388, 0.8506492, -0.4472199, 0.7236073, 0.5257253, -0.4472195, 0.276388, -0.8506492, 0.4472199, -0.7236073, -0.5257253, 0.4472195, -0.7236073, 0.5257253, 0.4472195, 0.276388, 0.8506492, 0.4472199, 0.8944262, 0, 0.4472156, 0, 0, 1, -0.1624555, -0.4999952, -0.8506544, 0.4253227, -0.3090114, -0.8506542, 0.2628688, -0.8090116, -0.5257377, 0.8506479, 0, -0.5257359, 0.4253227, 0.3090114, -0.8506542, -0.5257298, 0, -0.8506517, -0.6881894, -0.4999969, -0.5257362, -0.1624555, 0.4999952, -0.8506544, -0.6881894, 0.4999969, -0.5257362, 0.2628688, 0.8090116, -0.5257377, 0.9510579, -0.3090126, 0, 0.9510579, 0.3090126, 0, 0, -1, 0, 0.5877856, -0.8090167, 0, -0.9510579, -0.3090126, 0, -0.5877856, -0.8090167, 0, -0.5877856, 0.8090167, 0, -0.9510579, 0.3090126, 0, 0.5877856, 0.8090167, 0, 0, 1, 0, 0.6881894, -0.4999969, 0.5257362, -0.2628688, -0.8090116, 0.5257377, -0.8506479, 0, 0.5257359, -0.2628688, 0.8090116, 0.5257377, 0.6881894, 0.4999969, 0.5257362, 0.1624555, -0.4999952, 0.8506544, 0.5257298, 0, 0.8506517, -0.4253227, -0.3090114, 0.8506542, -0.4253227, 0.3090114, 0.8506542, 0.1624555, 0.4999952, 0.8506544 + }; + static const uint32_t icosphere_triangle_count = 80; + static const uint32_t icosphere_triangle_indices[icosphere_triangle_count * 3] = { + 0, 13, 12, 1, 13, 15, 0, 12, 17, 0, 17, 19, 0, 19, 16, 1, 15, 22, 2, 14, 24, 3, 18, 26, 4, 20, 28, 5, 21, 30, 1, 22, 25, 2, 24, 27, 3, 26, 29, 4, 28, 31, 5, 30, 23, 6, 32, 37, 7, 33, 39, 8, 34, 40, 9, 35, 41, 10, 36, 38, 38, 41, 11, 38, 36, 41, 36, 9, 41, 41, 40, 11, 41, 35, 40, 35, 8, 40, 40, 39, 11, 40, 34, 39, 34, 7, 39, 39, 37, 11, 39, 33, 37, 33, 6, 37, 37, 38, 11, 37, 32, 38, 32, 10, 38, 23, 36, 10, 23, 30, 36, 30, 9, 36, 31, 35, 9, 31, 28, 35, 28, 8, 35, 29, 34, 8, 29, 26, 34, 26, 7, 34, 27, 33, 7, 27, 24, 33, 24, 6, 33, 25, 32, 6, 25, 22, 32, 22, 10, 32, 30, 31, 9, 30, 21, 31, 21, 4, 31, 28, 29, 8, 28, 20, 29, 20, 3, 29, 26, 27, 7, 26, 18, 27, 18, 2, 27, 24, 25, 6, 24, 14, 25, 14, 1, 25, 22, 23, 10, 22, 15, 23, 15, 5, 23, 16, 21, 5, 16, 19, 21, 19, 4, 21, 19, 20, 4, 19, 17, 20, 17, 3, 20, 17, 18, 3, 17, 12, 18, 12, 2, 18, 15, 16, 5, 15, 13, 16, 13, 0, 16, 12, 14, 2, 12, 13, 14, 13, 1, 14 + }; + + Vector<uint8_t> vertex_data; + vertex_data.resize(sizeof(float) * icosphere_vertex_count * 3); + copymem(vertex_data.ptrw(), icosphere_vertices, vertex_data.size()); + + sphere_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); + + Vector<uint8_t> index_data; + index_data.resize(sizeof(uint32_t) * icosphere_triangle_count * 3); + copymem(index_data.ptrw(), icosphere_triangle_indices, index_data.size()); + + sphere_index_buffer = RD::get_singleton()->index_buffer_create(icosphere_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + + Vector<RID> buffers; + buffers.push_back(sphere_vertex_buffer); + + sphere_vertex_array = RD::get_singleton()->vertex_array_create(icosphere_vertex_count, vertex_format, buffers); + + sphere_index_array = RD::get_singleton()->index_array_create(sphere_index_buffer, 0, icosphere_triangle_count * 3); + + float min_d = 1e20; + for (uint32_t i = 0; i < icosphere_triangle_count; i++) { + Vector3 vertices[3]; + for (uint32_t j = 0; j < 3; j++) { + uint32_t index = icosphere_triangle_indices[i * 3 + j]; + for (uint32_t k = 0; k < 3; k++) { + vertices[j][k] = icosphere_vertices[index * 3 + k]; + } + } + Plane p(vertices[0], vertices[1], vertices[2]); + min_d = MIN(Math::abs(p.d), min_d); + } + sphere_overfit = 1.0 / min_d; + } + + { // CONE + static const uint32_t cone_vertex_count = 99; + static const float cone_vertices[cone_vertex_count * 3] = { + 0, 1, -1, 0.1950903, 0.9807853, -1, 0.3826835, 0.9238795, -1, 0.5555703, 0.8314696, -1, 0.7071068, 0.7071068, -1, 0.8314697, 0.5555702, -1, 0.9238795, 0.3826834, -1, 0.9807853, 0.1950903, -1, 1, 0, -1, 0.9807853, -0.1950902, -1, 0.9238796, -0.3826833, -1, 0.8314697, -0.5555702, -1, 0.7071068, -0.7071068, -1, 0.5555702, -0.8314697, -1, 0.3826833, -0.9238796, -1, 0.1950901, -0.9807853, -1, -3.25841e-7, -1, -1, -0.1950907, -0.9807852, -1, -0.3826839, -0.9238793, -1, -0.5555707, -0.8314693, -1, -0.7071073, -0.7071063, -1, -0.83147, -0.5555697, -1, -0.9238799, -0.3826827, -1, 0, 0, 0, -0.9807854, -0.1950894, -1, -1, 9.65599e-7, -1, -0.9807851, 0.1950913, -1, -0.9238791, 0.3826845, -1, -0.8314689, 0.5555713, -1, -0.7071059, 0.7071077, -1, -0.5555691, 0.8314704, -1, -0.3826821, 0.9238801, -1, -0.1950888, 0.9807856, -1 + }; + static const uint32_t cone_triangle_count = 62; + static const uint32_t cone_triangle_indices[cone_triangle_count * 3] = { + 0, 23, 1, 1, 23, 2, 2, 23, 3, 3, 23, 4, 4, 23, 5, 5, 23, 6, 6, 23, 7, 7, 23, 8, 8, 23, 9, 9, 23, 10, 10, 23, 11, 11, 23, 12, 12, 23, 13, 13, 23, 14, 14, 23, 15, 15, 23, 16, 16, 23, 17, 17, 23, 18, 18, 23, 19, 19, 23, 20, 20, 23, 21, 21, 23, 22, 22, 23, 24, 24, 23, 25, 25, 23, 26, 26, 23, 27, 27, 23, 28, 28, 23, 29, 29, 23, 30, 30, 23, 31, 31, 23, 32, 32, 23, 0, 7, 15, 24, 32, 0, 1, 1, 2, 3, 3, 4, 5, 5, 6, 3, 6, 7, 3, 7, 8, 9, 9, 10, 7, 10, 11, 7, 11, 12, 15, 12, 13, 15, 13, 14, 15, 15, 16, 17, 17, 18, 19, 19, 20, 24, 20, 21, 24, 21, 22, 24, 24, 25, 26, 26, 27, 28, 28, 29, 30, 30, 31, 32, 32, 1, 3, 15, 17, 24, 17, 19, 24, 24, 26, 32, 26, 28, 32, 28, 30, 32, 32, 3, 7, 7, 11, 15, 32, 7, 24 + }; + + Vector<uint8_t> vertex_data; + vertex_data.resize(sizeof(float) * cone_vertex_count * 3); + copymem(vertex_data.ptrw(), cone_vertices, vertex_data.size()); + + cone_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); + + Vector<uint8_t> index_data; + index_data.resize(sizeof(uint32_t) * cone_triangle_count * 3); + copymem(index_data.ptrw(), cone_triangle_indices, index_data.size()); + + cone_index_buffer = RD::get_singleton()->index_buffer_create(cone_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + + Vector<RID> buffers; + buffers.push_back(cone_vertex_buffer); + + cone_vertex_array = RD::get_singleton()->vertex_array_create(cone_vertex_count, vertex_format, buffers); + + cone_index_array = RD::get_singleton()->index_array_create(cone_index_buffer, 0, cone_triangle_count * 3); + + float min_d = 1e20; + for (uint32_t i = 0; i < cone_triangle_count; i++) { + Vector3 vertices[3]; + int32_t zero_index = -1; + for (uint32_t j = 0; j < 3; j++) { + uint32_t index = cone_triangle_indices[i * 3 + j]; + for (uint32_t k = 0; k < 3; k++) { + vertices[j][k] = cone_vertices[index * 3 + k]; + } + if (vertices[j] == Vector3()) { + zero_index = j; + } + } + + if (zero_index != -1) { + Vector3 a = vertices[(zero_index + 1) % 3]; + Vector3 b = vertices[(zero_index + 2) % 3]; + Vector3 c = a + Vector3(0, 0, 1); + Plane p(a, b, c); + min_d = MIN(Math::abs(p.d), min_d); + } + } + cone_overfit = 1.0 / min_d; + } + + { // BOX + static const uint32_t box_vertex_count = 8; + static const float box_vertices[box_vertex_count * 3] = { + -1, -1, -1, -1, -1, 1, -1, 1, -1, -1, 1, 1, 1, -1, -1, 1, -1, 1, 1, 1, -1, 1, 1, 1 + }; + static const uint32_t box_triangle_count = 12; + static const uint32_t box_triangle_indices[box_triangle_count * 3] = { + 1, 2, 0, 3, 6, 2, 7, 4, 6, 5, 0, 4, 6, 0, 2, 3, 5, 7, 1, 3, 2, 3, 7, 6, 7, 5, 4, 5, 1, 0, 6, 4, 0, 3, 1, 5 + }; + + Vector<uint8_t> vertex_data; + vertex_data.resize(sizeof(float) * box_vertex_count * 3); + copymem(vertex_data.ptrw(), box_vertices, vertex_data.size()); + + box_vertex_buffer = RD::get_singleton()->vertex_buffer_create(vertex_data.size(), vertex_data); + + Vector<uint8_t> index_data; + index_data.resize(sizeof(uint32_t) * box_triangle_count * 3); + copymem(index_data.ptrw(), box_triangle_indices, index_data.size()); + + box_index_buffer = RD::get_singleton()->index_buffer_create(box_triangle_count * 3, RD::INDEX_BUFFER_FORMAT_UINT32, index_data); + + Vector<RID> buffers; + buffers.push_back(box_vertex_buffer); + + box_vertex_array = RD::get_singleton()->vertex_array_create(box_vertex_count, vertex_format, buffers); + + box_index_array = RD::get_singleton()->index_array_create(box_index_buffer, 0, box_triangle_count * 3); + } +} +ClusterBuilderSharedDataRD::~ClusterBuilderSharedDataRD() { + RD::get_singleton()->free(sphere_vertex_buffer); + RD::get_singleton()->free(sphere_index_buffer); + RD::get_singleton()->free(cone_vertex_buffer); + RD::get_singleton()->free(cone_index_buffer); + RD::get_singleton()->free(box_vertex_buffer); + RD::get_singleton()->free(box_index_buffer); + + cluster_render.cluster_render_shader.version_free(cluster_render.shader_version); + cluster_store.cluster_store_shader.version_free(cluster_store.shader_version); + cluster_debug.cluster_debug_shader.version_free(cluster_debug.shader_version); +} + +///////////////////////////// + +void ClusterBuilderRD::_clear() { + if (cluster_buffer.is_null()) { + return; //nothing to clear + } + RD::get_singleton()->free(cluster_buffer); + RD::get_singleton()->free(cluster_render_buffer); + RD::get_singleton()->free(element_buffer); + cluster_buffer = RID(); + cluster_render_buffer = RID(); + element_buffer = RID(); + + memfree(render_elements); + + render_elements = nullptr; + render_element_max = 0; + render_element_count = 0; + + RD::get_singleton()->free(framebuffer); + framebuffer = RID(); + + cluster_render_uniform_set = RID(); + cluster_store_uniform_set = RID(); +} + +void ClusterBuilderRD::setup(Size2i p_screen_size, uint32_t p_max_elements, RID p_depth_buffer, RID p_depth_buffer_sampler, RID p_color_buffer) { + ERR_FAIL_COND(p_max_elements == 0); + ERR_FAIL_COND(p_screen_size.x < 1); + ERR_FAIL_COND(p_screen_size.y < 1); + + _clear(); + + screen_size = p_screen_size; + + cluster_screen_size.width = (p_screen_size.width - 1) / cluster_size + 1; + cluster_screen_size.height = (p_screen_size.height - 1) / cluster_size + 1; + + max_elements_by_type = p_max_elements; + if (max_elements_by_type % 32) { //need to be 32 aligned + max_elements_by_type += 32 - (max_elements_by_type % 32); + } + + cluster_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (max_elements_by_type / 32 + 32) * ELEMENT_TYPE_MAX * 4; + + render_element_max = max_elements_by_type * ELEMENT_TYPE_MAX; + + uint32_t element_tag_bits_size = render_element_max / 32; + uint32_t element_tag_depth_bits_size = render_element_max; + cluster_render_buffer_size = cluster_screen_size.x * cluster_screen_size.y * (element_tag_bits_size + element_tag_depth_bits_size) * 4; // tag bits (element was used) and tag depth (depth range in which it was used) + + cluster_render_buffer = RD::get_singleton()->storage_buffer_create(cluster_render_buffer_size); + cluster_buffer = RD::get_singleton()->storage_buffer_create(cluster_buffer_size); + + render_elements = (RenderElementData *)memalloc(sizeof(RenderElementData *) * render_element_max); + render_element_count = 0; + + element_buffer = RD::get_singleton()->storage_buffer_create(sizeof(RenderElementData) * render_element_max); + + uint32_t div_value = 1 << divisor; + if (use_msaa) { + framebuffer = RD::get_singleton()->framebuffer_create_empty(p_screen_size / div_value, RD::TEXTURE_SAMPLES_4); + } else { + framebuffer = RD::get_singleton()->framebuffer_create_empty(p_screen_size / div_value); + } + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 1; + u.ids.push_back(state_uniform); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 2; + u.ids.push_back(element_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 3; + u.ids.push_back(cluster_render_buffer); + uniforms.push_back(u); + } + + cluster_render_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shared->cluster_render.shader, 0); + } + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(cluster_render_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 2; + u.ids.push_back(cluster_buffer); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 3; + u.ids.push_back(element_buffer); + uniforms.push_back(u); + } + + cluster_store_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shared->cluster_store.shader, 0); + } + + if (p_color_buffer.is_valid()) { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(cluster_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(p_color_buffer); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 3; + u.ids.push_back(p_depth_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 4; + u.ids.push_back(p_depth_buffer_sampler); + uniforms.push_back(u); + } + + debug_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shared->cluster_debug.shader, 0); + } else { + debug_uniform_set = RID(); + } +} + +void ClusterBuilderRD::begin(const Transform &p_view_transform, const CameraMatrix &p_cam_projection, bool p_flip_y) { + view_xform = p_view_transform.affine_inverse(); + projection = p_cam_projection; + z_near = projection.get_z_near(); + z_far = projection.get_z_far(); + orthogonal = p_cam_projection.is_orthogonal(); + adjusted_projection = projection; + if (!orthogonal) { + adjusted_projection.adjust_perspective_znear(0.0001); + } + + CameraMatrix correction; + correction.set_depth_correction(p_flip_y); + projection = correction * projection; + adjusted_projection = correction * adjusted_projection; + + //reset counts + render_element_count = 0; + for (uint32_t i = 0; i < ELEMENT_TYPE_MAX; i++) { + cluster_count_by_type[i] = 0; + } +} + +void ClusterBuilderRD::bake_cluster() { + RENDER_TIMESTAMP(">Bake Cluster"); + + RD::get_singleton()->draw_command_begin_label("Bake Light Cluster"); + + //clear cluster buffer + RD::get_singleton()->buffer_clear(cluster_buffer, 0, cluster_buffer_size, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); + + if (render_element_count > 0) { + //clear render buffer + RD::get_singleton()->buffer_clear(cluster_render_buffer, 0, cluster_render_buffer_size, RD::BARRIER_MASK_RASTER); + + { //fill state uniform + + StateUniform state; + + RendererStorageRD::store_camera(adjusted_projection, state.projection); + state.inv_z_far = 1.0 / z_far; + state.screen_to_clusters_shift = get_shift_from_power_of_2(cluster_size); + state.screen_to_clusters_shift -= divisor; //screen is smaller, shift one less + + state.cluster_screen_width = cluster_screen_size.x; + state.cluster_depth_offset = (render_element_max / 32); + state.cluster_data_size = state.cluster_depth_offset + render_element_max; + + RD::get_singleton()->buffer_update(state_uniform, 0, sizeof(StateUniform), &state, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); + } + + //update instances + + RD::get_singleton()->buffer_update(element_buffer, 0, sizeof(RenderElementData) * render_element_count, render_elements, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); + + RENDER_TIMESTAMP("Render Elements"); + + //render elements + { + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); + ClusterBuilderSharedDataRD::ClusterRender::PushConstant push_constant = {}; + + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, shared->cluster_render.shader_pipelines[use_msaa ? ClusterBuilderSharedDataRD::ClusterRender::PIPELINE_MSAA : ClusterBuilderSharedDataRD::ClusterRender::PIPELINE_NORMAL]); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, cluster_render_uniform_set, 0); + + for (uint32_t i = 0; i < render_element_count;) { + push_constant.base_index = i; + switch (render_elements[i].type) { + case ELEMENT_TYPE_OMNI_LIGHT: { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->sphere_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->sphere_index_array); + } break; + case ELEMENT_TYPE_SPOT_LIGHT: { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->cone_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->cone_index_array); + } break; + case ELEMENT_TYPE_DECAL: + case ELEMENT_TYPE_REFLECTION_PROBE: { + RD::get_singleton()->draw_list_bind_vertex_array(draw_list, shared->box_vertex_array); + RD::get_singleton()->draw_list_bind_index_array(draw_list, shared->box_index_array); + } break; + } + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(ClusterBuilderSharedDataRD::ClusterRender::PushConstant)); + + uint32_t instances = 1; +#if 0 + for (uint32_t j = i+1; j < element_count; j++) { + if (elements[i].type!=elements[j].type) { + break; + } + instances++; + } +#endif + RD::get_singleton()->draw_list_draw(draw_list, true, instances); + i += instances; + } + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_COMPUTE); + } + //store elements + RENDER_TIMESTAMP("Pack Elements"); + + { + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, shared->cluster_store.shader_pipeline); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cluster_store_uniform_set, 0); + + ClusterBuilderSharedDataRD::ClusterStore::PushConstant push_constant; + push_constant.cluster_render_data_size = render_element_max / 32 + render_element_max; + push_constant.max_render_element_count_div_32 = render_element_max / 32; + push_constant.cluster_screen_size[0] = cluster_screen_size.x; + push_constant.cluster_screen_size[1] = cluster_screen_size.y; + push_constant.render_element_count_div_32 = render_element_count > 0 ? (render_element_count - 1) / 32 + 1 : 0; + push_constant.max_cluster_element_count_div_32 = max_elements_by_type / 32; + push_constant.pad1 = 0; + push_constant.pad2 = 0; + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ClusterBuilderSharedDataRD::ClusterStore::PushConstant)); + + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cluster_screen_size.x, cluster_screen_size.y, 1); + + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); + } + } else { + RD::get_singleton()->barrier(RD::BARRIER_MASK_TRANSFER, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); + } + RENDER_TIMESTAMP("<Bake Cluster"); + RD::get_singleton()->draw_command_end_label(); +} + +void ClusterBuilderRD::debug(ElementType p_element) { + ERR_FAIL_COND(debug_uniform_set.is_null()); + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, shared->cluster_debug.shader_pipeline); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, debug_uniform_set, 0); + + ClusterBuilderSharedDataRD::ClusterDebug::PushConstant push_constant; + push_constant.screen_size[0] = screen_size.x; + push_constant.screen_size[1] = screen_size.y; + push_constant.cluster_screen_size[0] = cluster_screen_size.x; + push_constant.cluster_screen_size[1] = cluster_screen_size.y; + push_constant.cluster_shift = get_shift_from_power_of_2(cluster_size); + push_constant.cluster_type = p_element; + push_constant.orthogonal = orthogonal; + push_constant.z_far = z_far; + push_constant.z_near = z_near; + push_constant.max_cluster_element_count_div_32 = max_elements_by_type / 32; + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ClusterBuilderSharedDataRD::ClusterDebug::PushConstant)); + + RD::get_singleton()->compute_list_dispatch_threads(compute_list, screen_size.x, screen_size.y, 1); + + RD::get_singleton()->compute_list_end(); +} + +RID ClusterBuilderRD::get_cluster_buffer() const { + return cluster_buffer; +} + +uint32_t ClusterBuilderRD::get_cluster_size() const { + return cluster_size; +} + +uint32_t ClusterBuilderRD::get_max_cluster_elements() const { + return max_elements_by_type; +} + +void ClusterBuilderRD::set_shared(ClusterBuilderSharedDataRD *p_shared) { + shared = p_shared; +} + +ClusterBuilderRD::ClusterBuilderRD() { + state_uniform = RD::get_singleton()->uniform_buffer_create(sizeof(StateUniform)); +} + +ClusterBuilderRD::~ClusterBuilderRD() { + _clear(); + RD::get_singleton()->free(state_uniform); +} diff --git a/servers/rendering/renderer_rd/cluster_builder_rd.h b/servers/rendering/renderer_rd/cluster_builder_rd.h new file mode 100644 index 0000000000..dc1707b534 --- /dev/null +++ b/servers/rendering/renderer_rd/cluster_builder_rd.h @@ -0,0 +1,378 @@ +/*************************************************************************/ +/* cluster_builder_rd.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef CLUSTER_BUILDER_RD_H +#define CLUSTER_BUILDER_RD_H + +#include "servers/rendering/renderer_rd/renderer_storage_rd.h" +#include "servers/rendering/renderer_rd/shaders/cluster_debug.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cluster_render.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cluster_store.glsl.gen.h" + +class ClusterBuilderSharedDataRD { + friend class ClusterBuilderRD; + + RID sphere_vertex_buffer; + RID sphere_vertex_array; + RID sphere_index_buffer; + RID sphere_index_array; + float sphere_overfit = 0.0; //because an icosphere is not a perfect sphere, we need to enlarge it to cover the sphere area + + RID cone_vertex_buffer; + RID cone_vertex_array; + RID cone_index_buffer; + RID cone_index_array; + float cone_overfit = 0.0; //because an cone mesh is not a perfect sphere, we need to enlarge it to cover the actual cone area + + RID box_vertex_buffer; + RID box_vertex_array; + RID box_index_buffer; + RID box_index_array; + + enum Divisor { + DIVISOR_1, + DIVISOR_2, + DIVISOR_4, + }; + + struct ClusterRender { + struct PushConstant { + uint32_t base_index; + uint32_t pad0; + uint32_t pad1; + uint32_t pad2; + }; + + ClusterRenderShaderRD cluster_render_shader; + RID shader_version; + RID shader; + enum PipelineVersion { + PIPELINE_NORMAL, + PIPELINE_MSAA, + PIPELINE_MAX + }; + + RID shader_pipelines[PIPELINE_MAX]; + } cluster_render; + + struct ClusterStore { + struct PushConstant { + uint32_t cluster_render_data_size; // how much data for a single cluster takes + uint32_t max_render_element_count_div_32; //divided by 32 + uint32_t cluster_screen_size[2]; + uint32_t render_element_count_div_32; //divided by 32 + uint32_t max_cluster_element_count_div_32; //divided by 32 + uint32_t pad1; + uint32_t pad2; + }; + + ClusterStoreShaderRD cluster_store_shader; + RID shader_version; + RID shader; + RID shader_pipeline; + } cluster_store; + + struct ClusterDebug { + struct PushConstant { + uint32_t screen_size[2]; + uint32_t cluster_screen_size[2]; + + uint32_t cluster_shift; + uint32_t cluster_type; + float z_near; + float z_far; + + uint32_t orthogonal; + uint32_t max_cluster_element_count_div_32; + uint32_t pad1; + uint32_t pad2; + }; + + ClusterDebugShaderRD cluster_debug_shader; + RID shader_version; + RID shader; + RID shader_pipeline; + } cluster_debug; + +public: + ClusterBuilderSharedDataRD(); + ~ClusterBuilderSharedDataRD(); +}; + +class ClusterBuilderRD { +public: + enum LightType { + LIGHT_TYPE_OMNI, + LIGHT_TYPE_SPOT + }; + + enum BoxType { + BOX_TYPE_REFLECTION_PROBE, + BOX_TYPE_DECAL, + }; + + enum ElementType { + ELEMENT_TYPE_OMNI_LIGHT, + ELEMENT_TYPE_SPOT_LIGHT, + ELEMENT_TYPE_DECAL, + ELEMENT_TYPE_REFLECTION_PROBE, + ELEMENT_TYPE_MAX, + + }; + +private: + ClusterBuilderSharedDataRD *shared = nullptr; + + struct RenderElementData { + uint32_t type; //0-4 + uint32_t touches_near; + uint32_t touches_far; + uint32_t original_index; + float transform_inv[12]; //transposed transform for less space + float scale[3]; + uint32_t pad; + }; + + uint32_t cluster_count_by_type[ELEMENT_TYPE_MAX] = {}; + uint32_t max_elements_by_type = 0; + + RenderElementData *render_elements = nullptr; + uint32_t render_element_count = 0; + uint32_t render_element_max = 0; + + Transform view_xform; + CameraMatrix adjusted_projection; + CameraMatrix projection; + float z_far = 0; + float z_near = 0; + bool orthogonal = false; + + enum Divisor { + DIVISOR_1, + DIVISOR_2, + DIVISOR_4, + }; + + uint32_t cluster_size = 32; + bool use_msaa = true; + Divisor divisor = DIVISOR_4; + + Size2i screen_size; + Size2i cluster_screen_size; + + RID framebuffer; + RID cluster_render_buffer; //used for creating + RID cluster_buffer; //used for rendering + RID element_buffer; //used for storing, to hint element touches far plane or near plane + uint32_t cluster_render_buffer_size = 0; + uint32_t cluster_buffer_size = 0; + + RID cluster_render_uniform_set; + RID cluster_store_uniform_set; + + //persistent data + + void _clear(); + + struct StateUniform { + float projection[16]; + float inv_z_far; + uint32_t screen_to_clusters_shift; // shift to obtain coordinates in block indices + uint32_t cluster_screen_width; // + uint32_t cluster_data_size; // how much data for a single cluster takes + uint32_t cluster_depth_offset; + uint32_t pad0; + uint32_t pad1; + uint32_t pad2; + }; + + RID state_uniform; + + RID debug_uniform_set; + +public: + void setup(Size2i p_screen_size, uint32_t p_max_elements, RID p_depth_buffer, RID p_depth_buffer_sampler, RID p_color_buffer); + + void begin(const Transform &p_view_transform, const CameraMatrix &p_cam_projection, bool p_flip_y); + + _FORCE_INLINE_ void add_light(LightType p_type, const Transform &p_transform, float p_radius, float p_spot_aperture) { + if (p_type == LIGHT_TYPE_OMNI && cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT] == max_elements_by_type) { + return; //max number elements reached + } + if (p_type == LIGHT_TYPE_SPOT && cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT] == max_elements_by_type) { + return; //max number elements reached + } + + RenderElementData &e = render_elements[render_element_count]; + + Transform xform = view_xform * p_transform; + + float radius = xform.basis.get_uniform_scale(); + if (radius > 0.98 || radius < 1.02) { + xform.basis.orthonormalize(); + } + + radius *= p_radius; + + if (p_type == LIGHT_TYPE_OMNI) { + radius *= shared->sphere_overfit; // overfit icosphere + + //omni + float depth = -xform.origin.z; + if (orthogonal) { + e.touches_near = (depth - radius) < z_near; + } else { + //contains camera inside light + float radius2 = radius * shared->sphere_overfit; // overfit again for outer size (camera may be outside actual sphere but behind an icosphere vertex) + e.touches_near = xform.origin.length_squared() < radius2 * radius2; + } + + e.touches_far = (depth + radius) > z_far; + e.scale[0] = radius; + e.scale[1] = radius; + e.scale[2] = radius; + e.type = ELEMENT_TYPE_OMNI_LIGHT; + e.original_index = cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT]; + + RendererStorageRD::store_transform_transposed_3x4(xform, e.transform_inv); + + cluster_count_by_type[ELEMENT_TYPE_OMNI_LIGHT]++; + + } else { + //spot + radius *= shared->cone_overfit; // overfit icosphere + + real_t len = Math::tan(Math::deg2rad(p_spot_aperture)) * radius; + //approximate, probably better to use a cone support function + float max_d = -1e20; + float min_d = 1e20; +#define CONE_MINMAX(m_x, m_y) \ + { \ + float d = -xform.xform(Vector3(len * m_x, len * m_y, -radius)).z; \ + min_d = MIN(d, min_d); \ + max_d = MAX(d, max_d); \ + } + + CONE_MINMAX(1, 1); + CONE_MINMAX(-1, 1); + CONE_MINMAX(-1, -1); + CONE_MINMAX(1, -1); + + if (orthogonal) { + e.touches_near = min_d < z_near; + } else { + //contains camera inside light + Plane base_plane(xform.origin, -xform.basis.get_axis(Vector3::AXIS_Z)); + float dist = base_plane.distance_to(Vector3()); + if (dist >= 0 && dist < radius) { + //inside, check angle + float angle = Math::rad2deg(Math::acos((-xform.origin.normalized()).dot(-xform.basis.get_axis(Vector3::AXIS_Z)))); + e.touches_near = angle < p_spot_aperture * 1.05; //overfit aperture a little due to cone overfit + } else { + e.touches_near = false; + } + } + + e.touches_far = max_d > z_far; + + e.scale[0] = len * shared->cone_overfit; + e.scale[1] = len * shared->cone_overfit; + e.scale[2] = radius; + + e.type = ELEMENT_TYPE_SPOT_LIGHT; + e.original_index = cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]; //use omni since they share index + + RendererStorageRD::store_transform_transposed_3x4(xform, e.transform_inv); + + cluster_count_by_type[ELEMENT_TYPE_SPOT_LIGHT]++; + } + + render_element_count++; + } + + _FORCE_INLINE_ void add_box(BoxType p_box_type, const Transform &p_transform, const Vector3 &p_half_extents) { + if (p_box_type == BOX_TYPE_DECAL && cluster_count_by_type[ELEMENT_TYPE_DECAL] == max_elements_by_type) { + return; //max number elements reached + } + if (p_box_type == BOX_TYPE_REFLECTION_PROBE && cluster_count_by_type[ELEMENT_TYPE_REFLECTION_PROBE] == max_elements_by_type) { + return; //max number elements reached + } + + RenderElementData &e = render_elements[render_element_count]; + Transform xform = view_xform * p_transform; + + //extract scale and scale the matrix by it, makes things simpler + Vector3 scale = p_half_extents; + for (uint32_t i = 0; i < 3; i++) { + float s = xform.basis.elements[i].length(); + scale[i] *= s; + xform.basis.elements[i] /= s; + }; + + float box_depth = Math::abs(xform.basis.xform_inv(Vector3(0, 0, -1)).dot(scale)); + float depth = -xform.origin.z; + + if (orthogonal) { + e.touches_near = depth - box_depth < z_near; + } else { + //contains camera inside box + Vector3 inside = xform.xform_inv(Vector3(0, 0, 0)).abs(); + e.touches_near = inside.x < scale.x && inside.y < scale.y && inside.z < scale.z; + } + + e.touches_far = depth + box_depth > z_far; + + e.scale[0] = scale.x; + e.scale[1] = scale.y; + e.scale[2] = scale.z; + + e.type = (p_box_type == BOX_TYPE_DECAL) ? ELEMENT_TYPE_DECAL : ELEMENT_TYPE_REFLECTION_PROBE; + e.original_index = cluster_count_by_type[e.type]; + + RendererStorageRD::store_transform_transposed_3x4(xform, e.transform_inv); + + cluster_count_by_type[e.type]++; + render_element_count++; + } + + void bake_cluster(); + void debug(ElementType p_element); + + RID get_cluster_buffer() const; + uint32_t get_cluster_size() const; + uint32_t get_max_cluster_elements() const; + + void set_shared(ClusterBuilderSharedDataRD *p_shared); + + ClusterBuilderRD(); + ~ClusterBuilderRD(); +}; + +#endif // CLUSTER_BUILDER_H diff --git a/servers/rendering/renderer_rd/effects_rd.cpp b/servers/rendering/renderer_rd/effects_rd.cpp index 6e1d61ff94..bc304aedd8 100644 --- a/servers/rendering/renderer_rd/effects_rd.cpp +++ b/servers/rendering/renderer_rd/effects_rd.cpp @@ -299,15 +299,12 @@ void EffectsRD::copy_to_rect(RID p_source_rd_texture, RID p_dest_texture, const copy.push_constant.target[0] = p_rect.position.x; copy.push_constant.target[1] = p_rect.position.y; - int32_t x_groups = (p_rect.size.width - 1) / 8 + 1; - int32_t y_groups = (p_rect.size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[p_8_bit_dst ? COPY_MODE_SIMPLY_COPY_8BIT : COPY_MODE_SIMPLY_COPY]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_rect.size.width, p_rect.size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -322,15 +319,12 @@ void EffectsRD::copy_cubemap_to_panorama(RID p_source_cube, RID p_dest_panorama, copy.push_constant.target[1] = 0; copy.push_constant.camera_z_far = p_lod; - int32_t x_groups = (p_panorama_size.width - 1) / 8 + 1; - int32_t y_groups = (p_panorama_size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[p_is_array ? COPY_MODE_CUBE_ARRAY_TO_PANORAMA : COPY_MODE_CUBE_TO_PANORAMA]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_cube), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_panorama), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_panorama_size.width, p_panorama_size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -349,15 +343,12 @@ void EffectsRD::copy_depth_to_rect_and_linearize(RID p_source_rd_texture, RID p_ copy.push_constant.camera_z_far = p_z_far; copy.push_constant.camera_z_near = p_z_near; - int32_t x_groups = (p_rect.size.width - 1) / 8 + 1; - int32_t y_groups = (p_rect.size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[COPY_MODE_LINEARIZE_DEPTH]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_rect.size.width, p_rect.size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -374,15 +365,12 @@ void EffectsRD::copy_depth_to_rect(RID p_source_rd_texture, RID p_dest_texture, copy.push_constant.target[0] = p_rect.position.x; copy.push_constant.target[1] = p_rect.position.y; - int32_t x_groups = (p_rect.size.width - 1) / 8 + 1; - int32_t y_groups = (p_rect.size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[COPY_MODE_SIMPLY_COPY_DEPTH]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_rect.size.width, p_rect.size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -400,14 +388,11 @@ void EffectsRD::set_color(RID p_dest_texture, const Color &p_color, const Rect2i copy.push_constant.set_color[2] = p_color.b; copy.push_constant.set_color[3] = p_color.a; - int32_t x_groups = (p_region.size.width - 1) / 8 + 1; - int32_t y_groups = (p_region.size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[p_8bit_dst ? COPY_MODE_SET_COLOR_8BIT : COPY_MODE_SET_COLOR]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_region.size.width, p_region.size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -420,8 +405,6 @@ void EffectsRD::gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back copy.push_constant.section[2] = p_region.size.width; copy.push_constant.section[3] = p_region.size.height; - int32_t x_groups = (p_region.size.width - 1) / 8 + 1; - int32_t y_groups = (p_region.size.height - 1) / 8 + 1; //HORIZONTAL RD::DrawListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[p_8bit_dst ? COPY_MODE_GAUSSIAN_COPY_8BIT : COPY_MODE_GAUSSIAN_COPY]); @@ -431,7 +414,7 @@ void EffectsRD::gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back copy.push_constant.flags = base_flags | COPY_FLAG_HORIZONTAL; RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_region.size.width, p_region.size.height, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); @@ -442,7 +425,7 @@ void EffectsRD::gaussian_blur(RID p_source_rd_texture, RID p_texture, RID p_back copy.push_constant.flags = base_flags; RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_region.size.width, p_region.size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -452,9 +435,6 @@ void EffectsRD::gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const CopyMode copy_mode = p_first_pass && p_auto_exposure.is_valid() ? COPY_MODE_GAUSSIAN_GLOW_AUTO_EXPOSURE : COPY_MODE_GAUSSIAN_GLOW; uint32_t base_flags = 0; - int32_t x_groups = (p_size.width + 7) / 8; - int32_t y_groups = (p_size.height + 7) / 8; - copy.push_constant.section[2] = p_size.x; copy.push_constant.section[3] = p_size.y; @@ -479,16 +459,13 @@ void EffectsRD::gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const copy.push_constant.flags = base_flags | (p_first_pass ? COPY_FLAG_GLOW_FIRST_PASS : 0) | (p_high_quality ? COPY_FLAG_HIGH_QUALITY_GLOW : 0); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_size.width, p_size.height, 1); RD::get_singleton()->compute_list_end(); } void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, RenderingServer::EnvironmentSSRRoughnessQuality p_roughness_quality, RID p_blur_radius, RID p_blur_radius2, RID p_metallic, const Color &p_metallic_mask, RID p_depth, RID p_scale_depth, RID p_scale_normal, RID p_output, RID p_output_blur, const Size2i &p_screen_size, int p_max_steps, float p_fade_in, float p_fade_out, float p_tolerance, const CameraMatrix &p_camera) { RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - int32_t x_groups = (p_screen_size.width - 1) / 8 + 1; - int32_t y_groups = (p_screen_size.height - 1) / 8 + 1; - { //scale color and depth to half ssr_scale.push_constant.camera_z_far = p_camera.get_z_far(); ssr_scale.push_constant.camera_z_near = p_camera.get_z_near(); @@ -506,7 +483,7 @@ void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, R RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssr_scale.push_constant, sizeof(ScreenSpaceReflectionScalePushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); } @@ -547,7 +524,7 @@ void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, R } RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_scale_normal), 2); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); } if (p_roughness_quality != RS::ENV_SSR_ROUGNESS_QUALITY_DISABLED) { @@ -585,7 +562,7 @@ void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, R RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssr_filter.push_constant, sizeof(ScreenSpaceReflectionFilterPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); @@ -600,7 +577,7 @@ void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, R RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssr_filter.push_constant, sizeof(ScreenSpaceReflectionFilterPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); } RD::get_singleton()->compute_list_end(); @@ -609,9 +586,6 @@ void EffectsRD::screen_space_reflection(RID p_diffuse, RID p_normal_roughness, R void EffectsRD::sub_surface_scattering(RID p_diffuse, RID p_diffuse2, RID p_depth, const CameraMatrix &p_camera, const Size2i &p_screen_size, float p_scale, float p_depth_scale, RenderingServer::SubSurfaceScatteringQuality p_quality) { RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - int32_t x_groups = (p_screen_size.width - 1) / 8 + 1; - int32_t y_groups = (p_screen_size.height - 1) / 8 + 1; - Plane p = p_camera.xform4(Plane(1, 0, -1, 1)); p.normal /= p.d; float unit_size = p.normal.x; @@ -635,7 +609,7 @@ void EffectsRD::sub_surface_scattering(RID p_diffuse, RID p_diffuse2, RID p_dept RD::get_singleton()->compute_list_set_push_constant(compute_list, &sss.push_constant, sizeof(SubSurfaceScatteringPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); @@ -646,7 +620,7 @@ void EffectsRD::sub_surface_scattering(RID p_diffuse, RID p_diffuse2, RID p_dept sss.push_constant.vertical = true; RD::get_singleton()->compute_list_set_push_constant(compute_list, &sss.push_constant, sizeof(SubSurfaceScatteringPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.width, p_screen_size.height, 1); RD::get_singleton()->compute_list_end(); } @@ -690,39 +664,33 @@ void EffectsRD::make_mipmap(RID p_source_rd_texture, RID p_dest_texture, const S copy.push_constant.section[2] = p_size.width; copy.push_constant.section[3] = p_size.height; - int32_t x_groups = (p_size.width - 1) / 8 + 1; - int32_t y_groups = (p_size.height - 1) / 8 + 1; - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, copy.pipelines[COPY_MODE_MIPMAP]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 3); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©.push_constant, sizeof(CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_size.width, p_size.height, 1); RD::get_singleton()->compute_list_end(); } -void EffectsRD::copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dest_texture, const Rect2i &p_rect, float p_z_near, float p_z_far, float p_bias, bool p_dp_flip) { +void EffectsRD::copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dst_framebuffer, const Rect2 &p_rect, float p_z_near, float p_z_far, bool p_dp_flip) { CopyToDPPushConstant push_constant; - push_constant.screen_size[0] = p_rect.size.x; - push_constant.screen_size[1] = p_rect.size.y; - push_constant.dest_offset[0] = p_rect.position.x; - push_constant.dest_offset[1] = p_rect.position.y; - push_constant.bias = p_bias; + push_constant.screen_rect[0] = p_rect.position.x; + push_constant.screen_rect[1] = p_rect.position.y; + push_constant.screen_rect[2] = p_rect.size.width; + push_constant.screen_rect[3] = p_rect.size.height; push_constant.z_far = p_z_far; push_constant.z_near = p_z_near; push_constant.z_flip = p_dp_flip; - int32_t x_groups = (p_rect.size.width - 1) / 8 + 1; - int32_t y_groups = (p_rect.size.height - 1) / 8 + 1; + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dst_framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, cube_to_dp.pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_rd_texture), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, cube_to_dp.pipeline); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_rd_texture), 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_texture), 1); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(CopyToDPPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); - RD::get_singleton()->compute_list_end(); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(CopyToDPPushConstant)); + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_TRANSFER); } void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const TonemapSettings &p_settings) { @@ -807,10 +775,7 @@ void EffectsRD::luminance_reduction(RID p_source_texture, const Size2i p_source_ RD::get_singleton()->compute_list_set_push_constant(compute_list, &luminance_reduce.push_constant, sizeof(LuminanceReducePushConstant)); - int32_t x_groups = (luminance_reduce.push_constant.source_size[0] - 1) / 8 + 1; - int32_t y_groups = (luminance_reduce.push_constant.source_size[1] - 1) / 8 + 1; - - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, luminance_reduce.push_constant.source_size[0], luminance_reduce.push_constant.source_size[1], 1); luminance_reduce.push_constant.source_size[0] = MAX(luminance_reduce.push_constant.source_size[0] / 8, 1); luminance_reduce.push_constant.source_size[1] = MAX(luminance_reduce.push_constant.source_size[1] / 8, 1); @@ -851,14 +816,12 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_depth_texture), 1); - int32_t x_groups = (p_base_texture_size.x - 1) / 8 + 1; - int32_t y_groups = (p_base_texture_size.y - 1) / 8 + 1; bokeh.push_constant.size[0] = p_base_texture_size.x; bokeh.push_constant.size[1] = p_base_texture_size.y; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); if (p_bokeh_shape == RS::DOF_BOKEH_BOX || p_bokeh_shape == RS::DOF_BOKEH_HEXAGON) { @@ -875,8 +838,6 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_halfsize_texture1), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_base_texture), 1); - x_groups = ((p_base_texture_size.x >> 1) - 1) / 8 + 1; - y_groups = ((p_base_texture_size.y >> 1) - 1) / 8 + 1; bokeh.push_constant.size[0] = p_base_texture_size.x >> 1; bokeh.push_constant.size[1] = p_base_texture_size.y >> 1; bokeh.push_constant.half_size = true; @@ -890,7 +851,7 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, bokeh.push_constant.size[0], bokeh.push_constant.size[1], 1); RD::get_singleton()->compute_list_add_barrier(compute_list); //third pass @@ -906,7 +867,7 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, bokeh.push_constant.size[0], bokeh.push_constant.size[1], 1); RD::get_singleton()->compute_list_add_barrier(compute_list); if (p_quality == RS::DOF_BLUR_QUALITY_VERY_LOW || p_quality == RS::DOF_BLUR_QUALITY_LOW) { @@ -917,8 +878,6 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_halfsize_texture2), 1); - x_groups = (p_base_texture_size.x - 1) / 8 + 1; - y_groups = (p_base_texture_size.y - 1) / 8 + 1; bokeh.push_constant.size[0] = p_base_texture_size.x; bokeh.push_constant.size[1] = p_base_texture_size.y; bokeh.push_constant.half_size = false; @@ -926,7 +885,7 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); } } else { //circle @@ -944,15 +903,13 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_halfsize_texture1), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_base_texture), 1); - x_groups = ((p_base_texture_size.x >> 1) - 1) / 8 + 1; - y_groups = ((p_base_texture_size.y >> 1) - 1) / 8 + 1; bokeh.push_constant.size[0] = p_base_texture_size.x >> 1; bokeh.push_constant.size[1] = p_base_texture_size.y >> 1; bokeh.push_constant.half_size = true; RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, bokeh.push_constant.size[0], bokeh.push_constant.size[1], 1); RD::get_singleton()->compute_list_add_barrier(compute_list); //circle is just one pass, then upscale @@ -964,8 +921,6 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_base_texture), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_halfsize_texture1), 1); - x_groups = (p_base_texture_size.x - 1) / 8 + 1; - y_groups = (p_base_texture_size.y - 1) / 8 + 1; bokeh.push_constant.size[0] = p_base_texture_size.x; bokeh.push_constant.size[1] = p_base_texture_size.y; bokeh.push_constant.half_size = false; @@ -973,7 +928,7 @@ void EffectsRD::bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i RD::get_singleton()->compute_list_set_push_constant(compute_list, &bokeh.push_constant, sizeof(BokehPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_base_texture_size.x, p_base_texture_size.y, 1); } RD::get_singleton()->compute_list_end(); @@ -998,20 +953,20 @@ void EffectsRD::gather_ssao(RD::ComputeListID p_compute_list, const Vector<RID> RD::get_singleton()->compute_list_bind_uniform_set(p_compute_list, _get_uniform_set_from_image(p_ao_slices[i]), 2); RD::get_singleton()->compute_list_set_push_constant(p_compute_list, &ssao.gather_push_constant, sizeof(SSAOGatherPushConstant)); - int x_groups = ((p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; - int y_groups = ((p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; + Size2i size = Size2i(p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1), p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1)); - RD::get_singleton()->compute_list_dispatch(p_compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(p_compute_list, size.x, size.y, 1); } RD::get_singleton()->compute_list_add_barrier(p_compute_list); } void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_depth_mipmaps_texture, const Vector<RID> &p_depth_mipmaps, RID p_ao, const Vector<RID> p_ao_slices, RID p_ao_pong, const Vector<RID> p_ao_pong_slices, RID p_upscale_buffer, RID p_importance_map, RID p_importance_map_pong, const CameraMatrix &p_projection, const SSAOSettings &p_settings, bool p_invalidate_uniform_sets) { RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - + RD::get_singleton()->draw_command_begin_label("SSAO"); /* FIRST PASS */ // Downsample and deinterleave the depth buffer. { + RD::get_singleton()->draw_command_begin_label("Downsample Depth"); if (p_invalidate_uniform_sets) { Vector<RD::Uniform> uniforms; { @@ -1074,16 +1029,17 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep } RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.downsample_push_constant, sizeof(SSAODownsamplePushConstant)); - int x_groups = (MAX(1, p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; - int y_groups = (MAX(1, p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; + Size2i size(MAX(1, p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1)), MAX(1, p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1))); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, size.x, size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); + RD::get_singleton()->draw_command_end_label(); // Downsample SSAO } /* SECOND PASS */ // Sample SSAO { + RD::get_singleton()->draw_command_begin_label("Gather Samples"); ssao.gather_push_constant.screen_size[0] = p_settings.full_screen_size.x; ssao.gather_push_constant.screen_size[1] = p_settings.full_screen_size.y; @@ -1184,6 +1140,7 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep } if (p_settings.quality == RS::ENV_SSAO_QUALITY_ULTRA) { + RD::get_singleton()->draw_command_begin_label("Generate Importance Map"); ssao.importance_map_push_constant.half_screen_pixel_size[0] = 1.0 / p_settings.half_screen_size.x; ssao.importance_map_push_constant.half_screen_pixel_size[1] = 1.0 / p_settings.half_screen_size.y; ssao.importance_map_push_constant.intensity = p_settings.intensity; @@ -1192,21 +1149,19 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_GATHER_BASE]); gather_ssao(compute_list, p_ao_pong_slices, p_settings, true); //generate importance map - int x_groups = (p_settings.quarter_screen_size.x - 1) / 8 + 1; - int y_groups = (p_settings.quarter_screen_size.y - 1) / 8 + 1; RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_GENERATE_IMPORTANCE_MAP]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_ao_pong), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_importance_map), 1); RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.importance_map_push_constant, sizeof(SSAOImportanceMapPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_settings.quarter_screen_size.x, p_settings.quarter_screen_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); //process importance map A RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_PROCESS_IMPORTANCE_MAPA]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_importance_map), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_importance_map_pong), 1); RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.importance_map_push_constant, sizeof(SSAOImportanceMapPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_settings.quarter_screen_size.x, p_settings.quarter_screen_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); //process Importance Map B RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_PROCESS_IMPORTANCE_MAPB]); @@ -1214,21 +1169,24 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_importance_map), 1); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, ssao.counter_uniform_set, 2); RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.importance_map_push_constant, sizeof(SSAOImportanceMapPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_settings.quarter_screen_size.x, p_settings.quarter_screen_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_GATHER_ADAPTIVE]); + RD::get_singleton()->draw_command_end_label(); // Importance Map } else { RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, ssao.pipelines[SSAO_GATHER]); } gather_ssao(compute_list, p_ao_slices, p_settings, false); + RD::get_singleton()->draw_command_end_label(); // Gather SSAO } // /* THIRD PASS */ // // Blur // { + RD::get_singleton()->draw_command_begin_label("Edge Aware Blur"); ssao.blur_push_constant.edge_sharpness = 1.0 - p_settings.sharpness; ssao.blur_push_constant.half_screen_pixel_size[0] = 1.0 / p_settings.half_screen_size.x; ssao.blur_push_constant.half_screen_pixel_size[1] = 1.0 / p_settings.half_screen_size.y; @@ -1268,22 +1226,22 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep } RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.blur_push_constant, sizeof(SSAOBlurPushConstant)); - int x_groups = ((p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; - int y_groups = ((p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1)) - 1) / 8 + 1; - - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + Size2i size(p_settings.full_screen_size.x >> (p_settings.half_size ? 2 : 1), p_settings.full_screen_size.y >> (p_settings.half_size ? 2 : 1)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, size.x, size.y, 1); } if (p_settings.quality > RS::ENV_SSAO_QUALITY_VERY_LOW) { RD::get_singleton()->compute_list_add_barrier(compute_list); } } + RD::get_singleton()->draw_command_end_label(); // Blur } /* FOURTH PASS */ // Interleave buffers // back to full size { + RD::get_singleton()->draw_command_begin_label("Interleave Buffers"); ssao.interleave_push_constant.inv_sharpness = 1.0 - p_settings.sharpness; ssao.interleave_push_constant.pixel_size[0] = 1.0 / p_settings.full_screen_size.x; ssao.interleave_push_constant.pixel_size[1] = 1.0 / p_settings.full_screen_size.y; @@ -1307,17 +1265,15 @@ void EffectsRD::generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_dep RD::get_singleton()->compute_list_set_push_constant(compute_list, &ssao.interleave_push_constant, sizeof(SSAOInterleavePushConstant)); - int x_groups = (p_settings.full_screen_size.x - 1) / 8 + 1; - int y_groups = (p_settings.full_screen_size.y - 1) / 8 + 1; - - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_settings.full_screen_size.x, p_settings.full_screen_size.y, 1); RD::get_singleton()->compute_list_add_barrier(compute_list); + RD::get_singleton()->draw_command_end_label(); // Interleave } - - RD::get_singleton()->compute_list_end(); + RD::get_singleton()->draw_command_end_label(); //SSAO + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_TRANSFER); //wait for upcoming transfer int zero[1] = { 0 }; - RD::get_singleton()->buffer_update(ssao.importance_map_load_counter, 0, sizeof(uint32_t), &zero, false); + RD::get_singleton()->buffer_update(ssao.importance_map_load_counter, 0, sizeof(uint32_t), &zero, 0); //no barrier } void EffectsRD::roughness_limit(RID p_source_normal, RID p_roughness, const Size2i &p_size, float p_curve) { @@ -1330,12 +1286,9 @@ void EffectsRD::roughness_limit(RID p_source_normal, RID p_roughness, const Size RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_normal), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_roughness), 1); - int x_groups = (p_size.x - 1) / 8 + 1; - int y_groups = (p_size.y - 1) / 8 + 1; - RD::get_singleton()->compute_list_set_push_constant(compute_list, &roughness_limiter.push_constant, sizeof(RoughnessLimiterPushConstant)); //not used but set anyway - RD::get_singleton()->compute_list_dispatch(compute_list, x_groups, y_groups, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_size.x, p_size.y, 1); RD::get_singleton()->compute_list_end(); } @@ -1448,7 +1401,7 @@ void EffectsRD::render_sky(RD::DrawListID p_list, float p_time, RID p_fb, RID p_ RD::get_singleton()->draw_list_draw(draw_list, true); } -void EffectsRD::resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_giprobe, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_giprobe, Vector2i p_screen_size, int p_samples) { +void EffectsRD::resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_giprobe, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_giprobe, Vector2i p_screen_size, int p_samples, uint32_t p_barrier) { ResolvePushConstant push_constant; push_constant.screen_size[0] = p_screen_size.x; push_constant.screen_size[1] = p_screen_size.y; @@ -1465,54 +1418,9 @@ void EffectsRD::resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RI RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ResolvePushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.x, p_screen_size.y, 1, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_screen_size.x, p_screen_size.y, 1); - RD::get_singleton()->compute_list_end(); -} - -void EffectsRD::reduce_shadow(RID p_source_shadow, RID p_dest_shadow, const Size2i &p_source_size, const Rect2i &p_source_rect, int p_shrink_limit, RD::ComputeListID compute_list) { - uint32_t push_constant[8] = { (uint32_t)p_source_size.x, (uint32_t)p_source_size.y, (uint32_t)p_source_rect.position.x, (uint32_t)p_source_rect.position.y, (uint32_t)p_shrink_limit, 0, 0, 0 }; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, shadow_reduce.pipelines[SHADOW_REDUCE_REDUCE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_image_pair(p_source_shadow, p_dest_shadow), 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(uint32_t) * 8); - - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_source_rect.size.width, p_source_rect.size.height, 1, 8, 8, 1); -} -void EffectsRD::filter_shadow(RID p_shadow, RID p_backing_shadow, const Size2i &p_source_size, const Rect2i &p_source_rect, RenderingServer::EnvVolumetricFogShadowFilter p_filter, RD::ComputeListID compute_list, bool p_vertical, bool p_horizontal) { - uint32_t push_constant[8] = { (uint32_t)p_source_size.x, (uint32_t)p_source_size.y, (uint32_t)p_source_rect.position.x, (uint32_t)p_source_rect.position.y, 0, 0, 0, 0 }; - - switch (p_filter) { - case RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_DISABLED: - case RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_LOW: { - push_constant[5] = 0; - } break; - case RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_MEDIUM: { - push_constant[5] = 9; - } break; - case RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_HIGH: { - push_constant[5] = 18; - } break; - } - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, shadow_reduce.pipelines[SHADOW_REDUCE_FILTER]); - if (p_vertical) { - push_constant[6] = 1; - push_constant[7] = 0; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_image_pair(p_shadow, p_backing_shadow), 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(uint32_t) * 8); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_source_rect.size.width, p_source_rect.size.height, 1, 8, 8, 1); - } - if (p_vertical && p_horizontal) { - RD::get_singleton()->compute_list_add_barrier(compute_list); - } - if (p_horizontal) { - push_constant[6] = 0; - push_constant[7] = 1; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_image_pair(p_backing_shadow, p_shadow), 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(uint32_t) * 8); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_source_rect.size.width, p_source_rect.size.height, 1, 8, 8, 1); - } + RD::get_singleton()->compute_list_end(p_barrier); } void EffectsRD::sort_buffer(RID p_uniform_set, int p_size) { @@ -1678,8 +1586,12 @@ EffectsRD::EffectsRD() { cube_to_dp.shader.initialize(copy_modes); cube_to_dp.shader_version = cube_to_dp.shader.version_create(); - - cube_to_dp.pipeline = RD::get_singleton()->compute_pipeline_create(cube_to_dp.shader.version_get_shader(cube_to_dp.shader_version, 0)); + RID shader = cube_to_dp.shader.version_get_shader(cube_to_dp.shader_version, 0); + RD::PipelineDepthStencilState dss; + dss.enable_depth_test = true; + dss.depth_compare_operator = RD::COMPARE_OP_ALWAYS; + dss.enable_depth_write = true; + cube_to_dp.pipeline.setup(shader, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), dss, RD::PipelineColorBlendState(), 0); } { @@ -1776,7 +1688,7 @@ EffectsRD::EffectsRD() { } } - RD::get_singleton()->buffer_update(ssao.gather_constants_buffer, 0, sizeof(SSAOGatherConstants), &gather_constants, false); + RD::get_singleton()->buffer_update(ssao.gather_constants_buffer, 0, sizeof(SSAOGatherConstants), &gather_constants); } { Vector<String> ssao_modes; @@ -1795,7 +1707,8 @@ EffectsRD::EffectsRD() { } ssao.importance_map_load_counter = RD::get_singleton()->storage_buffer_create(sizeof(uint32_t)); int zero[1] = { 0 }; - RD::get_singleton()->buffer_update(ssao.importance_map_load_counter, 0, sizeof(uint32_t), &zero, false); + RD::get_singleton()->buffer_update(ssao.importance_map_load_counter, 0, sizeof(uint32_t), &zero); + RD::get_singleton()->set_resource_name(ssao.importance_map_load_counter, "Importance Map Load Counter"); Vector<RD::Uniform> uniforms; { @@ -1806,6 +1719,7 @@ EffectsRD::EffectsRD() { uniforms.push_back(u); } ssao.counter_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, ssao.importance_map_shader.version_get_shader(ssao.importance_map_shader_version, 2), 2); + RD::get_singleton()->set_resource_name(ssao.counter_uniform_set, "Load Counter Uniform Set"); } { Vector<String> ssao_modes; @@ -1834,7 +1748,7 @@ EffectsRD::EffectsRD() { ssao.interleave_shader_version = ssao.interleave_shader.version_create(); for (int i = SSAO_INTERLEAVE; i <= SSAO_INTERLEAVE_HALF; i++) { ssao.pipelines[pipeline] = RD::get_singleton()->compute_pipeline_create(ssao.interleave_shader.version_get_shader(ssao.interleave_shader_version, i - SSAO_INTERLEAVE)); - + RD::get_singleton()->set_resource_name(ssao.pipelines[pipeline], "Interleave Pipeline " + itos(i)); pipeline++; } } @@ -1867,7 +1781,7 @@ EffectsRD::EffectsRD() { { // Initialize cubemap filter - filter.use_high_quality = GLOBAL_GET("rendering/quality/reflections/fast_filter_high_quality"); + filter.use_high_quality = GLOBAL_GET("rendering/reflections/sky_reflections/fast_filter_high_quality"); Vector<String> cubemap_filter_modes; cubemap_filter_modes.push_back("\n#define USE_HIGH_QUALITY\n"); @@ -1883,10 +1797,10 @@ EffectsRD::EffectsRD() { if (filter.use_high_quality) { filter.coefficient_buffer = RD::get_singleton()->storage_buffer_create(sizeof(high_quality_coeffs)); - RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(high_quality_coeffs), &high_quality_coeffs[0], false); + RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(high_quality_coeffs), &high_quality_coeffs[0]); } else { filter.coefficient_buffer = RD::get_singleton()->storage_buffer_create(sizeof(low_quality_coeffs)); - RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(low_quality_coeffs), &low_quality_coeffs[0], false); + RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(low_quality_coeffs), &low_quality_coeffs[0]); } Vector<RD::Uniform> uniforms; @@ -2005,20 +1919,6 @@ EffectsRD::EffectsRD() { } { - Vector<String> shadow_reduce_modes; - shadow_reduce_modes.push_back("\n#define MODE_REDUCE\n"); - shadow_reduce_modes.push_back("\n#define MODE_FILTER\n"); - - shadow_reduce.shader.initialize(shadow_reduce_modes); - - shadow_reduce.shader_version = shadow_reduce.shader.version_create(); - - for (int i = 0; i < SHADOW_REDUCE_MAX; i++) { - shadow_reduce.pipelines[i] = RD::get_singleton()->compute_pipeline_create(shadow_reduce.shader.version_get_shader(shadow_reduce.shader_version, i)); - } - } - - { Vector<String> sort_modes; sort_modes.push_back("\n#define MODE_SORT_BLOCK\n"); sort_modes.push_back("\n#define MODE_SORT_STEP\n"); @@ -2039,12 +1939,14 @@ EffectsRD::EffectsRD() { sampler.max_lod = 0; default_sampler = RD::get_singleton()->sampler_create(sampler); + RD::get_singleton()->set_resource_name(default_sampler, "Default Linear Sampler"); sampler.min_filter = RD::SAMPLER_FILTER_LINEAR; sampler.mip_filter = RD::SAMPLER_FILTER_LINEAR; sampler.max_lod = 1e20; default_mipmap_sampler = RD::get_singleton()->sampler_create(sampler); + RD::get_singleton()->set_resource_name(default_mipmap_sampler, "Default MipMap Sampler"); { //create index array for copy shaders Vector<uint8_t> pv; @@ -2104,5 +2006,4 @@ EffectsRD::~EffectsRD() { ssr_scale.shader.version_free(ssr_scale.shader_version); sss.shader.version_free(sss.shader_version); tonemap.shader.version_free(tonemap.shader_version); - shadow_reduce.shader.version_free(shadow_reduce.shader_version); } diff --git a/servers/rendering/renderer_rd/effects_rd.h b/servers/rendering/renderer_rd/effects_rd.h index e2cdd0c3d8..1ba25e301b 100644 --- a/servers/rendering/renderer_rd/effects_rd.h +++ b/servers/rendering/renderer_rd/effects_rd.h @@ -46,7 +46,6 @@ #include "servers/rendering/renderer_rd/shaders/screen_space_reflection.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/screen_space_reflection_filter.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/screen_space_reflection_scale.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/shadow_reduce.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/sort.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/specular_merge.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/ssao.glsl.gen.h" @@ -234,18 +233,17 @@ class EffectsRD { } luminance_reduce; struct CopyToDPPushConstant { - int32_t screen_size[2]; - int32_t dest_offset[2]; - float bias; float z_far; float z_near; uint32_t z_flip; + uint32_t pad; + float screen_rect[4]; }; struct CoptToDP { CubeToDpShaderRD shader; RID shader_version; - RID pipeline; + PipelineCacheRD pipeline; } cube_to_dp; struct BokehPushConstant { @@ -598,18 +596,6 @@ class EffectsRD { RID pipelines[RESOLVE_MODE_MAX]; //3 quality levels } resolve; - enum ShadowReduceMode { - SHADOW_REDUCE_REDUCE, - SHADOW_REDUCE_FILTER, - SHADOW_REDUCE_MAX - }; - - struct ShadowReduce { - ShadowReduceShaderRD shader; - RID shader_version; - RID pipelines[SHADOW_REDUCE_MAX]; - } shadow_reduce; - enum SortMode { SORT_MODE_BLOCK, SORT_MODE_STEP, @@ -687,7 +673,7 @@ public: void cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size); void make_mipmap(RID p_source_rd_texture, RID p_dest_texture, const Size2i &p_size); - void copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dest_texture, const Rect2i &p_rect, float p_z_near, float p_z_far, float p_bias, bool p_dp_flip); + void copy_cubemap_to_dp(RID p_source_rd_texture, RID p_dest_texture, const Rect2 &p_rect, float p_z_near, float p_z_far, bool p_dp_flip); void luminance_reduction(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set = false); void bokeh_dof(RID p_base_texture, RID p_depth_texture, const Size2i &p_base_texture_size, RID p_secondary_texture, RID p_bokeh_texture1, RID p_bokeh_texture2, bool p_dof_far, float p_dof_far_begin, float p_dof_far_size, bool p_dof_near, float p_dof_near_begin, float p_dof_near_size, float p_bokeh_size, RS::DOFBokehShape p_bokeh_shape, RS::DOFBlurQuality p_quality, bool p_use_jitter, float p_cam_znear, float p_cam_zfar, bool p_cam_orthogonal); @@ -764,10 +750,7 @@ public: void merge_specular(RID p_dest_framebuffer, RID p_specular, RID p_base, RID p_reflection); void sub_surface_scattering(RID p_diffuse, RID p_diffuse2, RID p_depth, const CameraMatrix &p_camera, const Size2i &p_screen_size, float p_scale, float p_depth_scale, RS::SubSurfaceScatteringQuality p_quality); - void resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_giprobe, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_giprobe, Vector2i p_screen_size, int p_samples); - - void reduce_shadow(RID p_source_shadow, RID p_dest_shadow, const Size2i &p_source_size, const Rect2i &p_source_rect, int p_shrink_limit, RenderingDevice::ComputeListID compute_list); - void filter_shadow(RID p_shadow, RID p_backing_shadow, const Size2i &p_source_size, const Rect2i &p_source_rect, RS::EnvVolumetricFogShadowFilter p_filter, RenderingDevice::ComputeListID compute_list, bool p_vertical = true, bool p_horizontal = true); + void resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_giprobe, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_giprobe, Vector2i p_screen_size, int p_samples, uint32_t p_barrier = RD::BARRIER_MASK_ALL); void sort_buffer(RID p_uniform_set, int p_size); diff --git a/servers/rendering/renderer_rd/forward_clustered/SCsub b/servers/rendering/renderer_rd/forward_clustered/SCsub new file mode 100644 index 0000000000..86681f9c74 --- /dev/null +++ b/servers/rendering/renderer_rd/forward_clustered/SCsub @@ -0,0 +1,5 @@ +#!/usr/bin/env python + +Import("env") + +env.add_source_files(env.servers_sources, "*.cpp") diff --git a/servers/rendering/renderer_rd/renderer_scene_render_forward.cpp b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp index c0939f23ef..cdff3139eb 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_forward.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -1,5 +1,5 @@ /*************************************************************************/ -/* renderer_scene_render_forward.cpp */ +/* render_forward_clustered.cpp */ /*************************************************************************/ /* This file is part of: */ /* GODOT ENGINE */ @@ -28,502 +28,18 @@ /* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ /*************************************************************************/ -#include "renderer_scene_render_forward.h" +#include "render_forward_clustered.h" #include "core/config/project_settings.h" #include "servers/rendering/rendering_device.h" #include "servers/rendering/rendering_server_default.h" -/* SCENE SHADER */ -void RendererSceneRenderForward::ShaderData::set_code(const String &p_code) { - //compile +using namespace RendererSceneRenderImplementation; - code = p_code; - valid = false; - ubo_size = 0; - uniforms.clear(); - uses_screen_texture = false; - - if (code == String()) { - return; //just invalid, but no error - } - - ShaderCompilerRD::GeneratedCode gen_code; - - int blend_mode = BLEND_MODE_MIX; - int depth_testi = DEPTH_TEST_ENABLED; - int alpha_antialiasing_mode = ALPHA_ANTIALIASING_OFF; - int cull = CULL_BACK; - - uses_point_size = false; - uses_alpha = false; - uses_blend_alpha = false; - uses_depth_pre_pass = false; - uses_discard = false; - uses_roughness = false; - uses_normal = false; - bool wireframe = false; - - unshaded = false; - uses_vertex = false; - uses_sss = false; - uses_transmittance = false; - uses_screen_texture = false; - uses_depth_texture = false; - uses_normal_texture = false; - uses_time = false; - writes_modelview_or_projection = false; - uses_world_coordinates = false; - - int depth_drawi = DEPTH_DRAW_OPAQUE; - - ShaderCompilerRD::IdentifierActions actions; - - actions.render_mode_values["blend_add"] = Pair<int *, int>(&blend_mode, BLEND_MODE_ADD); - actions.render_mode_values["blend_mix"] = Pair<int *, int>(&blend_mode, BLEND_MODE_MIX); - actions.render_mode_values["blend_sub"] = Pair<int *, int>(&blend_mode, BLEND_MODE_SUB); - actions.render_mode_values["blend_mul"] = Pair<int *, int>(&blend_mode, BLEND_MODE_MUL); - - actions.render_mode_values["alpha_to_coverage"] = Pair<int *, int>(&alpha_antialiasing_mode, ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE); - actions.render_mode_values["alpha_to_coverage_and_one"] = Pair<int *, int>(&alpha_antialiasing_mode, ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE); - - actions.render_mode_values["depth_draw_never"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_DISABLED); - actions.render_mode_values["depth_draw_opaque"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_OPAQUE); - actions.render_mode_values["depth_draw_always"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_ALWAYS); - - actions.render_mode_values["depth_test_disabled"] = Pair<int *, int>(&depth_testi, DEPTH_TEST_DISABLED); - - actions.render_mode_values["cull_disabled"] = Pair<int *, int>(&cull, CULL_DISABLED); - actions.render_mode_values["cull_front"] = Pair<int *, int>(&cull, CULL_FRONT); - actions.render_mode_values["cull_back"] = Pair<int *, int>(&cull, CULL_BACK); - - actions.render_mode_flags["unshaded"] = &unshaded; - actions.render_mode_flags["wireframe"] = &wireframe; - - actions.usage_flag_pointers["ALPHA"] = &uses_alpha; - actions.render_mode_flags["depth_prepass_alpha"] = &uses_depth_pre_pass; - - actions.usage_flag_pointers["SSS_STRENGTH"] = &uses_sss; - actions.usage_flag_pointers["SSS_TRANSMITTANCE_DEPTH"] = &uses_transmittance; - - actions.usage_flag_pointers["SCREEN_TEXTURE"] = &uses_screen_texture; - actions.usage_flag_pointers["DEPTH_TEXTURE"] = &uses_depth_texture; - actions.usage_flag_pointers["NORMAL_TEXTURE"] = &uses_normal_texture; - actions.usage_flag_pointers["DISCARD"] = &uses_discard; - actions.usage_flag_pointers["TIME"] = &uses_time; - actions.usage_flag_pointers["ROUGHNESS"] = &uses_roughness; - actions.usage_flag_pointers["NORMAL"] = &uses_normal; - actions.usage_flag_pointers["NORMAL_MAP"] = &uses_normal; - - actions.usage_flag_pointers["POINT_SIZE"] = &uses_point_size; - actions.usage_flag_pointers["POINT_COORD"] = &uses_point_size; - - actions.write_flag_pointers["MODELVIEW_MATRIX"] = &writes_modelview_or_projection; - actions.write_flag_pointers["PROJECTION_MATRIX"] = &writes_modelview_or_projection; - actions.write_flag_pointers["VERTEX"] = &uses_vertex; - - actions.uniforms = &uniforms; - - RendererSceneRenderForward *scene_singleton = (RendererSceneRenderForward *)RendererSceneRenderForward::singleton; - - Error err = scene_singleton->shader.compiler.compile(RS::SHADER_SPATIAL, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); - - if (version.is_null()) { - version = scene_singleton->shader.scene_shader.version_create(); - } - - depth_draw = DepthDraw(depth_drawi); - depth_test = DepthTest(depth_testi); - -#if 0 - print_line("**compiling shader:"); - print_line("**defines:\n"); - for (int i = 0; i < gen_code.defines.size(); i++) { - print_line(gen_code.defines[i]); - } - print_line("\n**uniforms:\n" + gen_code.uniforms); - print_line("\n**vertex_globals:\n" + gen_code.vertex_global); - print_line("\n**vertex_code:\n" + gen_code.vertex); - print_line("\n**fragment_globals:\n" + gen_code.fragment_global); - print_line("\n**fragment_code:\n" + gen_code.fragment); - print_line("\n**light_code:\n" + gen_code.light); -#endif - scene_singleton->shader.scene_shader.version_set_code(version, gen_code.uniforms, gen_code.vertex_global, gen_code.vertex, gen_code.fragment_global, gen_code.light, gen_code.fragment, gen_code.defines); - ERR_FAIL_COND(!scene_singleton->shader.scene_shader.version_is_valid(version)); - - ubo_size = gen_code.uniform_total_size; - ubo_offsets = gen_code.uniform_offsets; - texture_uniforms = gen_code.texture_uniforms; - - //blend modes - - // if any form of Alpha Antialiasing is enabled, set the blend mode to alpha to coverage - if (alpha_antialiasing_mode != ALPHA_ANTIALIASING_OFF) { - blend_mode = BLEND_MODE_ALPHA_TO_COVERAGE; - } - - RD::PipelineColorBlendState::Attachment blend_attachment; - - switch (blend_mode) { - case BLEND_MODE_MIX: { - blend_attachment.enable_blend = true; - blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; - blend_attachment.color_blend_op = RD::BLEND_OP_ADD; - blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; - blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_ONE; - blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; - - } break; - case BLEND_MODE_ADD: { - blend_attachment.enable_blend = true; - blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; - blend_attachment.color_blend_op = RD::BLEND_OP_ADD; - blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE; - blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE; - uses_blend_alpha = true; //force alpha used because of blend - - } break; - case BLEND_MODE_SUB: { - blend_attachment.enable_blend = true; - blend_attachment.alpha_blend_op = RD::BLEND_OP_SUBTRACT; - blend_attachment.color_blend_op = RD::BLEND_OP_SUBTRACT; - blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE; - blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE; - uses_blend_alpha = true; //force alpha used because of blend - - } break; - case BLEND_MODE_MUL: { - blend_attachment.enable_blend = true; - blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; - blend_attachment.color_blend_op = RD::BLEND_OP_ADD; - blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_DST_COLOR; - blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ZERO; - blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_DST_ALPHA; - blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ZERO; - uses_blend_alpha = true; //force alpha used because of blend - } break; - case BLEND_MODE_ALPHA_TO_COVERAGE: { - blend_attachment.enable_blend = true; - blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; - blend_attachment.color_blend_op = RD::BLEND_OP_ADD; - blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; - blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; - blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_ONE; - blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ZERO; - } - } - - RD::PipelineColorBlendState blend_state_blend; - blend_state_blend.attachments.push_back(blend_attachment); - RD::PipelineColorBlendState blend_state_opaque = RD::PipelineColorBlendState::create_disabled(1); - RD::PipelineColorBlendState blend_state_opaque_specular = RD::PipelineColorBlendState::create_disabled(2); - RD::PipelineColorBlendState blend_state_depth_normal_roughness = RD::PipelineColorBlendState::create_disabled(1); - RD::PipelineColorBlendState blend_state_depth_normal_roughness_giprobe = RD::PipelineColorBlendState::create_disabled(2); - - //update pipelines - - RD::PipelineDepthStencilState depth_stencil_state; - - if (depth_test != DEPTH_TEST_DISABLED) { - depth_stencil_state.enable_depth_test = true; - depth_stencil_state.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; - depth_stencil_state.enable_depth_write = depth_draw != DEPTH_DRAW_DISABLED ? true : false; - } - - for (int i = 0; i < CULL_VARIANT_MAX; i++) { - RD::PolygonCullMode cull_mode_rd_table[CULL_VARIANT_MAX][3] = { - { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_FRONT, RD::POLYGON_CULL_BACK }, - { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_BACK, RD::POLYGON_CULL_FRONT }, - { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_DISABLED } - }; - - RD::PolygonCullMode cull_mode_rd = cull_mode_rd_table[i][cull]; - - for (int j = 0; j < RS::PRIMITIVE_MAX; j++) { - RD::RenderPrimitive primitive_rd_table[RS::PRIMITIVE_MAX] = { - RD::RENDER_PRIMITIVE_POINTS, - RD::RENDER_PRIMITIVE_LINES, - RD::RENDER_PRIMITIVE_LINESTRIPS, - RD::RENDER_PRIMITIVE_TRIANGLES, - RD::RENDER_PRIMITIVE_TRIANGLE_STRIPS, - }; - - RD::RenderPrimitive primitive_rd = uses_point_size ? RD::RENDER_PRIMITIVE_POINTS : primitive_rd_table[j]; - - for (int k = 0; k < SHADER_VERSION_MAX; k++) { - if (!static_cast<RendererSceneRenderForward *>(singleton)->shader.scene_shader.is_variant_enabled(k)) { - continue; - } - RD::PipelineRasterizationState raster_state; - raster_state.cull_mode = cull_mode_rd; - raster_state.wireframe = wireframe; - - RD::PipelineColorBlendState blend_state; - RD::PipelineDepthStencilState depth_stencil = depth_stencil_state; - RD::PipelineMultisampleState multisample_state; - - if (uses_alpha || uses_blend_alpha) { - // only allow these flags to go through if we have some form of msaa - if (alpha_antialiasing_mode == ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE) { - multisample_state.enable_alpha_to_coverage = true; - } else if (alpha_antialiasing_mode == ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE) { - multisample_state.enable_alpha_to_coverage = true; - multisample_state.enable_alpha_to_one = true; - } - - if (k == SHADER_VERSION_COLOR_PASS || k == SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI || k == SHADER_VERSION_LIGHTMAP_COLOR_PASS) { - blend_state = blend_state_blend; - if (depth_draw == DEPTH_DRAW_OPAQUE) { - depth_stencil.enable_depth_write = false; //alpha does not draw depth - } - } else if (uses_depth_pre_pass && (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP || k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS || k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL)) { - if (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP) { - //none, blend state contains nothing - } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL) { - blend_state = RD::PipelineColorBlendState::create_disabled(5); //writes to normal and roughness in opaque way - } else { - blend_state = blend_state_opaque; //writes to normal and roughness in opaque way - } - } else { - pipelines[i][j][k].clear(); - continue; // do not use this version (will error if using it is attempted) - } - } else { - if (k == SHADER_VERSION_COLOR_PASS || k == SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI || k == SHADER_VERSION_LIGHTMAP_COLOR_PASS) { - blend_state = blend_state_opaque; - } else if (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP) { - //none, leave empty - } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS) { - blend_state = blend_state_depth_normal_roughness; - } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE) { - blend_state = blend_state_depth_normal_roughness_giprobe; - } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL) { - blend_state = RD::PipelineColorBlendState::create_disabled(5); //writes to normal and roughness in opaque way - } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_SDF) { - blend_state = RD::PipelineColorBlendState(); //no color targets for SDF - } else { - //specular write - blend_state = blend_state_opaque_specular; - depth_stencil.enable_depth_test = false; - depth_stencil.enable_depth_write = false; - } - } - - RID shader_variant = scene_singleton->shader.scene_shader.version_get_shader(version, k); - pipelines[i][j][k].setup(shader_variant, primitive_rd, raster_state, multisample_state, depth_stencil, blend_state, 0); - } - } - } - - valid = true; -} - -void RendererSceneRenderForward::ShaderData::set_default_texture_param(const StringName &p_name, RID p_texture) { - if (!p_texture.is_valid()) { - default_texture_params.erase(p_name); - } else { - default_texture_params[p_name] = p_texture; - } -} - -void RendererSceneRenderForward::ShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { - Map<int, StringName> order; - - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { - continue; - } - - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); - } else { - order[E->get().order] = E->key(); - } - } - - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); - p_param_list->push_back(pi); - } -} - -void RendererSceneRenderForward::ShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { - continue; - } - - RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); - p_param_list->push_back(p); - } -} - -bool RendererSceneRenderForward::ShaderData::is_param_texture(const StringName &p_param) const { - if (!uniforms.has(p_param)) { - return false; - } - - return uniforms[p_param].texture_order >= 0; -} - -bool RendererSceneRenderForward::ShaderData::is_animated() const { - return false; -} - -bool RendererSceneRenderForward::ShaderData::casts_shadows() const { - return false; -} - -Variant RendererSceneRenderForward::ShaderData::get_default_parameter(const StringName &p_parameter) const { - if (uniforms.has(p_parameter)) { - ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; - Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); - } - return Variant(); -} - -RendererSceneRenderForward::ShaderData::ShaderData() { - valid = false; - uses_screen_texture = false; -} - -RendererSceneRenderForward::ShaderData::~ShaderData() { - RendererSceneRenderForward *scene_singleton = (RendererSceneRenderForward *)RendererSceneRenderForward::singleton; - ERR_FAIL_COND(!scene_singleton); - //pipeline variants will clear themselves if shader is gone - if (version.is_valid()) { - scene_singleton->shader.scene_shader.version_free(version); - } -} - -RendererStorageRD::ShaderData *RendererSceneRenderForward::_create_shader_func() { - ShaderData *shader_data = memnew(ShaderData); - return shader_data; -} - -void RendererSceneRenderForward::MaterialData::set_render_priority(int p_priority) { - priority = p_priority - RS::MATERIAL_RENDER_PRIORITY_MIN; //8 bits -} - -void RendererSceneRenderForward::MaterialData::set_next_pass(RID p_pass) { - next_pass = p_pass; -} - -void RendererSceneRenderForward::MaterialData::update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty) { - RendererSceneRenderForward *scene_singleton = (RendererSceneRenderForward *)RendererSceneRenderForward::singleton; - - if ((uint32_t)ubo_data.size() != shader_data->ubo_size) { - p_uniform_dirty = true; - if (uniform_buffer.is_valid()) { - RD::get_singleton()->free(uniform_buffer); - uniform_buffer = RID(); - } - - ubo_data.resize(shader_data->ubo_size); - if (ubo_data.size()) { - uniform_buffer = RD::get_singleton()->uniform_buffer_create(ubo_data.size()); - memset(ubo_data.ptrw(), 0, ubo_data.size()); //clear - } - - //clear previous uniform set - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - uniform_set = RID(); - } - } - - //check whether buffer changed - if (p_uniform_dirty && ubo_data.size()) { - update_uniform_buffer(shader_data->uniforms, shader_data->ubo_offsets.ptr(), p_parameters, ubo_data.ptrw(), ubo_data.size(), false); - RD::get_singleton()->buffer_update(uniform_buffer, 0, ubo_data.size(), ubo_data.ptrw()); - } - - uint32_t tex_uniform_count = shader_data->texture_uniforms.size(); - - if ((uint32_t)texture_cache.size() != tex_uniform_count) { - texture_cache.resize(tex_uniform_count); - p_textures_dirty = true; - - //clear previous uniform set - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - uniform_set = RID(); - } - } - - if (p_textures_dirty && tex_uniform_count) { - update_textures(p_parameters, shader_data->default_texture_params, shader_data->texture_uniforms, texture_cache.ptrw(), true); - } - - if (shader_data->ubo_size == 0 && shader_data->texture_uniforms.size() == 0) { - // This material does not require an uniform set, so don't create it. - return; - } - - if (!p_textures_dirty && uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - //no reason to update uniform set, only UBO (or nothing) was needed to update - return; - } - - Vector<RD::Uniform> uniforms; - - { - if (shader_data->ubo_size) { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 0; - u.ids.push_back(uniform_buffer); - uniforms.push_back(u); - } - - const RID *textures = texture_cache.ptrw(); - for (uint32_t i = 0; i < tex_uniform_count; i++) { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 1 + i; - u.ids.push_back(textures[i]); - uniforms.push_back(u); - } - } - - uniform_set = RD::get_singleton()->uniform_set_create(uniforms, scene_singleton->shader.scene_shader.version_get_shader(shader_data->version, 0), MATERIAL_UNIFORM_SET); -} - -RendererSceneRenderForward::MaterialData::~MaterialData() { - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - } - - if (uniform_buffer.is_valid()) { - RD::get_singleton()->free(uniform_buffer); - } -} - -RendererStorageRD::MaterialData *RendererSceneRenderForward::_create_material_func(ShaderData *p_shader) { - MaterialData *material_data = memnew(MaterialData); - material_data->shader_data = p_shader; - material_data->last_frame = false; - //update will happen later anyway so do nothing. - return material_data; -} - -RendererSceneRenderForward::RenderBufferDataForward::~RenderBufferDataForward() { +RenderForwardClustered::RenderBufferDataForwardClustered::~RenderBufferDataForwardClustered() { clear(); } -void RendererSceneRenderForward::RenderBufferDataForward::ensure_specular() { +void RenderForwardClustered::RenderBufferDataForwardClustered::ensure_specular() { if (!specular.is_valid()) { RD::TextureFormat tf; tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; @@ -577,20 +93,7 @@ void RendererSceneRenderForward::RenderBufferDataForward::ensure_specular() { } } -void RendererSceneRenderForward::RenderBufferDataForward::ensure_gi() { - if (!reflection_buffer.is_valid()) { - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tf.width = width; - tf.height = height; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; - - reflection_buffer = RD::get_singleton()->texture_create(tf, RD::TextureView()); - ambient_buffer = RD::get_singleton()->texture_create(tf, RD::TextureView()); - } -} - -void RendererSceneRenderForward::RenderBufferDataForward::ensure_giprobe() { +void RenderForwardClustered::RenderBufferDataForwardClustered::ensure_giprobe() { if (!giprobe_buffer.is_valid()) { RD::TextureFormat tf; tf.format = RD::DATA_FORMAT_R8G8_UINT; @@ -626,17 +129,7 @@ void RendererSceneRenderForward::RenderBufferDataForward::ensure_giprobe() { } } -void RendererSceneRenderForward::RenderBufferDataForward::clear() { - if (ambient_buffer != RID() && ambient_buffer != color) { - RD::get_singleton()->free(ambient_buffer); - ambient_buffer = RID(); - } - - if (reflection_buffer != RID() && reflection_buffer != specular) { - RD::get_singleton()->free(reflection_buffer); - reflection_buffer = RID(); - } - +void RenderForwardClustered::RenderBufferDataForwardClustered::clear() { if (giprobe_buffer != RID()) { RD::get_singleton()->free(giprobe_buffer); giprobe_buffer = RID(); @@ -690,7 +183,7 @@ void RendererSceneRenderForward::RenderBufferDataForward::clear() { } } -void RendererSceneRenderForward::RenderBufferDataForward::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa) { +void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa) { clear(); msaa = p_msaa; @@ -757,7 +250,7 @@ void RendererSceneRenderForward::RenderBufferDataForward::configure(RID p_color_ } } -void RendererSceneRenderForward::_allocate_normal_roughness_texture(RenderBufferDataForward *rb) { +void RenderForwardClustered::_allocate_normal_roughness_texture(RenderBufferDataForwardClustered *rb) { if (rb->normal_roughness_buffer.is_valid()) { return; } @@ -795,11 +288,11 @@ void RendererSceneRenderForward::_allocate_normal_roughness_texture(RenderBuffer _render_buffers_clear_uniform_set(rb); } -RendererSceneRenderRD::RenderBufferData *RendererSceneRenderForward::_create_render_buffer_data() { - return memnew(RenderBufferDataForward); +RendererSceneRenderRD::RenderBufferData *RenderForwardClustered::_create_render_buffer_data() { + return memnew(RenderBufferDataForwardClustered); } -bool RendererSceneRenderForward::free(RID p_rid) { +bool RenderForwardClustered::free(RID p_rid) { if (RendererSceneRenderRD::free(p_rid)) { return true; } @@ -808,15 +301,15 @@ bool RendererSceneRenderForward::free(RID p_rid) { /// RENDERING /// -template <RendererSceneRenderForward::PassMode p_pass_mode> -void RendererSceneRenderForward::_render_list_template(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { +template <RenderForwardClustered::PassMode p_pass_mode> +void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { RD::DrawListID draw_list = p_draw_list; RD::FramebufferFormatID framebuffer_format = p_framebuffer_Format; //global scope bindings RD::get_singleton()->draw_list_bind_uniform_set(draw_list, render_base_uniform_set, SCENE_UNIFORM_SET); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_params->render_pass_uniform_set, RENDER_PASS_UNIFORM_SET); - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, default_vec4_xform_uniform_set, TRANSFORMS_UNIFORM_SET); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, scene_shader.default_vec4_xform_uniform_set, TRANSFORMS_UNIFORM_SET); RID prev_material_uniform_set; @@ -827,16 +320,26 @@ void RendererSceneRenderForward::_render_list_template(RenderingDevice::DrawList bool shadow_pass = (p_params->pass_mode == PASS_MODE_SHADOW) || (p_params->pass_mode == PASS_MODE_SHADOW_DP); - float old_offset[2] = { 0, 0 }; + SceneState::PushConstant push_constant; + + if (p_params->pass_mode == PASS_MODE_DEPTH_MATERIAL) { + push_constant.uv_offset = Math::make_half_float(p_params->uv_offset.y) << 16; + push_constant.uv_offset |= Math::make_half_float(p_params->uv_offset.x); + } else { + push_constant.uv_offset = 0; + } for (uint32_t i = p_from_element; i < p_to_element; i++) { const GeometryInstanceSurfaceDataCache *surf = p_params->elements[i]; + const RenderElementInfo &element_info = p_params->element_info[i]; + + push_constant.base_index = i + p_params->element_offset; RID material_uniform_set; - ShaderData *shader; + SceneShaderForwardClustered::ShaderData *shader; void *mesh_surface; - if (shadow_pass) { + if (shadow_pass || p_params->pass_mode == PASS_MODE_DEPTH) { //regular depth pass can use these too material_uniform_set = surf->material_uniform_set_shadow; shader = surf->shader_shadow; mesh_surface = surf->surface_shadow; @@ -851,67 +354,60 @@ void RendererSceneRenderForward::_render_list_template(RenderingDevice::DrawList continue; } - if (p_params->pass_mode == PASS_MODE_DEPTH_MATERIAL) { - old_offset[0] = surf->owner->push_constant.lightmap_uv_scale[0]; - old_offset[1] = surf->owner->push_constant.lightmap_uv_scale[1]; - surf->owner->push_constant.lightmap_uv_scale[0] = p_params->uv_offset.x; - surf->owner->push_constant.lightmap_uv_scale[1] = p_params->uv_offset.y; - } - //find cull variant - ShaderData::CullVariant cull_variant; + SceneShaderForwardClustered::ShaderData::CullVariant cull_variant; if (p_params->pass_mode == PASS_MODE_DEPTH_MATERIAL || p_params->pass_mode == PASS_MODE_SDF || ((p_params->pass_mode == PASS_MODE_SHADOW || p_params->pass_mode == PASS_MODE_SHADOW_DP) && surf->flags & GeometryInstanceSurfaceDataCache::FLAG_USES_DOUBLE_SIDED_SHADOWS)) { - cull_variant = ShaderData::CULL_VARIANT_DOUBLE_SIDED; + cull_variant = SceneShaderForwardClustered::ShaderData::CULL_VARIANT_DOUBLE_SIDED; } else { bool mirror = surf->owner->mirror; if (p_params->reverse_cull) { mirror = !mirror; } - cull_variant = mirror ? ShaderData::CULL_VARIANT_REVERSED : ShaderData::CULL_VARIANT_NORMAL; + cull_variant = mirror ? SceneShaderForwardClustered::ShaderData::CULL_VARIANT_REVERSED : SceneShaderForwardClustered::ShaderData::CULL_VARIANT_NORMAL; } RS::PrimitiveType primitive = surf->primitive; RID xforms_uniform_set = surf->owner->transforms_uniform_set; - ShaderVersion shader_version = SHADER_VERSION_MAX; // Assigned to silence wrong -Wmaybe-initialized. + SceneShaderForwardClustered::ShaderVersion shader_version = SceneShaderForwardClustered::SHADER_VERSION_MAX; // Assigned to silence wrong -Wmaybe-initialized. switch (p_params->pass_mode) { case PASS_MODE_COLOR: case PASS_MODE_COLOR_TRANSPARENT: { - if (surf->sort.uses_lightmap) { - shader_version = SHADER_VERSION_LIGHTMAP_COLOR_PASS; - } else if (surf->sort.uses_forward_gi) { - shader_version = SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI; + if (element_info.uses_lightmap) { + shader_version = SceneShaderForwardClustered::SHADER_VERSION_LIGHTMAP_COLOR_PASS; + } else if (element_info.uses_forward_gi) { + shader_version = SceneShaderForwardClustered::SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI; } else { - shader_version = SHADER_VERSION_COLOR_PASS; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_COLOR_PASS; } } break; case PASS_MODE_COLOR_SPECULAR: { - if (surf->sort.uses_lightmap) { - shader_version = SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR; + if (element_info.uses_lightmap) { + shader_version = SceneShaderForwardClustered::SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR; } else { - shader_version = SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR; } } break; case PASS_MODE_SHADOW: case PASS_MODE_DEPTH: { - shader_version = SHADER_VERSION_DEPTH_PASS; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS; } break; case PASS_MODE_SHADOW_DP: { - shader_version = SHADER_VERSION_DEPTH_PASS_DP; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS_DP; } break; case PASS_MODE_DEPTH_NORMAL_ROUGHNESS: { - shader_version = SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS; } break; case PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE: { - shader_version = SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE; } break; case PASS_MODE_DEPTH_MATERIAL: { - shader_version = SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL; } break; case PASS_MODE_SDF: { - shader_version = SHADER_VERSION_DEPTH_PASS_WITH_SDF; + shader_version = SceneShaderForwardClustered::SHADER_VERSION_DEPTH_PASS_WITH_SDF; } break; } @@ -930,31 +426,7 @@ void RendererSceneRenderForward::_render_list_template(RenderingDevice::DrawList storage->mesh_surface_get_vertex_arrays_and_format(mesh_surface, pipeline->get_vertex_input_mask(), vertex_array_rd, vertex_format); } - if (p_params->screen_lod_threshold > 0.0 && storage->mesh_surface_has_lod(mesh_surface)) { - //lod - Vector3 support_min = surf->owner->transformed_aabb.get_support(-p_params->lod_plane.normal); - Vector3 support_max = surf->owner->transformed_aabb.get_support(p_params->lod_plane.normal); - - float distance_min = p_params->lod_plane.distance_to(support_min); - float distance_max = p_params->lod_plane.distance_to(support_max); - - float distance = 0.0; - - if (distance_min * distance_max < 0.0) { - //crossing plane - distance = 0.0; - } else if (distance_min >= 0.0) { - distance = distance_min; - } else if (distance_max <= 0.0) { - distance = -distance_max; - } - - index_array_rd = storage->mesh_surface_get_index_array_with_lod(mesh_surface, surf->owner->lod_model_scale * surf->owner->lod_bias, distance * p_params->lod_distance_multiplier, p_params->screen_lod_threshold); - - } else { - //no lod - index_array_rd = storage->mesh_surface_get_index_array(mesh_surface); - } + index_array_rd = storage->mesh_surface_get_index_array(mesh_surface, element_info.lod_index); if (prev_vertex_array_rd != vertex_array_rd) { RD::get_singleton()->draw_list_bind_vertex_array(draw_list, vertex_array_rd); @@ -991,18 +463,15 @@ void RendererSceneRenderForward::_render_list_template(RenderingDevice::DrawList prev_material_uniform_set = material_uniform_set; } - RD::get_singleton()->draw_list_set_push_constant(draw_list, &surf->owner->push_constant, sizeof(GeometryInstanceForward::PushConstant)); - - RD::get_singleton()->draw_list_draw(draw_list, index_array_rd.is_valid(), surf->owner->instance_count); + RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(SceneState::PushConstant)); - if (p_params->pass_mode == PASS_MODE_DEPTH_MATERIAL) { - surf->owner->push_constant.lightmap_uv_scale[0] = old_offset[0]; - surf->owner->push_constant.lightmap_uv_scale[1] = old_offset[1]; - } + uint32_t instance_count = surf->owner->instance_count > 1 ? surf->owner->instance_count : element_info.repeat; + RD::get_singleton()->draw_list_draw(draw_list, index_array_rd.is_valid(), instance_count); + i += element_info.repeat - 1; //skip equal elements } } -void RendererSceneRenderForward::_render_list(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { +void RenderForwardClustered::_render_list(RenderingDevice::DrawListID p_draw_list, RenderingDevice::FramebufferFormatID p_framebuffer_Format, RenderListParameters *p_params, uint32_t p_from_element, uint32_t p_to_element) { //use template for faster performance (pass mode comparisons are inlined) switch (p_params->pass_mode) { @@ -1039,7 +508,7 @@ void RendererSceneRenderForward::_render_list(RenderingDevice::DrawListID p_draw } } -void RendererSceneRenderForward::_render_list_thread_function(uint32_t p_thread, RenderListParameters *p_params) { +void RenderForwardClustered::_render_list_thread_function(uint32_t p_thread, RenderListParameters *p_params) { uint32_t render_total = p_params->element_count; uint32_t total_threads = RendererThreadPool::singleton->thread_work_pool.get_thread_count(); uint32_t render_from = p_thread * render_total / total_threads; @@ -1047,7 +516,7 @@ void RendererSceneRenderForward::_render_list_thread_function(uint32_t p_thread, _render_list(thread_draw_lists[p_thread], p_params->framebuffer_format, p_params, render_from, render_to); } -void RendererSceneRenderForward::_render_list_with_threads(RenderListParameters *p_params, RID p_framebuffer, RD::InitialAction p_initial_color_action, RD::FinalAction p_final_color_action, RD::InitialAction p_initial_depth_action, RD::FinalAction p_final_depth_action, const Vector<Color> &p_clear_color_values, float p_clear_depth, uint32_t p_clear_stencil, const Rect2 &p_region, const Vector<RID> &p_storage_textures) { +void RenderForwardClustered::_render_list_with_threads(RenderListParameters *p_params, RID p_framebuffer, RD::InitialAction p_initial_color_action, RD::FinalAction p_final_color_action, RD::InitialAction p_initial_depth_action, RD::FinalAction p_final_depth_action, const Vector<Color> &p_clear_color_values, float p_clear_depth, uint32_t p_clear_stencil, const Rect2 &p_region, const Vector<RID> &p_storage_textures) { RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(p_framebuffer); p_params->framebuffer_format = fb_format; @@ -1055,17 +524,17 @@ void RendererSceneRenderForward::_render_list_with_threads(RenderListParameters //multi threaded thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); RD::get_singleton()->draw_list_begin_split(p_framebuffer, thread_draw_lists.size(), thread_draw_lists.ptr(), p_initial_color_action, p_final_color_action, p_initial_depth_action, p_final_depth_action, p_clear_color_values, p_clear_depth, p_clear_stencil, p_region, p_storage_textures); - RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RendererSceneRenderForward::_render_list_thread_function, p_params); - RD::get_singleton()->draw_list_end(); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardClustered::_render_list_thread_function, p_params); + RD::get_singleton()->draw_list_end(p_params->barrier); } else { //single threaded RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_framebuffer, p_initial_color_action, p_final_color_action, p_initial_depth_action, p_final_depth_action, p_clear_color_values, p_clear_depth, p_clear_stencil, p_region, p_storage_textures); _render_list(draw_list, fb_format, p_params, 0, p_params->element_count); - RD::get_singleton()->draw_list_end(); + RD::get_singleton()->draw_list_end(p_params->barrier); } } -void RendererSceneRenderForward::_setup_environment(RID p_environment, RID p_render_buffers, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_reflection_probe, bool p_no_fog, const Size2 &p_screen_pixel_size, RID p_shadow_atlas, bool p_flip_y, const Color &p_default_bg_color, float p_znear, float p_zfar, bool p_opaque_render_buffers, bool p_pancake_shadows) { +void RenderForwardClustered::_setup_environment(RID p_environment, RID p_render_buffers, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_reflection_probe, bool p_no_fog, const Size2i &p_screen_size, uint32_t p_cluster_size, uint32_t p_max_cluster_elements, RID p_shadow_atlas, bool p_flip_y, const Color &p_default_bg_color, float p_znear, float p_zfar, bool p_opaque_render_buffers, bool p_pancake_shadows, int p_index) { //CameraMatrix projection = p_cam_projection; //projection.flip_y(); // Vulkan and modern APIs use Y-Down CameraMatrix correction; @@ -1093,8 +562,18 @@ void RendererSceneRenderForward::_setup_environment(RID p_environment, RID p_ren scene_state.ubo.penumbra_shadow_samples = penumbra_shadow_samples_get(); scene_state.ubo.soft_shadow_samples = soft_shadow_samples_get(); - scene_state.ubo.screen_pixel_size[0] = p_screen_pixel_size.x; - scene_state.ubo.screen_pixel_size[1] = p_screen_pixel_size.y; + Size2 screen_pixel_size = Vector2(1.0, 1.0) / Size2(p_screen_size); + scene_state.ubo.screen_pixel_size[0] = screen_pixel_size.x; + scene_state.ubo.screen_pixel_size[1] = screen_pixel_size.y; + + scene_state.ubo.cluster_shift = get_shift_from_power_of_2(p_cluster_size); + scene_state.ubo.max_cluster_element_count_div_32 = p_max_cluster_elements / 32; + { + uint32_t cluster_screen_width = (p_screen_size.width - 1) / p_cluster_size + 1; + uint32_t cluster_screen_height = (p_screen_size.height - 1) / p_cluster_size + 1; + scene_state.ubo.cluster_type_size = cluster_screen_width * cluster_screen_height * (scene_state.ubo.max_cluster_element_count_div_32 + 32); + scene_state.ubo.cluster_width = cluster_screen_width; + } if (p_shadow_atlas.is_valid()) { Vector2 sas = shadow_atlas_get_size(p_shadow_atlas); @@ -1114,7 +593,7 @@ void RendererSceneRenderForward::_setup_environment(RID p_environment, RID p_ren scene_state.ubo.fog_enabled = false; if (p_render_buffers.is_valid()) { - RenderBufferDataForward *render_buffers = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); + RenderBufferDataForwardClustered *render_buffers = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffers); if (render_buffers->msaa != RS::VIEWPORT_MSAA_DISABLED) { scene_state.ubo.gi_upscale_for_msaa = true; } @@ -1161,7 +640,7 @@ void RendererSceneRenderForward::_setup_environment(RID p_environment, RID p_ren //vec2 tex_pixel_size = 1.0 / vec2(ivec2( (OCT_SIZE+2) * params.probe_axis_size * params.probe_axis_size, (OCT_SIZE+2) * params.probe_axis_size ) ); //vec3 probe_uv_offset = (ivec3(OCT_SIZE+2,OCT_SIZE+2,(OCT_SIZE+2) * params.probe_axis_size)) * tex_pixel_size.xyx; - uint32_t oct_size = sdfgi_get_lightprobe_octahedron_size(); + uint32_t oct_size = gi.sdfgi_get_lightprobe_octahedron_size(); scene_state.ubo.sdfgi_lightprobe_tex_pixel_size[0] = 1.0 / ((oct_size + 2) * scene_state.ubo.sdfgi_probe_axis_size * scene_state.ubo.sdfgi_probe_axis_size); scene_state.ubo.sdfgi_lightprobe_tex_pixel_size[1] = 1.0 / ((oct_size + 2) * scene_state.ubo.sdfgi_probe_axis_size); @@ -1294,27 +773,125 @@ void RendererSceneRenderForward::_setup_environment(RID p_environment, RID p_ren scene_state.ubo.roughness_limiter_amount = screen_space_roughness_limiter_get_amount(); scene_state.ubo.roughness_limiter_limit = screen_space_roughness_limiter_get_limit(); - RD::get_singleton()->buffer_update(scene_state.uniform_buffer, 0, sizeof(SceneState::UBO), &scene_state.ubo, true); + if (p_index >= (int)scene_state.uniform_buffers.size()) { + uint32_t from = scene_state.uniform_buffers.size(); + scene_state.uniform_buffers.resize(p_index + 1); + render_pass_uniform_sets.resize(p_index + 1); + for (uint32_t i = from; i < scene_state.uniform_buffers.size(); i++) { + scene_state.uniform_buffers[i] = RD::get_singleton()->uniform_buffer_create(sizeof(SceneState::UBO)); + } + } + RD::get_singleton()->buffer_update(scene_state.uniform_buffers[p_index], 0, sizeof(SceneState::UBO), &scene_state.ubo, RD::BARRIER_MASK_RASTER); +} + +void RenderForwardClustered::_update_instance_data_buffer(RenderListType p_render_list) { + if (scene_state.instance_data[p_render_list].size() > 0) { + if (scene_state.instance_buffer[p_render_list] == RID() || scene_state.instance_buffer_size[p_render_list] < scene_state.instance_data[p_render_list].size()) { + if (scene_state.instance_buffer[p_render_list] != RID()) { + RD::get_singleton()->free(scene_state.instance_buffer[p_render_list]); + } + uint32_t new_size = nearest_power_of_2_templated(MAX(uint64_t(INSTANCE_DATA_BUFFER_MIN_SIZE), scene_state.instance_data[p_render_list].size())); + scene_state.instance_buffer[p_render_list] = RD::get_singleton()->storage_buffer_create(new_size * sizeof(SceneState::InstanceData)); + scene_state.instance_buffer_size[p_render_list] = new_size; + } + RD::get_singleton()->buffer_update(scene_state.instance_buffer[p_render_list], 0, sizeof(SceneState::InstanceData) * scene_state.instance_data[p_render_list].size(), scene_state.instance_data[p_render_list].ptr(), RD::BARRIER_MASK_RASTER); + } +} +void RenderForwardClustered::_fill_instance_data(RenderListType p_render_list, uint32_t p_offset, int32_t p_max_elements, bool p_update_buffer) { + RenderList *rl = &render_list[p_render_list]; + uint32_t element_total = p_max_elements >= 0 ? uint32_t(p_max_elements) : rl->elements.size(); + + scene_state.instance_data[p_render_list].resize(p_offset + element_total); + rl->element_info.resize(p_offset + element_total); + + uint32_t repeats = 0; + GeometryInstanceSurfaceDataCache *prev_surface = nullptr; + for (uint32_t i = 0; i < element_total; i++) { + GeometryInstanceSurfaceDataCache *surface = rl->elements[i + p_offset]; + GeometryInstanceForwardClustered *inst = surface->owner; + + SceneState::InstanceData &instance_data = scene_state.instance_data[p_render_list][i + p_offset]; + + if (inst->store_transform_cache) { + RendererStorageRD::store_transform(inst->transform, instance_data.transform); + } else { + RendererStorageRD::store_transform(Transform(), instance_data.transform); + } + + instance_data.flags = inst->flags_cache; + instance_data.gi_offset = inst->gi_offset_cache; + instance_data.layer_mask = inst->layer_mask; + instance_data.instance_uniforms_ofs = uint32_t(inst->shader_parameters_offset); + instance_data.lightmap_uv_scale[0] = inst->lightmap_uv_scale.position.x; + instance_data.lightmap_uv_scale[1] = inst->lightmap_uv_scale.position.y; + instance_data.lightmap_uv_scale[2] = inst->lightmap_uv_scale.size.x; + instance_data.lightmap_uv_scale[3] = inst->lightmap_uv_scale.size.y; + + bool cant_repeat = instance_data.flags & INSTANCE_DATA_FLAG_MULTIMESH || inst->mesh_instance.is_valid(); + + if (prev_surface != nullptr && !cant_repeat && prev_surface->sort.sort_key1 == surface->sort.sort_key1 && prev_surface->sort.sort_key2 == surface->sort.sort_key2) { + //this element is the same as the previous one, count repeats to draw it using instancing + repeats++; + } else { + if (repeats > 0) { + for (uint32_t j = 1; j <= repeats; j++) { + rl->element_info[p_offset + i - j].repeat = j; + } + } + repeats = 1; + } + + RenderElementInfo &element_info = rl->element_info[p_offset + i]; + + element_info.lod_index = surface->sort.lod_index; + element_info.uses_forward_gi = surface->sort.uses_forward_gi; + element_info.uses_lightmap = surface->sort.uses_lightmap; + + if (cant_repeat) { + prev_surface = nullptr; + } else { + prev_surface = surface; + } + } + + if (repeats > 0) { + for (uint32_t j = 1; j <= repeats; j++) { + rl->element_info[p_offset + element_total - j].repeat = j; + } + } + + if (p_update_buffer) { + _update_instance_data_buffer(p_render_list); + } } -void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInstance *> &p_instances, PassMode p_pass_mode, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, bool p_using_sdfgi, bool p_using_opaque_gi) { - scene_state.used_sss = false; - scene_state.used_screen_texture = false; - scene_state.used_normal_texture = false; - scene_state.used_depth_texture = false; +void RenderForwardClustered::_fill_render_list(RenderListType p_render_list, const PagedArray<GeometryInstance *> &p_instances, PassMode p_pass_mode, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, bool p_using_sdfgi, bool p_using_opaque_gi, const Plane &p_lod_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold, bool p_append) { + if (p_render_list == RENDER_LIST_OPAQUE) { + scene_state.used_sss = false; + scene_state.used_screen_texture = false; + scene_state.used_normal_texture = false; + scene_state.used_depth_texture = false; + } + uint32_t lightmap_captures_used = 0; Plane near_plane(p_cam_transform.origin, -p_cam_transform.basis.get_axis(Vector3::AXIS_Z)); near_plane.d += p_cam_projection.get_z_near(); float z_max = p_cam_projection.get_z_far() - p_cam_projection.get_z_near(); - uint32_t lightmap_captures_used = 0; + RenderList *rl = &render_list[p_render_list]; _update_dirty_geometry_instances(); - render_list.clear(); + + if (!p_append) { + rl->clear(); + if (p_render_list == RENDER_LIST_OPAQUE) { + render_list[RENDER_LIST_ALPHA].clear(); //opaque fills alpha too + } + } //fill list for (int i = 0; i < (int)p_instances.size(); i++) { - GeometryInstanceForward *inst = static_cast<GeometryInstanceForward *>(p_instances[i]); + GeometryInstanceForwardClustered *inst = static_cast<GeometryInstanceForwardClustered *>(p_instances[i]); Vector3 support_min = inst->transformed_aabb.get_support(-near_plane.normal); inst->depth = near_plane.distance_to(support_min); @@ -1325,7 +902,7 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst bool uses_lightmap = false; bool uses_gi = false; - if (p_pass_mode == PASS_MODE_COLOR) { + if (p_render_list == RENDER_LIST_OPAQUE) { //setup GI if (inst->lightmap_instance.is_valid()) { @@ -1337,15 +914,15 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst } } if (lightmap_cull_index >= 0) { - inst->push_constant.gi_offset &= 0xFFFF; - inst->push_constant.gi_offset |= lightmap_cull_index; + inst->gi_offset_cache = inst->lightmap_slice_index << 16; + inst->gi_offset_cache |= lightmap_cull_index; flags |= INSTANCE_DATA_FLAG_USE_LIGHTMAP; if (scene_state.lightmap_has_sh[lightmap_cull_index]) { flags |= INSTANCE_DATA_FLAG_USE_SH_LIGHTMAP; } uses_lightmap = true; } else { - inst->push_constant.gi_offset = 0xFFFFFFFF; + inst->gi_offset_cache = 0xFFFFFFFF; } } else if (inst->lightmap_sh) { @@ -1359,7 +936,7 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst lcd.sh[j * 4 + 3] = src_capture[j].a; } flags |= INSTANCE_DATA_FLAG_USE_LIGHTMAP_CAPTURE; - inst->push_constant.gi_offset = lightmap_captures_used; + inst->gi_offset_cache = lightmap_captures_used; lightmap_captures_used++; uses_lightmap = true; } @@ -1386,18 +963,19 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst SWAP(probe0_index, probe1_index); } - inst->push_constant.gi_offset = probe0_index | (probe1_index << 16); + inst->gi_offset_cache = probe0_index | (probe1_index << 16); + flags |= INSTANCE_DATA_FLAG_USE_GIPROBE; uses_gi = true; } else { if (p_using_sdfgi && inst->can_sdfgi) { flags |= INSTANCE_DATA_FLAG_USE_SDFGI; uses_gi = true; } - inst->push_constant.gi_offset = 0xFFFFFFFF; + inst->gi_offset_cache = 0xFFFFFFFF; } } } - inst->push_constant.flags = flags; + inst->flags_cache = flags; GeometryInstanceSurfaceDataCache *surf = inst->surface_caches; @@ -1405,12 +983,39 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst surf->sort.uses_forward_gi = 0; surf->sort.uses_lightmap = 0; + // LOD + + if (p_screen_lod_threshold > 0.0 && storage->mesh_surface_has_lod(surf->surface)) { + //lod + Vector3 lod_support_min = inst->transformed_aabb.get_support(-p_lod_plane.normal); + Vector3 lod_support_max = inst->transformed_aabb.get_support(p_lod_plane.normal); + + float distance_min = p_lod_plane.distance_to(lod_support_min); + float distance_max = p_lod_plane.distance_to(lod_support_max); + + float distance = 0.0; + + if (distance_min * distance_max < 0.0) { + //crossing plane + distance = 0.0; + } else if (distance_min >= 0.0) { + distance = distance_min; + } else if (distance_max <= 0.0) { + distance = -distance_max; + } + + surf->sort.lod_index = storage->mesh_surface_get_lod(surf->surface, inst->lod_model_scale * inst->lod_bias, distance * p_lod_distance_multiplier, p_screen_lod_threshold); + } else { + surf->sort.lod_index = 0; + } + + // ADD Element if (p_pass_mode == PASS_MODE_COLOR) { if (surf->flags & (GeometryInstanceSurfaceDataCache::FLAG_PASS_DEPTH | GeometryInstanceSurfaceDataCache::FLAG_PASS_OPAQUE)) { - render_list.add_element(surf); + rl->add_element(surf); } if (surf->flags & GeometryInstanceSurfaceDataCache::FLAG_PASS_ALPHA) { - render_list.add_alpha_element(surf); + render_list[RENDER_LIST_ALPHA].add_element(surf); if (uses_gi) { surf->sort.uses_forward_gi = 1; } @@ -1435,11 +1040,11 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst } else if (p_pass_mode == PASS_MODE_SHADOW || p_pass_mode == PASS_MODE_SHADOW_DP) { if (surf->flags & GeometryInstanceSurfaceDataCache::FLAG_PASS_SHADOW) { - render_list.add_element(surf); + rl->add_element(surf); } } else { if (surf->flags & (GeometryInstanceSurfaceDataCache::FLAG_PASS_DEPTH | GeometryInstanceSurfaceDataCache::FLAG_PASS_OPAQUE)) { - render_list.add_element(surf); + rl->add_element(surf); } } @@ -1449,19 +1054,19 @@ void RendererSceneRenderForward::_fill_render_list(const PagedArray<GeometryInst } } - if (lightmap_captures_used) { - RD::get_singleton()->buffer_update(scene_state.lightmap_capture_buffer, 0, sizeof(LightmapCaptureData) * lightmap_captures_used, scene_state.lightmap_captures, true); + if (p_render_list == RENDER_LIST_OPAQUE && lightmap_captures_used) { + RD::get_singleton()->buffer_update(scene_state.lightmap_capture_buffer, 0, sizeof(LightmapCaptureData) * lightmap_captures_used, scene_state.lightmap_captures, RD::BARRIER_MASK_RASTER); } } -void RendererSceneRenderForward::_setup_giprobes(const PagedArray<RID> &p_giprobes) { +void RenderForwardClustered::_setup_giprobes(const PagedArray<RID> &p_giprobes) { scene_state.giprobes_used = MIN(p_giprobes.size(), uint32_t(MAX_GI_PROBES)); for (uint32_t i = 0; i < scene_state.giprobes_used; i++) { scene_state.giprobe_ids[i] = p_giprobes[i]; } } -void RendererSceneRenderForward::_setup_lightmaps(const PagedArray<RID> &p_lightmaps, const Transform &p_cam_transform) { +void RenderForwardClustered::_setup_lightmaps(const PagedArray<RID> &p_lightmaps, const Transform &p_cam_transform) { scene_state.lightmaps_used = 0; for (int i = 0; i < (int)p_lightmaps.size(); i++) { if (i >= (int)scene_state.max_lightmaps) { @@ -1479,29 +1084,22 @@ void RendererSceneRenderForward::_setup_lightmaps(const PagedArray<RID> &p_light scene_state.lightmaps_used++; } if (scene_state.lightmaps_used > 0) { - RD::get_singleton()->buffer_update(scene_state.lightmap_buffer, 0, sizeof(LightmapData) * scene_state.lightmaps_used, scene_state.lightmaps, true); + RD::get_singleton()->buffer_update(scene_state.lightmap_buffer, 0, sizeof(LightmapData) * scene_state.lightmaps_used, scene_state.lightmaps, RD::BARRIER_MASK_RASTER); } } -void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, int p_directional_light_count, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_bg_color, float p_screen_lod_threshold) { - RenderBufferDataForward *render_buffer = nullptr; +void RenderForwardClustered::_render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_cluster_buffer, uint32_t p_cluster_size, uint32_t p_max_cluster_elements, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_bg_color, float p_screen_lod_threshold) { + RenderBufferDataForwardClustered *render_buffer = nullptr; if (p_render_buffer.is_valid()) { - render_buffer = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffer); + render_buffer = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffer); } + RendererSceneEnvironmentRD *env = get_environment(p_environment); //first of all, make a new render pass - render_pass++; - //fill up ubo RENDER_TIMESTAMP("Setup 3D Scene"); - if (p_reflection_probe.is_valid()) { - scene_state.ubo.reflection_multiplier = 0.0; - } else { - scene_state.ubo.reflection_multiplier = 1.0; - } - float lod_distance_multiplier = p_cam_projection.get_lod_multiplier(); Plane lod_camera_plane(p_cam_transform.get_origin(), -p_cam_transform.basis.get_axis(Vector3::AXIS_Z)); @@ -1514,9 +1112,8 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf Vector2 vp_he = p_cam_projection.get_viewport_half_extents(); scene_state.ubo.viewport_size[0] = vp_he.x; scene_state.ubo.viewport_size[1] = vp_he.y; - scene_state.ubo.directional_light_count = p_directional_light_count; + scene_state.ubo.directional_light_count = 0; - Size2 screen_pixel_size; Size2i screen_size; RID opaque_framebuffer; RID opaque_specular_framebuffer; @@ -1531,8 +1128,6 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf bool using_giprobe = false; if (render_buffer) { - screen_pixel_size.width = 1.0 / render_buffer->width; - screen_pixel_size.height = 1.0 / render_buffer->height; screen_size.x = render_buffer->width; screen_size.y = render_buffer->height; @@ -1540,7 +1135,6 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf if (!low_end && p_gi_probes.size() > 0) { using_giprobe = true; - render_buffer->ensure_gi(); } if (!p_environment.is_valid() && using_giprobe) { @@ -1550,7 +1144,6 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf if (environment_is_sdfgi_enabled(p_environment)) { depth_pass_mode = using_giprobe ? PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE : PASS_MODE_DEPTH_NORMAL_ROUGHNESS; // also giprobe using_sdfgi = true; - render_buffer->ensure_gi(); } else { depth_pass_mode = using_giprobe ? PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE : PASS_MODE_DEPTH_NORMAL_ROUGHNESS; } @@ -1589,8 +1182,6 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf alpha_framebuffer = opaque_framebuffer; } else if (p_reflection_probe.is_valid()) { uint32_t resolution = reflection_probe_instance_get_resolution(p_reflection_probe); - screen_pixel_size.width = 1.0 / resolution; - screen_pixel_size.height = 1.0 / resolution; screen_size.x = resolution; screen_size.y = resolution; @@ -1605,13 +1196,21 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf ERR_FAIL(); //bug? } + RD::get_singleton()->draw_command_begin_label("Render Setup"); + _setup_lightmaps(p_lightmaps, p_cam_transform); _setup_giprobes(p_gi_probes); - _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_pixel_size, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), false); + _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_size, p_cluster_size, p_max_cluster_elements, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), false); _update_render_base_uniform_set(); //may have changed due to the above (light buffer enlarged, as an example) - _fill_render_list(p_instances, PASS_MODE_COLOR, p_cam_projection, p_cam_transform, using_sdfgi, using_sdfgi || using_giprobe); + _fill_render_list(RENDER_LIST_OPAQUE, p_instances, PASS_MODE_COLOR, p_cam_projection, p_cam_transform, using_sdfgi, using_sdfgi || using_giprobe, lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); + render_list[RENDER_LIST_OPAQUE].sort_by_key(); + render_list[RENDER_LIST_ALPHA].sort_by_depth(); + _fill_instance_data(RENDER_LIST_OPAQUE); + _fill_instance_data(RENDER_LIST_ALPHA); + + RD::get_singleton()->draw_command_end_label(); bool using_sss = !low_end && render_buffer && scene_state.used_sss && sub_surface_scattering_get_quality() != RS::SUB_SURFACE_SCATTERING_QUALITY_DISABLED; @@ -1641,7 +1240,7 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf clear_color.b *= bg_energy; if (render_buffers_has_volumetric_fog(p_render_buffer) || environment_is_fog_enabled(p_environment)) { draw_sky_fog_only = true; - storage->material_set_param(sky_scene_state.fog_material, "clear_color", Variant(clear_color.to_linear())); + storage->material_set_param(sky.sky_scene_state.fog_material, "clear_color", Variant(clear_color.to_linear())); } } break; case RS::ENV_BG_COLOR: { @@ -1651,7 +1250,7 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf clear_color.b *= bg_energy; if (render_buffers_has_volumetric_fog(p_render_buffer) || environment_is_fog_enabled(p_environment)) { draw_sky_fog_only = true; - storage->material_set_param(sky_scene_state.fog_material, "clear_color", Variant(clear_color.to_linear())); + storage->material_set_param(sky.sky_scene_state.fog_material, "clear_color", Variant(clear_color.to_linear())); } } break; case RS::ENV_BG_SKY: { @@ -1671,6 +1270,7 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf // setup sky if used for ambient, reflections, or background if (draw_sky || draw_sky_fog_only || environment_get_reflection_source(p_environment) == RS::ENV_REFLECTION_SOURCE_SKY || environment_get_ambient_source(p_environment) == RS::ENV_AMBIENT_SOURCE_SKY) { RENDER_TIMESTAMP("Setup Sky"); + RD::get_singleton()->draw_command_begin_label("Setup Sky"); CameraMatrix projection = p_cam_projection; if (p_reflection_probe.is_valid()) { CameraMatrix correction; @@ -1678,65 +1278,87 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf projection = correction * p_cam_projection; } - _setup_sky(p_environment, p_render_buffer, projection, p_cam_transform, screen_size); + sky.setup(env, p_render_buffer, projection, p_cam_transform, screen_size, this); - RID sky = environment_get_sky(p_environment); - if (sky.is_valid()) { - _update_sky(p_environment, projection, p_cam_transform); - radiance_texture = sky_get_radiance_texture_rd(sky); + RID sky_rid = env->sky; + if (sky_rid.is_valid()) { + sky.update(env, projection, p_cam_transform, time); + radiance_texture = sky.sky_get_radiance_texture_rd(sky_rid); } else { // do not try to draw sky if invalid draw_sky = false; } + RD::get_singleton()->draw_command_end_label(); } } else { clear_color = p_default_bg_color; } - render_list.sort_by_key(false); - bool debug_giprobes = get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_ALBEDO || get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_LIGHTING || get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_EMISSION; bool debug_sdfgi_probes = get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_SDFGI_PROBES; - bool depth_pre_pass = !low_end && depth_framebuffer.is_valid(); bool using_ssao = depth_pre_pass && p_render_buffer.is_valid() && p_environment.is_valid() && environment_is_ssao_enabled(p_environment); bool continue_depth = false; if (depth_pre_pass) { //depth pre pass - RENDER_TIMESTAMP("Render Depth Pre-Pass"); - RID rp_uniform_set = _setup_render_pass_uniform_set(RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); + bool needs_pre_resolve = _needs_post_prepass_render(using_sdfgi || using_giprobe); + if (needs_pre_resolve) { + RENDER_TIMESTAMP("GI + Render Depth Pre-Pass (parallel)"); + } else { + RENDER_TIMESTAMP("Render Depth Pre-Pass"); + } + if (needs_pre_resolve) { + //pre clear the depth framebuffer, as AMD (and maybe others?) use compute for it, and barrier other compute shaders. + RD::get_singleton()->draw_list_begin(depth_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_CONTINUE, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_CONTINUE, depth_pass_clear); + RD::get_singleton()->draw_list_end(); + //start compute processes here, so they run at the same time as depth pre-pass + _post_prepass_render(using_sdfgi || using_giprobe); + } + + RD::get_singleton()->draw_command_begin_label("Render Depth Pre-Pass"); + + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, RID(), RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); bool finish_depth = using_ssao || using_sdfgi || using_giprobe; - RenderListParameters render_list_params(render_list.elements, render_list.element_count, false, depth_pass_mode, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); - _render_list_with_threads(&render_list_params, depth_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, finish_depth ? RD::FINAL_ACTION_READ : RD::FINAL_ACTION_CONTINUE, depth_pass_clear); + RenderListParameters render_list_params(render_list[RENDER_LIST_OPAQUE].elements.ptr(), render_list[RENDER_LIST_OPAQUE].element_info.ptr(), render_list[RENDER_LIST_OPAQUE].elements.size(), false, depth_pass_mode, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); + _render_list_with_threads(&render_list_params, depth_framebuffer, needs_pre_resolve ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, needs_pre_resolve ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_CLEAR, finish_depth ? RD::FINAL_ACTION_READ : RD::FINAL_ACTION_CONTINUE, needs_pre_resolve ? Vector<Color>() : depth_pass_clear); + + RD::get_singleton()->draw_command_end_label(); + + if (needs_pre_resolve) { + _pre_resolve_render(using_sdfgi || using_giprobe); + } if (render_buffer && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { RENDER_TIMESTAMP("Resolve Depth Pre-Pass"); + RD::get_singleton()->draw_command_begin_label("Resolve Depth Pre-Pass"); if (depth_pass_mode == PASS_MODE_DEPTH_NORMAL_ROUGHNESS || depth_pass_mode == PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE) { + if (needs_pre_resolve) { + RD::get_singleton()->barrier(RD::BARRIER_MASK_RASTER, RD::BARRIER_MASK_COMPUTE); + } static int texture_samples[RS::VIEWPORT_MSAA_MAX] = { 1, 2, 4, 8, 16 }; storage->get_effects()->resolve_gi(render_buffer->depth_msaa, render_buffer->normal_roughness_buffer_msaa, using_giprobe ? render_buffer->giprobe_buffer_msaa : RID(), render_buffer->depth, render_buffer->normal_roughness_buffer, using_giprobe ? render_buffer->giprobe_buffer : RID(), Vector2i(render_buffer->width, render_buffer->height), texture_samples[render_buffer->msaa]); } else if (finish_depth) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->depth_msaa, render_buffer->depth, true); + RD::get_singleton()->texture_resolve_multisample(render_buffer->depth_msaa, render_buffer->depth); } + RD::get_singleton()->draw_command_end_label(); } continue_depth = !finish_depth; } - if (using_ssao) { - _process_ssao(p_render_buffer, p_environment, render_buffer->normal_roughness_buffer, p_cam_projection); - } + _pre_opaque_render(using_ssao, using_sdfgi || using_giprobe, render_buffer ? render_buffer->normal_roughness_buffer : RID(), render_buffer ? render_buffer->giprobe_buffer : RID()); - if (using_sdfgi || using_giprobe) { - _process_gi(p_render_buffer, render_buffer->normal_roughness_buffer, render_buffer->ambient_buffer, render_buffer->reflection_buffer, render_buffer->giprobe_buffer, p_environment, p_cam_projection, p_cam_transform, p_gi_probes); - } + RD::get_singleton()->draw_command_begin_label("Render Opaque Pass"); - _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_pixel_size, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), p_render_buffer.is_valid()); + scene_state.ubo.directional_light_count = _get_render_state_directional_light_count(); + + _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_size, p_cluster_size, p_max_cluster_elements, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), p_render_buffer.is_valid()); RENDER_TIMESTAMP("Render Opaque Pass"); - RID rp_uniform_set = _setup_render_pass_uniform_set(p_render_buffer, radiance_texture, p_shadow_atlas, p_reflection_atlas, p_gi_probes, p_lightmaps); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, p_render_buffer, radiance_texture, p_shadow_atlas, p_reflection_atlas, p_cluster_buffer, p_gi_probes, p_lightmaps, true); bool can_continue_color = !scene_state.used_screen_texture && !using_ssr && !using_sss; bool can_continue_depth = !scene_state.used_depth_texture && !using_ssr && !using_sss; @@ -1757,10 +1379,8 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf } RID framebuffer = using_separate_specular ? opaque_specular_framebuffer : opaque_framebuffer; - RenderListParameters render_list_params(render_list.elements, render_list.element_count, false, using_separate_specular ? PASS_MODE_COLOR_SPECULAR : PASS_MODE_COLOR, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); - - _render_list_with_threads(&render_list_params, framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, depth_pre_pass ? (continue_depth ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CONTINUE) : RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); - + RenderListParameters render_list_params(render_list[RENDER_LIST_OPAQUE].elements.ptr(), render_list[RENDER_LIST_OPAQUE].element_info.ptr(), render_list[RENDER_LIST_OPAQUE].elements.size(), false, using_separate_specular ? PASS_MODE_COLOR_SPECULAR : PASS_MODE_COLOR, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); + _render_list_with_threads(&render_list_params, framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, depth_pre_pass ? (continue_depth ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP) : RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); if (will_continue_color && using_separate_specular) { // close the specular framebuffer, as it's no longer used RD::get_singleton()->draw_list_begin(render_buffer->specular_only_fb, RD::INITIAL_ACTION_CONTINUE, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, RD::FINAL_ACTION_CONTINUE); @@ -1768,6 +1388,8 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf } } + RD::get_singleton()->draw_command_end_label(); + if (debug_giprobes) { //debug giprobes bool will_continue_color = (can_continue_color || draw_sky || draw_sky_fog_only); @@ -1777,9 +1399,11 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf dc.set_depth_correction(true); CameraMatrix cm = (dc * p_cam_projection) * CameraMatrix(p_cam_transform.affine_inverse()); RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(opaque_framebuffer, RD::INITIAL_ACTION_CONTINUE, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ); + RD::get_singleton()->draw_command_begin_label("Debug GIProbes"); for (int i = 0; i < (int)p_gi_probes.size(); i++) { - _debug_giprobe(p_gi_probes[i], draw_list, opaque_framebuffer, cm, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_LIGHTING, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_EMISSION, 1.0); + gi.debug_giprobe(p_gi_probes[i], draw_list, opaque_framebuffer, cm, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_LIGHTING, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_GI_PROBE_EMISSION, 1.0); } + RD::get_singleton()->draw_command_end_label(); RD::get_singleton()->draw_list_end(); } @@ -1792,7 +1416,9 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf dc.set_depth_correction(true); CameraMatrix cm = (dc * p_cam_projection) * CameraMatrix(p_cam_transform.affine_inverse()); RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(opaque_framebuffer, RD::INITIAL_ACTION_CONTINUE, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ); + RD::get_singleton()->draw_command_begin_label("Debug SDFGI"); _debug_sdfgi_probes(p_render_buffer, draw_list, opaque_framebuffer, cm); + RD::get_singleton()->draw_command_end_label(); RD::get_singleton()->draw_list_end(); } @@ -1805,30 +1431,35 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf correction.set_depth_correction(true); projection = correction * p_cam_projection; } - - _draw_sky(can_continue_color, can_continue_depth, opaque_framebuffer, p_environment, projection, p_cam_transform); + RD::get_singleton()->draw_command_begin_label("Draw Sky"); + sky.draw(env, can_continue_color, can_continue_depth, opaque_framebuffer, projection, p_cam_transform, time); + RD::get_singleton()->draw_command_end_label(); } if (render_buffer && !can_continue_color && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color, true); + RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color); if (using_separate_specular) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->specular_msaa, render_buffer->specular, true); + RD::get_singleton()->texture_resolve_multisample(render_buffer->specular_msaa, render_buffer->specular); } } if (render_buffer && !can_continue_depth && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->depth_msaa, render_buffer->depth, true); + RD::get_singleton()->texture_resolve_multisample(render_buffer->depth_msaa, render_buffer->depth); } if (using_separate_specular) { if (using_sss) { RENDER_TIMESTAMP("Sub Surface Scattering"); + RD::get_singleton()->draw_command_begin_label("Process Sub Surface Scattering"); _process_sss(p_render_buffer, p_cam_projection); + RD::get_singleton()->draw_command_end_label(); } if (using_ssr) { RENDER_TIMESTAMP("Screen Space Reflection"); + RD::get_singleton()->draw_command_begin_label("Process Screen Space Reflections"); _process_ssr(p_render_buffer, render_buffer->color_fb, render_buffer->normal_roughness_buffer, render_buffer->specular, render_buffer->specular, Color(0, 0, 0, 1), p_environment, p_cam_projection, render_buffer->msaa == RS::VIEWPORT_MSAA_DISABLED); + RD::get_singleton()->draw_command_end_label(); } else { //just mix specular back RENDER_TIMESTAMP("Merge Specular"); @@ -1838,30 +1469,44 @@ void RendererSceneRenderForward::_render_scene(RID p_render_buffer, const Transf RENDER_TIMESTAMP("Render Transparent Pass"); - _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_pixel_size, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), false); + RD::get_singleton()->draw_command_begin_label("Render Transparent Pass"); + + rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_ALPHA, p_render_buffer, radiance_texture, p_shadow_atlas, p_reflection_atlas, p_cluster_buffer, p_gi_probes, p_lightmaps, true); - render_list.sort_by_reverse_depth_and_priority(true); + _setup_environment(p_environment, p_render_buffer, p_cam_projection, p_cam_transform, p_reflection_probe, p_reflection_probe.is_valid(), screen_size, p_cluster_size, p_max_cluster_elements, p_shadow_atlas, !p_reflection_probe.is_valid(), p_default_bg_color, p_cam_projection.get_z_near(), p_cam_projection.get_z_far(), false); { - RenderListParameters render_list_params(&render_list.elements[render_list.max_elements - render_list.alpha_element_count], render_list.alpha_element_count, false, PASS_MODE_COLOR, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); + RenderListParameters render_list_params(render_list[RENDER_LIST_ALPHA].elements.ptr(), render_list[RENDER_LIST_ALPHA].element_info.ptr(), render_list[RENDER_LIST_ALPHA].elements.size(), false, PASS_MODE_COLOR, render_buffer == nullptr, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), lod_camera_plane, lod_distance_multiplier, p_screen_lod_threshold); _render_list_with_threads(&render_list_params, alpha_framebuffer, can_continue_color ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, can_continue_depth ? RD::INITIAL_ACTION_CONTINUE : RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ); } + RD::get_singleton()->draw_command_end_label(); + + RD::get_singleton()->draw_command_begin_label("Resolve"); + if (render_buffer && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color, true); + RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color); } -} -void RendererSceneRenderForward::_render_shadow(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool p_use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold) { - RENDER_TIMESTAMP("Setup Rendering Shadow"); + RD::get_singleton()->draw_command_end_label(); +} +void RenderForwardClustered::_render_shadow_begin() { + scene_state.shadow_passes.clear(); + RD::get_singleton()->draw_command_begin_label("Shadow Setup"); _update_render_base_uniform_set(); - render_pass++; + render_list[RENDER_LIST_SECONDARY].clear(); + scene_state.instance_data[RENDER_LIST_SECONDARY].clear(); +} +void RenderForwardClustered::_render_shadow_append(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool p_use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold, const Rect2i &p_rect, bool p_flip_y, bool p_clear_region, bool p_begin, bool p_end) { + uint32_t shadow_pass_index = scene_state.shadow_passes.size(); + + SceneState::ShadowPass shadow_pass; scene_state.ubo.dual_paraboloid_side = p_use_dp_flip ? -1 : 1; - _setup_environment(RID(), RID(), p_projection, p_transform, RID(), true, Vector2(1, 1), RID(), true, Color(), 0, p_zfar, false, p_use_pancake); + _setup_environment(RID(), RID(), p_projection, p_transform, RID(), true, Vector2(1, 1), 1, 32, RID(), !p_flip_y, Color(), 0, p_zfar, false, p_use_pancake, shadow_pass_index); if (get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_DISABLE_LOD) { p_screen_lod_threshold = 0.0; @@ -1869,72 +1514,116 @@ void RendererSceneRenderForward::_render_shadow(RID p_framebuffer, const PagedAr PassMode pass_mode = p_use_dp ? PASS_MODE_SHADOW_DP : PASS_MODE_SHADOW; - _fill_render_list(p_instances, pass_mode, p_projection, p_transform); + uint32_t render_list_from = render_list[RENDER_LIST_SECONDARY].elements.size(); + _fill_render_list(RENDER_LIST_SECONDARY, p_instances, pass_mode, p_projection, p_transform, false, false, p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold, true); + uint32_t render_list_size = render_list[RENDER_LIST_SECONDARY].elements.size() - render_list_from; + render_list[RENDER_LIST_SECONDARY].sort_by_key_range(render_list_from, render_list_size); + _fill_instance_data(RENDER_LIST_SECONDARY, render_list_from, render_list_size, false); + + { + //regular forward for now + bool flip_cull = p_use_dp_flip; + if (p_flip_y) { + flip_cull = !flip_cull; + } - RID rp_uniform_set = _setup_render_pass_uniform_set(RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); + shadow_pass.element_from = render_list_from; + shadow_pass.element_count = render_list_size; + shadow_pass.flip_cull = flip_cull; + shadow_pass.pass_mode = pass_mode; - RENDER_TIMESTAMP("Render Shadow"); + shadow_pass.rp_uniform_set = RID(); //will be filled later when instance buffer is complete + shadow_pass.camera_plane = p_camera_plane; + shadow_pass.screen_lod_threshold = p_screen_lod_threshold; + shadow_pass.lod_distance_multiplier = p_lod_distance_multiplier; - render_list.sort_by_key(false); + shadow_pass.framebuffer = p_framebuffer; + shadow_pass.initial_depth_action = p_begin ? (p_clear_region ? RD::INITIAL_ACTION_CLEAR_REGION : RD::INITIAL_ACTION_CLEAR) : (p_clear_region ? RD::INITIAL_ACTION_CLEAR_REGION_CONTINUE : RD::INITIAL_ACTION_CONTINUE); + shadow_pass.final_depth_action = p_end ? RD::FINAL_ACTION_READ : RD::FINAL_ACTION_CONTINUE; + shadow_pass.rect = p_rect; - { - //regular forward for now - RenderListParameters render_list_params(render_list.elements, render_list.element_count, p_use_dp_flip, pass_mode, true, rp_uniform_set, false, Vector2(), p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold); - _render_list_with_threads(&render_list_params, p_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ); + scene_state.shadow_passes.push_back(shadow_pass); } } -void RendererSceneRenderForward::_render_particle_collider_heightfield(RID p_fb, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, const PagedArray<GeometryInstance *> &p_instances) { - RENDER_TIMESTAMP("Setup Render Collider Heightfield"); +void RenderForwardClustered::_render_shadow_process() { + _update_instance_data_buffer(RENDER_LIST_SECONDARY); + //render shadows one after the other, so this can be done un-barriered and the driver can optimize (as well as allow us to run compute at the same time) - _update_render_base_uniform_set(); + for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { + //render passes need to be configured after instance buffer is done, since they need the latest version + SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + shadow_pass.rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, RID(), RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>(), false, i); + } + + RD::get_singleton()->draw_command_end_label(); +} +void RenderForwardClustered::_render_shadow_end(uint32_t p_barrier) { + RD::get_singleton()->draw_command_begin_label("Shadow Render"); - render_pass++; + for (uint32_t i = 0; i < scene_state.shadow_passes.size(); i++) { + SceneState::ShadowPass &shadow_pass = scene_state.shadow_passes[i]; + RenderListParameters render_list_parameters(render_list[RENDER_LIST_SECONDARY].elements.ptr() + shadow_pass.element_from, render_list[RENDER_LIST_SECONDARY].element_info.ptr() + shadow_pass.element_from, shadow_pass.element_count, shadow_pass.flip_cull, shadow_pass.pass_mode, true, shadow_pass.rp_uniform_set, false, Vector2(), shadow_pass.camera_plane, shadow_pass.lod_distance_multiplier, shadow_pass.screen_lod_threshold, shadow_pass.element_from, RD::BARRIER_MASK_NO_BARRIER); + _render_list_with_threads(&render_list_parameters, shadow_pass.framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, shadow_pass.initial_depth_action, shadow_pass.final_depth_action, Vector<Color>(), 1.0, 0, shadow_pass.rect); + } + if (p_barrier != RD::BARRIER_MASK_NO_BARRIER) { + RD::get_singleton()->barrier(RD::BARRIER_MASK_RASTER, p_barrier); + } + RD::get_singleton()->draw_command_end_label(); +} + +void RenderForwardClustered::_render_particle_collider_heightfield(RID p_fb, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, const PagedArray<GeometryInstance *> &p_instances) { + RENDER_TIMESTAMP("Setup Render Collider Heightfield"); + + RD::get_singleton()->draw_command_begin_label("Render Collider Heightfield"); + + _update_render_base_uniform_set(); scene_state.ubo.dual_paraboloid_side = 0; - _setup_environment(RID(), RID(), p_cam_projection, p_cam_transform, RID(), true, Vector2(1, 1), RID(), true, Color(), 0, p_cam_projection.get_z_far(), false, false); + _setup_environment(RID(), RID(), p_cam_projection, p_cam_transform, RID(), true, Vector2(1, 1), 1, 32, RID(), true, Color(), 0, p_cam_projection.get_z_far(), false, false); PassMode pass_mode = PASS_MODE_SHADOW; - _fill_render_list(p_instances, pass_mode, p_cam_projection, p_cam_transform); - - RID rp_uniform_set = _setup_render_pass_uniform_set(RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); + _fill_render_list(RENDER_LIST_SECONDARY, p_instances, pass_mode, p_cam_projection, p_cam_transform); + render_list[RENDER_LIST_SECONDARY].sort_by_key(); + _fill_instance_data(RENDER_LIST_SECONDARY); - RENDER_TIMESTAMP("Render Collider Heightield"); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, RID(), RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); - render_list.sort_by_key(false); + RENDER_TIMESTAMP("Render Collider Heightfield"); { //regular forward for now - RenderListParameters render_list_params(render_list.elements, render_list.element_count, false, pass_mode, true, rp_uniform_set); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), false, pass_mode, true, rp_uniform_set); _render_list_with_threads(&render_list_params, p_fb, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ); } + RD::get_singleton()->draw_command_end_label(); } -void RendererSceneRenderForward::_render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) { +void RenderForwardClustered::_render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) { RENDER_TIMESTAMP("Setup Rendering Material"); - _update_render_base_uniform_set(); + RD::get_singleton()->draw_command_begin_label("Render Material"); - render_pass++; + _update_render_base_uniform_set(); scene_state.ubo.dual_paraboloid_side = 0; - scene_state.ubo.material_uv2_mode = true; + scene_state.ubo.material_uv2_mode = false; - _setup_environment(RID(), RID(), p_cam_projection, p_cam_transform, RID(), true, Vector2(1, 1), RID(), false, Color(), 0, 0); + _setup_environment(RID(), RID(), p_cam_projection, p_cam_transform, RID(), true, Vector2(1, 1), 1, 32, RID(), false, Color(), 0, 0); PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; - _fill_render_list(p_instances, pass_mode, p_cam_projection, p_cam_transform); + _fill_render_list(RENDER_LIST_SECONDARY, p_instances, pass_mode, p_cam_projection, p_cam_transform); + render_list[RENDER_LIST_SECONDARY].sort_by_key(); + _fill_instance_data(RENDER_LIST_SECONDARY); - RID rp_uniform_set = _setup_render_pass_uniform_set(RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, RID(), RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); RENDER_TIMESTAMP("Render Material"); - render_list.sort_by_key(false); - { - RenderListParameters render_list_params(render_list.elements, render_list.element_count, true, pass_mode, true, rp_uniform_set); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, true, rp_uniform_set); //regular forward for now Vector<Color> clear; clear.push_back(Color(0, 0, 0, 0)); @@ -1946,31 +1635,33 @@ void RendererSceneRenderForward::_render_material(const Transform &p_cam_transfo _render_list(draw_list, RD::get_singleton()->framebuffer_get_format(p_framebuffer), &render_list_params, 0, render_list_params.element_count); RD::get_singleton()->draw_list_end(); } + + RD::get_singleton()->draw_command_end_label(); } -void RendererSceneRenderForward::_render_uv2(const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) { +void RenderForwardClustered::_render_uv2(const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) { RENDER_TIMESTAMP("Setup Rendering UV2"); - _update_render_base_uniform_set(); + RD::get_singleton()->draw_command_begin_label("Render UV2"); - render_pass++; + _update_render_base_uniform_set(); scene_state.ubo.dual_paraboloid_side = 0; scene_state.ubo.material_uv2_mode = true; - _setup_environment(RID(), RID(), CameraMatrix(), Transform(), RID(), true, Vector2(1, 1), RID(), false, Color(), 0, 0); + _setup_environment(RID(), RID(), CameraMatrix(), Transform(), RID(), true, Vector2(1, 1), 1, 32, RID(), false, Color(), 0, 0); PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; - _fill_render_list(p_instances, pass_mode, CameraMatrix(), Transform()); + _fill_render_list(RENDER_LIST_SECONDARY, p_instances, pass_mode, CameraMatrix(), Transform()); + render_list[RENDER_LIST_SECONDARY].sort_by_key(); + _fill_instance_data(RENDER_LIST_SECONDARY); - RID rp_uniform_set = _setup_render_pass_uniform_set(RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, RID(), RID(), RID(), RID(), RID(), PagedArray<RID>(), PagedArray<RID>()); RENDER_TIMESTAMP("Render Material"); - render_list.sort_by_key(false); - { - RenderListParameters render_list_params(render_list.elements, render_list.element_count, true, pass_mode, true, rp_uniform_set, true); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, true, rp_uniform_set, true); //regular forward for now Vector<Color> clear; clear.push_back(Color(0, 0, 0, 0)); @@ -2006,23 +1697,24 @@ void RendererSceneRenderForward::_render_uv2(const PagedArray<GeometryInstance * RD::get_singleton()->draw_list_end(); } + + RD::get_singleton()->draw_command_end_label(); } -void RendererSceneRenderForward::_render_sdfgi(RID p_render_buffers, const Vector3i &p_from, const Vector3i &p_size, const AABB &p_bounds, const PagedArray<GeometryInstance *> &p_instances, const RID &p_albedo_texture, const RID &p_emission_texture, const RID &p_emission_aniso_texture, const RID &p_geom_facing_texture) { +void RenderForwardClustered::_render_sdfgi(RID p_render_buffers, const Vector3i &p_from, const Vector3i &p_size, const AABB &p_bounds, const PagedArray<GeometryInstance *> &p_instances, const RID &p_albedo_texture, const RID &p_emission_texture, const RID &p_emission_aniso_texture, const RID &p_geom_facing_texture) { RENDER_TIMESTAMP("Render SDFGI"); + RD::get_singleton()->draw_command_begin_label("Render SDFGI Voxel"); + _update_render_base_uniform_set(); - RenderBufferDataForward *render_buffer = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); + RenderBufferDataForwardClustered *render_buffer = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffers); ERR_FAIL_COND(!render_buffer); - render_pass++; - PassMode pass_mode = PASS_MODE_SDF; - _fill_render_list(p_instances, pass_mode, CameraMatrix(), Transform()); - render_list.sort_by_key(false); - - RID rp_uniform_set = _setup_sdfgi_render_pass_uniform_set(p_albedo_texture, p_emission_texture, p_emission_aniso_texture, p_geom_facing_texture); + _fill_render_list(RENDER_LIST_SECONDARY, p_instances, pass_mode, CameraMatrix(), Transform()); + render_list[RENDER_LIST_SECONDARY].sort_by_key(); + _fill_instance_data(RENDER_LIST_SECONDARY); Vector3 half_extents = p_bounds.size * 0.5; Vector3 center = p_bounds.position + half_extents; @@ -2073,7 +1765,9 @@ void RendererSceneRenderForward::_render_sdfgi(RID p_render_buffers, const Vecto RendererStorageRD::store_transform(to_bounds.affine_inverse() * cam_xform, scene_state.ubo.sdf_to_bounds); - _setup_environment(RID(), RID(), camera_proj, cam_xform, RID(), true, Vector2(1, 1), RID(), false, Color(), 0, 0); + _setup_environment(RID(), RID(), camera_proj, cam_xform, RID(), true, Vector2(1, 1), 1, 32, RID(), false, Color(), 0, 0); + + RID rp_uniform_set = _setup_sdfgi_render_pass_uniform_set(p_albedo_texture, p_emission_texture, p_emission_aniso_texture, p_geom_facing_texture); Map<Size2i, RID>::Element *E = sdfgi_framebuffer_size_cache.find(fb_size); if (!E) { @@ -2081,19 +1775,21 @@ void RendererSceneRenderForward::_render_sdfgi(RID p_render_buffers, const Vecto E = sdfgi_framebuffer_size_cache.insert(fb_size, fb); } - RenderListParameters render_list_params(render_list.elements, render_list.element_count, true, pass_mode, true, rp_uniform_set, false); + RenderListParameters render_list_params(render_list[RENDER_LIST_SECONDARY].elements.ptr(), render_list[RENDER_LIST_SECONDARY].element_info.ptr(), render_list[RENDER_LIST_SECONDARY].elements.size(), true, pass_mode, true, rp_uniform_set, false); _render_list_with_threads(&render_list_params, E->get(), RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, Vector<Color>(), 1.0, 0, Rect2(), sbs); } + + RD::get_singleton()->draw_command_end_label(); } -void RendererSceneRenderForward::_base_uniforms_changed() { +void RenderForwardClustered::_base_uniforms_changed() { if (!render_base_uniform_set.is_null() && RD::get_singleton()->uniform_set_is_valid(render_base_uniform_set)) { RD::get_singleton()->free(render_base_uniform_set); } render_base_uniform_set = RID(); } -void RendererSceneRenderForward::_update_render_base_uniform_set() { +void RenderForwardClustered::_update_render_base_uniform_set() { if (render_base_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(render_base_uniform_set) || (lightmap_texture_array_version != storage->lightmap_array_get_version())) { if (render_base_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(render_base_uniform_set)) { RD::get_singleton()->free(render_base_uniform_set); @@ -2128,57 +1824,56 @@ void RendererSceneRenderForward::_update_render_base_uniform_set() { RD::Uniform u; u.binding = 2; u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.ids.push_back(shadow_sampler); + u.ids.push_back(scene_shader.shadow_sampler); uniforms.push_back(u); } { RD::Uniform u; u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(scene_state.uniform_buffer); + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(get_omni_light_buffer()); uniforms.push_back(u); } - { RD::Uniform u; - u.binding = 5; + u.binding = 4; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(get_positional_light_buffer()); + u.ids.push_back(get_spot_light_buffer()); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 6; + u.binding = 5; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.ids.push_back(get_reflection_probe_buffer()); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 7; + u.binding = 6; u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; u.ids.push_back(get_directional_light_buffer()); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 10; + u.binding = 7; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.ids.push_back(scene_state.lightmap_buffer); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 11; + u.binding = 8; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.ids.push_back(scene_state.lightmap_capture_buffer); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 12; + u.binding = 9; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID decal_atlas = storage->decal_atlas_get_texture(); u.ids.push_back(decal_atlas); @@ -2186,7 +1881,7 @@ void RendererSceneRenderForward::_update_render_base_uniform_set() { } { RD::Uniform u; - u.binding = 13; + u.binding = 10; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID decal_atlas = storage->decal_atlas_get_texture_srgb(); u.ids.push_back(decal_atlas); @@ -2194,7 +1889,7 @@ void RendererSceneRenderForward::_update_render_base_uniform_set() { } { RD::Uniform u; - u.binding = 14; + u.binding = 11; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.ids.push_back(get_decal_buffer()); uniforms.push_back(u); @@ -2202,35 +1897,8 @@ void RendererSceneRenderForward::_update_render_base_uniform_set() { { RD::Uniform u; - u.binding = 15; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(get_cluster_builder_texture()); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 16; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(get_cluster_builder_indices_buffer()); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.binding = 17; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - if (directional_shadow_get_texture().is_valid()) { - u.ids.push_back(directional_shadow_get_texture()); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); - } - uniforms.push_back(u); - } - - { - RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 18; + u.binding = 12; u.ids.push_back(storage->global_variables_get_storage_buffer()); uniforms.push_back(u); } @@ -2238,23 +1906,22 @@ void RendererSceneRenderForward::_update_render_base_uniform_set() { if (!low_end) { RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 19; + u.binding = 13; u.ids.push_back(sdfgi_get_ubo()); uniforms.push_back(u); } - render_base_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, default_shader_rd, SCENE_UNIFORM_SET); + render_base_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, scene_shader.default_shader_rd, SCENE_UNIFORM_SET); } } -RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buffers, RID p_radiance_texture, RID p_shadow_atlas, RID p_reflection_atlas, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps) { - if (render_pass_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(render_pass_uniform_set)) { - RD::get_singleton()->free(render_pass_uniform_set); - } +RID RenderForwardClustered::_setup_render_pass_uniform_set(RenderListType p_render_list, RID p_render_buffers, RID p_radiance_texture, RID p_shadow_atlas, RID p_reflection_atlas, RID p_cluster_buffer, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, bool p_use_directional_shadow_atlas, int p_index) { + //there should always be enough uniform buffers for render passes, otherwise bugs + ERR_FAIL_INDEX_V(p_index, (int)scene_state.uniform_buffers.size(), RID()); - RenderBufferDataForward *rb = nullptr; + RenderBufferDataForwardClustered *rb = nullptr; if (p_render_buffers.is_valid()) { - rb = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); + rb = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffers); } //default render buffer and scene state uniform set @@ -2262,6 +1929,24 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff Vector<RD::Uniform> uniforms; { + RD::Uniform u; + u.binding = 0; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(scene_state.uniform_buffers[p_index]); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + RID instance_buffer = scene_state.instance_buffer[p_render_list]; + if (instance_buffer == RID()) { + instance_buffer = scene_shader.default_vec4_xform_buffer; // any buffer will do since its not used + } + u.ids.push_back(instance_buffer); + uniforms.push_back(u); + } + { RID radiance_texture; if (p_radiance_texture.is_valid()) { radiance_texture = p_radiance_texture; @@ -2269,7 +1954,7 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff radiance_texture = storage->texture_rd_get_default(is_using_radiance_cubemap_array() ? RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_ARRAY_BLACK : RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK); } RD::Uniform u; - u.binding = 0; + u.binding = 2; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.push_back(radiance_texture); uniforms.push_back(u); @@ -2278,7 +1963,7 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff { RID ref_texture = p_reflection_atlas.is_valid() ? reflection_atlas_get_texture(p_reflection_atlas) : RID(); RD::Uniform u; - u.binding = 1; + u.binding = 3; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; if (ref_texture.is_valid()) { u.ids.push_back(ref_texture); @@ -2290,7 +1975,7 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff { RD::Uniform u; - u.binding = 2; + u.binding = 4; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID texture; if (p_shadow_atlas.is_valid()) { @@ -2304,7 +1989,18 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } { RD::Uniform u; - u.binding = 3; + u.binding = 5; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + if (p_use_directional_shadow_atlas && directional_shadow_get_texture().is_valid()) { + u.ids.push_back(directional_shadow_get_texture()); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 6; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.resize(scene_state.max_lightmaps); RID default_tex = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_2D_ARRAY_WHITE); @@ -2323,13 +2019,13 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } { RD::Uniform u; - u.binding = 4; + u.binding = 7; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.resize(MAX_GI_PROBES); RID default_tex = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); for (int i = 0; i < MAX_GI_PROBES; i++) { if (i < (int)p_gi_probes.size()) { - RID tex = gi_probe_instance_get_texture(p_gi_probes[i]); + RID tex = gi.gi_probe_instance_get_texture(p_gi_probes[i]); if (!tex.is_valid()) { tex = default_tex; } @@ -2344,7 +2040,16 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff { RD::Uniform u; - u.binding = 5; + u.binding = 8; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + RID cb = p_cluster_buffer.is_valid() ? p_cluster_buffer : scene_shader.default_vec4_xform_buffer; + u.ids.push_back(cb); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.binding = 9; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID texture = (false && rb && rb->depth.is_valid()) ? rb->depth : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); u.ids.push_back(texture); @@ -2352,17 +2057,18 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } { RD::Uniform u; - u.binding = 6; + u.binding = 10; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID bbt = rb ? render_buffers_get_back_buffer_texture(p_render_buffers) : RID(); RID texture = bbt.is_valid() ? bbt : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); u.ids.push_back(texture); uniforms.push_back(u); } + if (!low_end) { { RD::Uniform u; - u.binding = 7; + u.binding = 11; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID texture = rb && rb->normal_roughness_buffer.is_valid() ? rb->normal_roughness_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_NORMAL); u.ids.push_back(texture); @@ -2371,7 +2077,7 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff { RD::Uniform u; - u.binding = 8; + u.binding = 12; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID aot = rb ? render_buffers_get_ao_texture(p_render_buffers) : RID(); RID texture = aot.is_valid() ? aot : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); @@ -2381,24 +2087,26 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff { RD::Uniform u; - u.binding = 9; + u.binding = 13; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - RID texture = rb && rb->ambient_buffer.is_valid() ? rb->ambient_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); + RID ambient_buffer = p_render_buffers.is_valid() ? render_buffers_get_gi_ambient_texture(p_render_buffers) : RID(); + RID texture = ambient_buffer.is_valid() ? ambient_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); u.ids.push_back(texture); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 10; + u.binding = 14; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - RID texture = rb && rb->reflection_buffer.is_valid() ? rb->reflection_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); + RID reflection_buffer = p_render_buffers.is_valid() ? render_buffers_get_gi_reflection_texture(p_render_buffers) : RID(); + RID texture = reflection_buffer.is_valid() ? reflection_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); u.ids.push_back(texture); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 11; + u.binding = 15; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID t; if (rb && render_buffers_is_sdfgi_enabled(p_render_buffers)) { @@ -2411,7 +2119,7 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } { RD::Uniform u; - u.binding = 12; + u.binding = 16; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; if (rb && render_buffers_is_sdfgi_enabled(p_render_buffers)) { u.ids.push_back(render_buffers_get_sdfgi_occlusion_texture(p_render_buffers)); @@ -2422,14 +2130,14 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } { RD::Uniform u; - u.binding = 13; + u.binding = 17; u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; u.ids.push_back(rb ? render_buffers_get_gi_probe_buffer(p_render_buffers) : render_buffers_get_default_gi_probe_buffer()); uniforms.push_back(u); } { RD::Uniform u; - u.binding = 14; + u.binding = 18; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID vfog = RID(); if (rb && render_buffers_has_volumetric_fog(p_render_buffers)) { @@ -2445,11 +2153,19 @@ RID RendererSceneRenderForward::_setup_render_pass_uniform_set(RID p_render_buff } } - render_pass_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, default_shader_rd, RENDER_PASS_UNIFORM_SET); - return render_pass_uniform_set; + if (p_index >= (int)render_pass_uniform_sets.size()) { + render_pass_uniform_sets.resize(p_index + 1); + } + + if (render_pass_uniform_sets[p_index].is_valid() && RD::get_singleton()->uniform_set_is_valid(render_pass_uniform_sets[p_index])) { + RD::get_singleton()->free(render_pass_uniform_sets[p_index]); + } + + render_pass_uniform_sets[p_index] = RD::get_singleton()->uniform_set_create(uniforms, scene_shader.default_shader_rd, RENDER_PASS_UNIFORM_SET); + return render_pass_uniform_sets[p_index]; } -RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albedo_texture, RID p_emission_texture, RID p_emission_aniso_texture, RID p_geom_facing_texture) { +RID RenderForwardClustered::_setup_sdfgi_render_pass_uniform_set(RID p_albedo_texture, RID p_emission_texture, RID p_emission_aniso_texture, RID p_geom_facing_texture) { if (sdfgi_pass_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(sdfgi_pass_uniform_set)) { RD::get_singleton()->free(sdfgi_pass_uniform_set); } @@ -2457,10 +2173,24 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed Vector<RD::Uniform> uniforms; { + RD::Uniform u; + u.binding = 0; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(scene_state.uniform_buffers[0]); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(scene_state.instance_buffer[RENDER_LIST_SECONDARY]); + uniforms.push_back(u); + } + { // No radiance texture. RID radiance_texture = storage->texture_rd_get_default(is_using_radiance_cubemap_array() ? RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_ARRAY_BLACK : RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK); RD::Uniform u; - u.binding = 0; + u.binding = 2; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.push_back(radiance_texture); uniforms.push_back(u); @@ -2470,7 +2200,7 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed // No reflection atlas. RID ref_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_ARRAY_BLACK); RD::Uniform u; - u.binding = 1; + u.binding = 3; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.push_back(ref_texture); uniforms.push_back(u); @@ -2479,7 +2209,17 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed { // No shadow atlas. RD::Uniform u; - u.binding = 2; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + RID texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); + u.ids.push_back(texture); + uniforms.push_back(u); + } + + { + // No directional shadow atlas. + RD::Uniform u; + u.binding = 5; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; RID texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); u.ids.push_back(texture); @@ -2489,7 +2229,7 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed { // No Lightmaps RD::Uniform u; - u.binding = 3; + u.binding = 6; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.resize(scene_state.max_lightmaps); RID default_tex = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_2D_ARRAY_WHITE); @@ -2503,7 +2243,7 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed { // No GIProbes RD::Uniform u; - u.binding = 4; + u.binding = 7; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.ids.resize(MAX_GI_PROBES); RID default_tex = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); @@ -2513,77 +2253,70 @@ RID RendererSceneRenderForward::_setup_sdfgi_render_pass_uniform_set(RID p_albed uniforms.push_back(u); } + + { + RD::Uniform u; + u.binding = 8; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + RID cb = scene_shader.default_vec4_xform_buffer; + u.ids.push_back(cb); + uniforms.push_back(u); + } + // actual sdfgi stuff { RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 5; + u.binding = 9; u.ids.push_back(p_albedo_texture); uniforms.push_back(u); } { RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 6; + u.binding = 10; u.ids.push_back(p_emission_texture); uniforms.push_back(u); } { RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 7; + u.binding = 11; u.ids.push_back(p_emission_aniso_texture); uniforms.push_back(u); } { RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 8; + u.binding = 12; u.ids.push_back(p_geom_facing_texture); uniforms.push_back(u); } - sdfgi_pass_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, default_shader_sdfgi_rd, RENDER_PASS_UNIFORM_SET); + sdfgi_pass_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, scene_shader.default_shader_sdfgi_rd, RENDER_PASS_UNIFORM_SET); return sdfgi_pass_uniform_set; } -void RendererSceneRenderForward::_render_buffers_clear_uniform_set(RenderBufferDataForward *rb) { +void RenderForwardClustered::_render_buffers_clear_uniform_set(RenderBufferDataForwardClustered *rb) { } -void RendererSceneRenderForward::_render_buffers_uniform_set_changed(RID p_render_buffers) { - RenderBufferDataForward *rb = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); +void RenderForwardClustered::_render_buffers_uniform_set_changed(RID p_render_buffers) { + RenderBufferDataForwardClustered *rb = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffers); _render_buffers_clear_uniform_set(rb); } -RID RendererSceneRenderForward::_render_buffers_get_normal_texture(RID p_render_buffers) { - RenderBufferDataForward *rb = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); +RID RenderForwardClustered::_render_buffers_get_normal_texture(RID p_render_buffers) { + RenderBufferDataForwardClustered *rb = (RenderBufferDataForwardClustered *)render_buffers_get_data(p_render_buffers); return rb->normal_roughness_buffer; } -RID RendererSceneRenderForward::_render_buffers_get_ambient_texture(RID p_render_buffers) { - RenderBufferDataForward *rb = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); - - return rb->ambient_buffer; -} - -RID RendererSceneRenderForward::_render_buffers_get_reflection_texture(RID p_render_buffers) { - RenderBufferDataForward *rb = (RenderBufferDataForward *)render_buffers_get_data(p_render_buffers); +RenderForwardClustered *RenderForwardClustered::singleton = nullptr; - return rb->reflection_buffer; -} - -RendererSceneRenderForward *RendererSceneRenderForward::singleton = nullptr; - -void RendererSceneRenderForward::set_time(double p_time, double p_step) { - time = p_time; - RendererSceneRenderRD::set_time(p_time, p_step); -} - -void RendererSceneRenderForward::_geometry_instance_mark_dirty(GeometryInstance *p_geometry_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::_geometry_instance_mark_dirty(GeometryInstance *p_geometry_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); if (ginstance->dirty_list_element.in_list()) { return; } @@ -2602,7 +2335,7 @@ void RendererSceneRenderForward::_geometry_instance_mark_dirty(GeometryInstance geometry_instance_dirty_list.add(&ginstance->dirty_list_element); } -void RendererSceneRenderForward::_geometry_instance_add_surface_with_material(GeometryInstanceForward *ginstance, uint32_t p_surface, MaterialData *p_material, uint32_t p_material_id, uint32_t p_shader_id, RID p_mesh) { +void RenderForwardClustered::_geometry_instance_add_surface_with_material(GeometryInstanceForwardClustered *ginstance, uint32_t p_surface, SceneShaderForwardClustered::MaterialData *p_material, uint32_t p_material_id, uint32_t p_shader_id, RID p_mesh) { bool has_read_screen_alpha = p_material->shader_data->uses_screen_texture || p_material->shader_data->uses_depth_texture || p_material->shader_data->uses_normal_texture; bool has_base_alpha = (p_material->shader_data->uses_alpha || has_read_screen_alpha); bool has_blend_alpha = p_material->shader_data->uses_blend_alpha; @@ -2626,14 +2359,14 @@ void RendererSceneRenderForward::_geometry_instance_add_surface_with_material(Ge flags |= GeometryInstanceSurfaceDataCache::FLAG_USES_NORMAL_TEXTURE; } - if (ginstance->data->cast_double_sided_shaodows) { + if (ginstance->data->cast_double_sided_shadows) { flags |= GeometryInstanceSurfaceDataCache::FLAG_USES_DOUBLE_SIDED_SHADOWS; } - if (has_alpha || has_read_screen_alpha || p_material->shader_data->depth_draw == ShaderData::DEPTH_DRAW_DISABLED || p_material->shader_data->depth_test == ShaderData::DEPTH_TEST_DISABLED) { + if (has_alpha || has_read_screen_alpha || p_material->shader_data->depth_draw == SceneShaderForwardClustered::ShaderData::DEPTH_DRAW_DISABLED || p_material->shader_data->depth_test == SceneShaderForwardClustered::ShaderData::DEPTH_TEST_DISABLED) { //material is only meant for alpha pass flags |= GeometryInstanceSurfaceDataCache::FLAG_PASS_ALPHA; - if (p_material->shader_data->uses_depth_pre_pass && !(p_material->shader_data->depth_draw == ShaderData::DEPTH_DRAW_DISABLED || p_material->shader_data->depth_test == ShaderData::DEPTH_TEST_DISABLED)) { + if (p_material->shader_data->uses_depth_pre_pass && !(p_material->shader_data->depth_draw == SceneShaderForwardClustered::ShaderData::DEPTH_DRAW_DISABLED || p_material->shader_data->depth_test == SceneShaderForwardClustered::ShaderData::DEPTH_TEST_DISABLED)) { flags |= GeometryInstanceSurfaceDataCache::FLAG_PASS_DEPTH; flags |= GeometryInstanceSurfaceDataCache::FLAG_PASS_SHADOW; } @@ -2643,11 +2376,18 @@ void RendererSceneRenderForward::_geometry_instance_add_surface_with_material(Ge flags |= GeometryInstanceSurfaceDataCache::FLAG_PASS_SHADOW; } - MaterialData *material_shadow = nullptr; - //void *surface_shadow = nullptr; + SceneShaderForwardClustered::MaterialData *material_shadow = nullptr; + void *surface_shadow = nullptr; if (!p_material->shader_data->writes_modelview_or_projection && !p_material->shader_data->uses_vertex && !p_material->shader_data->uses_discard && !p_material->shader_data->uses_depth_pre_pass) { flags |= GeometryInstanceSurfaceDataCache::FLAG_USES_SHARED_SHADOW_MATERIAL; - material_shadow = (MaterialData *)storage->material_get_data(default_material, RendererStorageRD::SHADER_TYPE_3D); + material_shadow = (SceneShaderForwardClustered::MaterialData *)storage->material_get_data(scene_shader.default_material, RendererStorageRD::SHADER_TYPE_3D); + + RID shadow_mesh = storage->mesh_get_shadow_mesh(p_mesh); + + if (shadow_mesh.is_valid()) { + surface_shadow = storage->mesh_get_surface(shadow_mesh, p_surface); + } + } else { material_shadow = p_material; } @@ -2669,7 +2409,8 @@ void RendererSceneRenderForward::_geometry_instance_add_surface_with_material(Ge //shadow sdcache->shader_shadow = material_shadow->shader_data; sdcache->material_uniform_set_shadow = material_shadow->uniform_set; - sdcache->surface_shadow = sdcache->surface; //when adding special shadow meshes, will use this + + sdcache->surface_shadow = surface_shadow ? surface_shadow : sdcache->surface; sdcache->owner = ginstance; @@ -2681,23 +2422,24 @@ void RendererSceneRenderForward::_geometry_instance_add_surface_with_material(Ge sdcache->sort.sort_key1 = 0; sdcache->sort.sort_key2 = 0; - sdcache->sort.surface_type = ginstance->data->base_type; - sdcache->sort.material_id = p_material_id; + sdcache->sort.surface_index = p_surface; + sdcache->sort.material_id_low = p_material_id & 0x3FFF; + sdcache->sort.material_id_hi = p_material_id >> 14; sdcache->sort.shader_id = p_shader_id; - sdcache->sort.geometry_id = p_mesh.get_local_index(); + sdcache->sort.geometry_id = p_mesh.get_local_index(); //only meshes can repeat anyway sdcache->sort.uses_forward_gi = ginstance->can_sdfgi; sdcache->sort.priority = p_material->priority; } -void RendererSceneRenderForward::_geometry_instance_add_surface(GeometryInstanceForward *ginstance, uint32_t p_surface, RID p_material, RID p_mesh) { +void RenderForwardClustered::_geometry_instance_add_surface(GeometryInstanceForwardClustered *ginstance, uint32_t p_surface, RID p_material, RID p_mesh) { RID m_src; m_src = ginstance->data->material_override.is_valid() ? ginstance->data->material_override : p_material; - MaterialData *material = nullptr; + SceneShaderForwardClustered::MaterialData *material = nullptr; if (m_src.is_valid()) { - material = (MaterialData *)storage->material_get_data(m_src, RendererStorageRD::SHADER_TYPE_3D); + material = (SceneShaderForwardClustered::MaterialData *)storage->material_get_data(m_src, RendererStorageRD::SHADER_TYPE_3D); if (!material || !material->shader_data->valid) { material = nullptr; } @@ -2708,8 +2450,8 @@ void RendererSceneRenderForward::_geometry_instance_add_surface(GeometryInstance storage->material_update_dependency(m_src, &ginstance->data->dependency_tracker); } } else { - material = (MaterialData *)storage->material_get_data(default_material, RendererStorageRD::SHADER_TYPE_3D); - m_src = default_material; + material = (SceneShaderForwardClustered::MaterialData *)storage->material_get_data(scene_shader.default_material, RendererStorageRD::SHADER_TYPE_3D); + m_src = scene_shader.default_material; } ERR_FAIL_COND(!material); @@ -2718,7 +2460,7 @@ void RendererSceneRenderForward::_geometry_instance_add_surface(GeometryInstance while (material->next_pass.is_valid()) { RID next_pass = material->next_pass; - material = (MaterialData *)storage->material_get_data(next_pass, RendererStorageRD::SHADER_TYPE_3D); + material = (SceneShaderForwardClustered::MaterialData *)storage->material_get_data(next_pass, RendererStorageRD::SHADER_TYPE_3D); if (!material || !material->shader_data->valid) { break; } @@ -2729,8 +2471,8 @@ void RendererSceneRenderForward::_geometry_instance_add_surface(GeometryInstance } } -void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_geometry_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::_geometry_instance_update(GeometryInstance *p_geometry_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); if (ginstance->data->dirty_dependencies) { ginstance->data->dependency_tracker.update_begin(); @@ -2790,8 +2532,9 @@ void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_g for (int j = 0; j < draw_passes; j++) { RID mesh = storage->particles_get_draw_pass_mesh(ginstance->data->base, j); - if (!mesh.is_valid()) + if (!mesh.is_valid()) { continue; + } const RID *materials = nullptr; uint32_t surface_count; @@ -2814,11 +2557,6 @@ void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_g //Fill push constant - ginstance->push_constant.instance_uniforms_ofs = ginstance->data->shader_parameters_offset >= 0 ? ginstance->data->shader_parameters_offset : 0; - ginstance->push_constant.layer_mask = ginstance->data->layer_mask; - ginstance->push_constant.flags = 0; - ginstance->push_constant.gi_offset = 0xFFFFFFFF; //disabled - bool store_transform = true; if (ginstance->data->base_type == RS::INSTANCE_MULTIMESH) { @@ -2840,7 +2578,7 @@ void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_g } ginstance->base_flags |= (stride << INSTANCE_DATA_FLAGS_MULTIMESH_STRIDE_SHIFT); - ginstance->transforms_uniform_set = storage->multimesh_get_3d_uniform_set(ginstance->data->base, default_shader_rd, TRANSFORMS_UNIFORM_SET); + ginstance->transforms_uniform_set = storage->multimesh_get_3d_uniform_set(ginstance->data->base, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); } else if (ginstance->data->base_type == RS::INSTANCE_PARTICLES) { ginstance->base_flags |= INSTANCE_DATA_FLAG_MULTIMESH; @@ -2863,33 +2601,22 @@ void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_g if (!storage->particles_is_using_local_coords(ginstance->data->base)) { store_transform = false; } - ginstance->transforms_uniform_set = storage->particles_get_instance_buffer_uniform_set(ginstance->data->base, default_shader_rd, TRANSFORMS_UNIFORM_SET); + ginstance->transforms_uniform_set = storage->particles_get_instance_buffer_uniform_set(ginstance->data->base, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); } else if (ginstance->data->base_type == RS::INSTANCE_MESH) { if (storage->skeleton_is_valid(ginstance->data->skeleton)) { ginstance->base_flags |= INSTANCE_DATA_FLAG_SKELETON; - ginstance->transforms_uniform_set = storage->skeleton_get_3d_uniform_set(ginstance->data->skeleton, default_shader_rd, TRANSFORMS_UNIFORM_SET); + ginstance->transforms_uniform_set = storage->skeleton_get_3d_uniform_set(ginstance->data->skeleton, scene_shader.default_shader_rd, TRANSFORMS_UNIFORM_SET); if (ginstance->data->dirty_dependencies) { storage->skeleton_update_dependency(ginstance->data->skeleton, &ginstance->data->dependency_tracker); } } } - if (store_transform) { - RendererStorageRD::store_transform(ginstance->data->transform, ginstance->push_constant.transform); - } else { - RendererStorageRD::store_transform(Transform(), ginstance->push_constant.transform); - } - + ginstance->store_transform_cache = store_transform; ginstance->can_sdfgi = false; - if (lightmap_instance_is_valid(ginstance->lightmap_instance)) { - ginstance->push_constant.gi_offset = ginstance->data->lightmap_slice_index << 16; - ginstance->push_constant.lightmap_uv_scale[0] = ginstance->data->lightmap_uv_scale.position.x; - ginstance->push_constant.lightmap_uv_scale[1] = ginstance->data->lightmap_uv_scale.position.y; - ginstance->push_constant.lightmap_uv_scale[2] = ginstance->data->lightmap_uv_scale.size.width; - ginstance->push_constant.lightmap_uv_scale[3] = ginstance->data->lightmap_uv_scale.size.height; - } else if (!low_end) { + if (!lightmap_instance_is_valid(ginstance->lightmap_instance) && !low_end) { if (ginstance->gi_probes[0].is_null() && (ginstance->data->use_baked_light || ginstance->data->use_dynamic_gi)) { ginstance->can_sdfgi = true; } @@ -2903,24 +2630,24 @@ void RendererSceneRenderForward::_geometry_instance_update(GeometryInstance *p_g ginstance->dirty_list_element.remove_from_list(); } -void RendererSceneRenderForward::_update_dirty_geometry_instances() { +void RenderForwardClustered::_update_dirty_geometry_instances() { while (geometry_instance_dirty_list.first()) { _geometry_instance_update(geometry_instance_dirty_list.first()->self()); } } -void RendererSceneRenderForward::_geometry_instance_dependency_changed(RendererStorage::DependencyChangedNotification p_notification, RendererStorage::DependencyTracker *p_tracker) { +void RenderForwardClustered::_geometry_instance_dependency_changed(RendererStorage::DependencyChangedNotification p_notification, RendererStorage::DependencyTracker *p_tracker) { switch (p_notification) { case RendererStorage::DEPENDENCY_CHANGED_MATERIAL: case RendererStorage::DEPENDENCY_CHANGED_MESH: case RendererStorage::DEPENDENCY_CHANGED_MULTIMESH: case RendererStorage::DEPENDENCY_CHANGED_SKELETON_DATA: { - static_cast<RendererSceneRenderForward *>(singleton)->_geometry_instance_mark_dirty(static_cast<GeometryInstance *>(p_tracker->userdata)); + static_cast<RenderForwardClustered *>(singleton)->_geometry_instance_mark_dirty(static_cast<GeometryInstance *>(p_tracker->userdata)); } break; case RendererStorage::DEPENDENCY_CHANGED_MULTIMESH_VISIBLE_INSTANCES: { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_tracker->userdata); + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_tracker->userdata); if (ginstance->data->base_type == RS::INSTANCE_MULTIMESH) { - ginstance->instance_count = static_cast<RendererSceneRenderForward *>(singleton)->storage->multimesh_get_instances_to_draw(ginstance->data->base); + ginstance->instance_count = static_cast<RenderForwardClustered *>(singleton)->storage->multimesh_get_instances_to_draw(ginstance->data->base); } } break; default: { @@ -2928,16 +2655,16 @@ void RendererSceneRenderForward::_geometry_instance_dependency_changed(RendererS } break; } } -void RendererSceneRenderForward::_geometry_instance_dependency_deleted(const RID &p_dependency, RendererStorage::DependencyTracker *p_tracker) { - static_cast<RendererSceneRenderForward *>(singleton)->_geometry_instance_mark_dirty(static_cast<GeometryInstance *>(p_tracker->userdata)); +void RenderForwardClustered::_geometry_instance_dependency_deleted(const RID &p_dependency, RendererStorage::DependencyTracker *p_tracker) { + static_cast<RenderForwardClustered *>(singleton)->_geometry_instance_mark_dirty(static_cast<GeometryInstance *>(p_tracker->userdata)); } -RendererSceneRender::GeometryInstance *RendererSceneRenderForward::geometry_instance_create(RID p_base) { +RendererSceneRender::GeometryInstance *RenderForwardClustered::geometry_instance_create(RID p_base) { RS::InstanceType type = storage->get_base_type(p_base); ERR_FAIL_COND_V(!((1 << type) & RS::INSTANCE_GEOMETRY_MASK), nullptr); - GeometryInstanceForward *ginstance = geometry_instance_alloc.alloc(); - ginstance->data = memnew(GeometryInstanceForward::Data); + GeometryInstanceForwardClustered *ginstance = geometry_instance_alloc.alloc(); + ginstance->data = memnew(GeometryInstanceForwardClustered::Data); ginstance->data->base = p_base; ginstance->data->base_type = type; @@ -2949,38 +2676,37 @@ RendererSceneRender::GeometryInstance *RendererSceneRenderForward::geometry_inst return ginstance; } -void RendererSceneRenderForward::geometry_instance_set_skeleton(GeometryInstance *p_geometry_instance, RID p_skeleton) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_skeleton(GeometryInstance *p_geometry_instance, RID p_skeleton) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->data->skeleton = p_skeleton; _geometry_instance_mark_dirty(ginstance); ginstance->data->dirty_dependencies = true; } -void RendererSceneRenderForward::geometry_instance_set_material_override(GeometryInstance *p_geometry_instance, RID p_override) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_material_override(GeometryInstance *p_geometry_instance, RID p_override) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->data->material_override = p_override; _geometry_instance_mark_dirty(ginstance); ginstance->data->dirty_dependencies = true; } -void RendererSceneRenderForward::geometry_instance_set_surface_materials(GeometryInstance *p_geometry_instance, const Vector<RID> &p_materials) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_surface_materials(GeometryInstance *p_geometry_instance, const Vector<RID> &p_materials) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->data->surface_materials = p_materials; _geometry_instance_mark_dirty(ginstance); ginstance->data->dirty_dependencies = true; } -void RendererSceneRenderForward::geometry_instance_set_mesh_instance(GeometryInstance *p_geometry_instance, RID p_mesh_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_mesh_instance(GeometryInstance *p_geometry_instance, RID p_mesh_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->mesh_instance = p_mesh_instance; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_transform(GeometryInstance *p_geometry_instance, const Transform &p_transform, const AABB &p_aabb, const AABB &p_transformed_aabb) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_transform(GeometryInstance *p_geometry_instance, const Transform &p_transform, const AABB &p_aabb, const AABB &p_transformed_aabb) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); - RendererStorageRD::store_transform(p_transform, ginstance->push_constant.transform); - ginstance->data->transform = p_transform; + ginstance->transform = p_transform; ginstance->mirror = p_transform.basis.determinant() < 0; ginstance->data->aabb = p_aabb; ginstance->transformed_aabb = p_transformed_aabb; @@ -2994,33 +2720,33 @@ void RendererSceneRenderForward::geometry_instance_set_transform(GeometryInstanc ginstance->lod_model_scale = max_scale; } -void RendererSceneRenderForward::geometry_instance_set_lod_bias(GeometryInstance *p_geometry_instance, float p_lod_bias) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_lod_bias(GeometryInstance *p_geometry_instance, float p_lod_bias) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->lod_bias = p_lod_bias; } -void RendererSceneRenderForward::geometry_instance_set_use_baked_light(GeometryInstance *p_geometry_instance, bool p_enable) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_use_baked_light(GeometryInstance *p_geometry_instance, bool p_enable) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->data->use_baked_light = p_enable; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_use_dynamic_gi(GeometryInstance *p_geometry_instance, bool p_enable) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_use_dynamic_gi(GeometryInstance *p_geometry_instance, bool p_enable) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->data->use_dynamic_gi = p_enable; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_use_lightmap(GeometryInstance *p_geometry_instance, RID p_lightmap_instance, const Rect2 &p_lightmap_uv_scale, int p_lightmap_slice_index) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_use_lightmap(GeometryInstance *p_geometry_instance, RID p_lightmap_instance, const Rect2 &p_lightmap_uv_scale, int p_lightmap_slice_index) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); ginstance->lightmap_instance = p_lightmap_instance; - ginstance->data->lightmap_uv_scale = p_lightmap_uv_scale; - ginstance->data->lightmap_slice_index = p_lightmap_slice_index; + ginstance->lightmap_uv_scale = p_lightmap_uv_scale; + ginstance->lightmap_slice_index = p_lightmap_slice_index; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_lightmap_capture(GeometryInstance *p_geometry_instance, const Color *p_sh9) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_lightmap_capture(GeometryInstance *p_geometry_instance, const Color *p_sh9) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); if (p_sh9) { if (ginstance->lightmap_sh == nullptr) { @@ -3036,29 +2762,28 @@ void RendererSceneRenderForward::geometry_instance_set_lightmap_capture(Geometry } _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_instance_shader_parameters_offset(GeometryInstance *p_geometry_instance, int32_t p_offset) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_instance_shader_parameters_offset(GeometryInstance *p_geometry_instance, int32_t p_offset) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); - ginstance->data->shader_parameters_offset = p_offset; + ginstance->shader_parameters_offset = p_offset; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_cast_double_sided_shadows(GeometryInstance *p_geometry_instance, bool p_enable) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_cast_double_sided_shadows(GeometryInstance *p_geometry_instance, bool p_enable) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); - ginstance->data->cast_double_sided_shaodows = p_enable; + ginstance->data->cast_double_sided_shadows = p_enable; _geometry_instance_mark_dirty(ginstance); } -void RendererSceneRenderForward::geometry_instance_set_layer_mask(GeometryInstance *p_geometry_instance, uint32_t p_layer_mask) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_set_layer_mask(GeometryInstance *p_geometry_instance, uint32_t p_layer_mask) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); - ginstance->data->layer_mask = p_layer_mask; - ginstance->push_constant.layer_mask = p_layer_mask; + ginstance->layer_mask = p_layer_mask; } -void RendererSceneRenderForward::geometry_instance_free(GeometryInstance *p_geometry_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_free(GeometryInstance *p_geometry_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); if (ginstance->lightmap_sh != nullptr) { geometry_instance_lightmap_sh.free(ginstance->lightmap_sh); @@ -3073,29 +2798,29 @@ void RendererSceneRenderForward::geometry_instance_free(GeometryInstance *p_geom geometry_instance_alloc.free(ginstance); } -uint32_t RendererSceneRenderForward::geometry_instance_get_pair_mask() { +uint32_t RenderForwardClustered::geometry_instance_get_pair_mask() { return (1 << RS::INSTANCE_GI_PROBE); } -void RendererSceneRenderForward::geometry_instance_pair_light_instances(GeometryInstance *p_geometry_instance, const RID *p_light_instances, uint32_t p_light_instance_count) { +void RenderForwardClustered::geometry_instance_pair_light_instances(GeometryInstance *p_geometry_instance, const RID *p_light_instances, uint32_t p_light_instance_count) { } -void RendererSceneRenderForward::geometry_instance_pair_reflection_probe_instances(GeometryInstance *p_geometry_instance, const RID *p_reflection_probe_instances, uint32_t p_reflection_probe_instance_count) { +void RenderForwardClustered::geometry_instance_pair_reflection_probe_instances(GeometryInstance *p_geometry_instance, const RID *p_reflection_probe_instances, uint32_t p_reflection_probe_instance_count) { } -void RendererSceneRenderForward::geometry_instance_pair_decal_instances(GeometryInstance *p_geometry_instance, const RID *p_decal_instances, uint32_t p_decal_instance_count) { +void RenderForwardClustered::geometry_instance_pair_decal_instances(GeometryInstance *p_geometry_instance, const RID *p_decal_instances, uint32_t p_decal_instance_count) { } -Transform RendererSceneRenderForward::geometry_instance_get_transform(GeometryInstance *p_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_instance); +Transform RenderForwardClustered::geometry_instance_get_transform(GeometryInstance *p_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_instance); ERR_FAIL_COND_V(!ginstance, Transform()); - return ginstance->data->transform; + return ginstance->transform; } -AABB RendererSceneRenderForward::geometry_instance_get_aabb(GeometryInstance *p_instance) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_instance); +AABB RenderForwardClustered::geometry_instance_get_aabb(GeometryInstance *p_instance) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_instance); ERR_FAIL_COND_V(!ginstance, AABB()); return ginstance->data->aabb; } -void RendererSceneRenderForward::geometry_instance_pair_gi_probe_instances(GeometryInstance *p_geometry_instance, const RID *p_gi_probe_instances, uint32_t p_gi_probe_instance_count) { - GeometryInstanceForward *ginstance = static_cast<GeometryInstanceForward *>(p_geometry_instance); +void RenderForwardClustered::geometry_instance_pair_gi_probe_instances(GeometryInstance *p_geometry_instance, const RID *p_gi_probe_instances, uint32_t p_gi_probe_instance_count) { + GeometryInstanceForwardClustered *ginstance = static_cast<GeometryInstanceForwardClustered *>(p_geometry_instance); ERR_FAIL_COND(!ginstance); if (p_gi_probe_instance_count > 0) { ginstance->gi_probes[0] = p_gi_probe_instances[0]; @@ -3110,11 +2835,9 @@ void RendererSceneRenderForward::geometry_instance_pair_gi_probe_instances(Geome } } -RendererSceneRenderForward::RendererSceneRenderForward(RendererStorageRD *p_storage) : +RenderForwardClustered::RenderForwardClustered(RendererStorageRD *p_storage) : RendererSceneRenderRD(p_storage) { singleton = this; - low_end = is_low_end(); - storage = p_storage; /* SCENE SHADER */ @@ -3128,7 +2851,7 @@ RendererSceneRenderForward::RendererSceneRenderForward(RendererStorageRD *p_stor if (is_using_radiance_cubemap_array()) { defines += "\n#define USE_RADIANCE_CUBEMAP_ARRAY \n"; } - defines += "\n#define SDFGI_OCT_SIZE " + itos(sdfgi_get_lightprobe_octahedron_size()) + "\n"; + defines += "\n#define SDFGI_OCT_SIZE " + itos(gi.sdfgi_get_lightprobe_octahedron_size()) + "\n"; defines += "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(get_max_directional_lights()) + "\n"; { @@ -3149,294 +2872,37 @@ RendererSceneRenderForward::RendererSceneRenderForward(RendererStorageRD *p_stor defines += "\n#define MATERIAL_UNIFORM_SET " + itos(MATERIAL_UNIFORM_SET) + "\n"; } - Vector<String> shader_versions; - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n"); - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_DUAL_PARABOLOID\n"); - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_NORMAL_ROUGHNESS\n"); - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_NORMAL_ROUGHNESS\n#define MODE_RENDER_GIPROBE\n"); - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_MATERIAL\n"); - shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_SDF\n"); - shader_versions.push_back(""); - shader_versions.push_back("\n#define USE_FORWARD_GI\n"); - shader_versions.push_back("\n#define MODE_MULTIPLE_RENDER_TARGETS\n"); - shader_versions.push_back("\n#define USE_LIGHTMAP\n"); - shader_versions.push_back("\n#define MODE_MULTIPLE_RENDER_TARGETS\n#define USE_LIGHTMAP\n"); - shader.scene_shader.initialize(shader_versions, defines); - - if (is_low_end()) { - //disable the high end versions - shader.scene_shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS, false); - shader.scene_shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE, false); - shader.scene_shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_SDF, false); - shader.scene_shader.set_variant_enabled(SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI, false); - shader.scene_shader.set_variant_enabled(SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR, false); - shader.scene_shader.set_variant_enabled(SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR, false); - } - } - - storage->shader_set_data_request_function(RendererStorageRD::SHADER_TYPE_3D, _create_shader_funcs); - storage->material_set_data_request_function(RendererStorageRD::SHADER_TYPE_3D, _create_material_funcs); - - { - //shader compiler - ShaderCompilerRD::DefaultIdentifierActions actions; - - actions.renames["WORLD_MATRIX"] = "world_matrix"; - actions.renames["WORLD_NORMAL_MATRIX"] = "world_normal_matrix"; - actions.renames["INV_CAMERA_MATRIX"] = "scene_data.inv_camera_matrix"; - actions.renames["CAMERA_MATRIX"] = "scene_data.camera_matrix"; - actions.renames["PROJECTION_MATRIX"] = "projection_matrix"; - actions.renames["INV_PROJECTION_MATRIX"] = "scene_data.inv_projection_matrix"; - actions.renames["MODELVIEW_MATRIX"] = "modelview"; - actions.renames["MODELVIEW_NORMAL_MATRIX"] = "modelview_normal"; - - actions.renames["VERTEX"] = "vertex"; - actions.renames["NORMAL"] = "normal"; - actions.renames["TANGENT"] = "tangent"; - actions.renames["BINORMAL"] = "binormal"; - actions.renames["POSITION"] = "position"; - actions.renames["UV"] = "uv_interp"; - actions.renames["UV2"] = "uv2_interp"; - actions.renames["COLOR"] = "color_interp"; - actions.renames["POINT_SIZE"] = "gl_PointSize"; - actions.renames["INSTANCE_ID"] = "gl_InstanceIndex"; - - actions.renames["ALPHA_SCISSOR_THRESHOLD"] = "alpha_scissor_threshold"; - actions.renames["ALPHA_HASH_SCALE"] = "alpha_hash_scale"; - actions.renames["ALPHA_ANTIALIASING_EDGE"] = "alpha_antialiasing_edge"; - actions.renames["ALPHA_TEXTURE_COORDINATE"] = "alpha_texture_coordinate"; - - //builtins - - actions.renames["TIME"] = "scene_data.time"; - actions.renames["VIEWPORT_SIZE"] = "scene_data.viewport_size"; - - actions.renames["FRAGCOORD"] = "gl_FragCoord"; - actions.renames["FRONT_FACING"] = "gl_FrontFacing"; - actions.renames["NORMAL_MAP"] = "normal_map"; - actions.renames["NORMAL_MAP_DEPTH"] = "normal_map_depth"; - actions.renames["ALBEDO"] = "albedo"; - actions.renames["ALPHA"] = "alpha"; - actions.renames["METALLIC"] = "metallic"; - actions.renames["SPECULAR"] = "specular"; - actions.renames["ROUGHNESS"] = "roughness"; - actions.renames["RIM"] = "rim"; - actions.renames["RIM_TINT"] = "rim_tint"; - actions.renames["CLEARCOAT"] = "clearcoat"; - actions.renames["CLEARCOAT_GLOSS"] = "clearcoat_gloss"; - actions.renames["ANISOTROPY"] = "anisotropy"; - actions.renames["ANISOTROPY_FLOW"] = "anisotropy_flow"; - actions.renames["SSS_STRENGTH"] = "sss_strength"; - actions.renames["SSS_TRANSMITTANCE_COLOR"] = "transmittance_color"; - actions.renames["SSS_TRANSMITTANCE_DEPTH"] = "transmittance_depth"; - actions.renames["SSS_TRANSMITTANCE_CURVE"] = "transmittance_curve"; - actions.renames["SSS_TRANSMITTANCE_BOOST"] = "transmittance_boost"; - actions.renames["BACKLIGHT"] = "backlight"; - actions.renames["AO"] = "ao"; - actions.renames["AO_LIGHT_AFFECT"] = "ao_light_affect"; - actions.renames["EMISSION"] = "emission"; - actions.renames["POINT_COORD"] = "gl_PointCoord"; - actions.renames["INSTANCE_CUSTOM"] = "instance_custom"; - actions.renames["SCREEN_UV"] = "screen_uv"; - actions.renames["SCREEN_TEXTURE"] = "color_buffer"; - actions.renames["DEPTH_TEXTURE"] = "depth_buffer"; - actions.renames["NORMAL_ROUGHNESS_TEXTURE"] = "normal_roughness_buffer"; - actions.renames["DEPTH"] = "gl_FragDepth"; - actions.renames["OUTPUT_IS_SRGB"] = "true"; - actions.renames["FOG"] = "custom_fog"; - actions.renames["RADIANCE"] = "custom_radiance"; - actions.renames["IRRADIANCE"] = "custom_irradiance"; - actions.renames["BONE_INDICES"] = "bone_attrib"; - actions.renames["BONE_WEIGHTS"] = "weight_attrib"; - actions.renames["CUSTOM0"] = "custom0_attrib"; - actions.renames["CUSTOM1"] = "custom1_attrib"; - actions.renames["CUSTOM2"] = "custom2_attrib"; - actions.renames["CUSTOM3"] = "custom3_attrib"; - - //for light - actions.renames["VIEW"] = "view"; - actions.renames["LIGHT_COLOR"] = "light_color"; - actions.renames["LIGHT"] = "light"; - actions.renames["ATTENUATION"] = "attenuation"; - actions.renames["SHADOW_ATTENUATION"] = "shadow_attenuation"; - actions.renames["DIFFUSE_LIGHT"] = "diffuse_light"; - actions.renames["SPECULAR_LIGHT"] = "specular_light"; - - actions.usage_defines["NORMAL"] = "#define NORMAL_USED\n"; - actions.usage_defines["TANGENT"] = "#define TANGENT_USED\n"; - actions.usage_defines["BINORMAL"] = "@TANGENT"; - actions.usage_defines["RIM"] = "#define LIGHT_RIM_USED\n"; - actions.usage_defines["RIM_TINT"] = "@RIM"; - actions.usage_defines["CLEARCOAT"] = "#define LIGHT_CLEARCOAT_USED\n"; - actions.usage_defines["CLEARCOAT_GLOSS"] = "@CLEARCOAT"; - actions.usage_defines["ANISOTROPY"] = "#define LIGHT_ANISOTROPY_USED\n"; - actions.usage_defines["ANISOTROPY_FLOW"] = "@ANISOTROPY"; - actions.usage_defines["AO"] = "#define AO_USED\n"; - actions.usage_defines["AO_LIGHT_AFFECT"] = "#define AO_USED\n"; - actions.usage_defines["UV"] = "#define UV_USED\n"; - actions.usage_defines["UV2"] = "#define UV2_USED\n"; - actions.usage_defines["BONE_INDICES"] = "#define BONES_USED\n"; - actions.usage_defines["BONE_WEIGHTS"] = "#define WEIGHTS_USED\n"; - actions.usage_defines["CUSTOM0"] = "#define CUSTOM0\n"; - actions.usage_defines["CUSTOM1"] = "#define CUSTOM1\n"; - actions.usage_defines["CUSTOM2"] = "#define CUSTOM2\n"; - actions.usage_defines["CUSTOM3"] = "#define CUSTOM3\n"; - actions.usage_defines["NORMAL_MAP"] = "#define NORMAL_MAP_USED\n"; - actions.usage_defines["NORMAL_MAP_DEPTH"] = "@NORMAL_MAP"; - actions.usage_defines["COLOR"] = "#define COLOR_USED\n"; - actions.usage_defines["INSTANCE_CUSTOM"] = "#define ENABLE_INSTANCE_CUSTOM\n"; - actions.usage_defines["POSITION"] = "#define OVERRIDE_POSITION\n"; - - actions.usage_defines["ALPHA_SCISSOR_THRESHOLD"] = "#define ALPHA_SCISSOR_USED\n"; - actions.usage_defines["ALPHA_HASH_SCALE"] = "#define ALPHA_HASH_USED\n"; - actions.usage_defines["ALPHA_ANTIALIASING_EDGE"] = "#define ALPHA_ANTIALIASING_EDGE_USED\n"; - actions.usage_defines["ALPHA_TEXTURE_COORDINATE"] = "@ALPHA_ANTIALIASING_EDGE"; - - actions.usage_defines["SSS_STRENGTH"] = "#define ENABLE_SSS\n"; - actions.usage_defines["SSS_TRANSMITTANCE_DEPTH"] = "#define ENABLE_TRANSMITTANCE\n"; - actions.usage_defines["BACKLIGHT"] = "#define LIGHT_BACKLIGHT_USED\n"; - actions.usage_defines["SCREEN_TEXTURE"] = "#define SCREEN_TEXTURE_USED\n"; - actions.usage_defines["SCREEN_UV"] = "#define SCREEN_UV_USED\n"; - - actions.usage_defines["DIFFUSE_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; - actions.usage_defines["SPECULAR_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; - - actions.usage_defines["FOG"] = "#define CUSTOM_FOG_USED\n"; - actions.usage_defines["RADIANCE"] = "#define CUSTOM_RADIANCE_USED\n"; - actions.usage_defines["IRRADIANCE"] = "#define CUSTOM_IRRADIANCE_USED\n"; - - actions.render_mode_defines["skip_vertex_transform"] = "#define SKIP_TRANSFORM_USED\n"; - actions.render_mode_defines["world_vertex_coords"] = "#define VERTEX_WORLD_COORDS_USED\n"; - actions.render_mode_defines["ensure_correct_normals"] = "#define ENSURE_CORRECT_NORMALS\n"; - actions.render_mode_defines["cull_front"] = "#define DO_SIDE_CHECK\n"; - actions.render_mode_defines["cull_disabled"] = "#define DO_SIDE_CHECK\n"; - - bool force_lambert = GLOBAL_GET("rendering/quality/shading/force_lambert_over_burley"); - - if (!force_lambert) { - actions.render_mode_defines["diffuse_burley"] = "#define DIFFUSE_BURLEY\n"; - } - - actions.render_mode_defines["diffuse_oren_nayar"] = "#define DIFFUSE_OREN_NAYAR\n"; - actions.render_mode_defines["diffuse_lambert_wrap"] = "#define DIFFUSE_LAMBERT_WRAP\n"; - actions.render_mode_defines["diffuse_toon"] = "#define DIFFUSE_TOON\n"; - - actions.render_mode_defines["sss_mode_skin"] = "#define SSS_MODE_SKIN\n"; - - bool force_blinn = GLOBAL_GET("rendering/quality/shading/force_blinn_over_ggx"); - - if (!force_blinn) { - actions.render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n"; - } else { - actions.render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_BLINN\n"; - } - - actions.render_mode_defines["specular_blinn"] = "#define SPECULAR_BLINN\n"; - actions.render_mode_defines["specular_phong"] = "#define SPECULAR_PHONG\n"; - actions.render_mode_defines["specular_toon"] = "#define SPECULAR_TOON\n"; - actions.render_mode_defines["specular_disabled"] = "#define SPECULAR_DISABLED\n"; - actions.render_mode_defines["shadows_disabled"] = "#define SHADOWS_DISABLED\n"; - actions.render_mode_defines["ambient_light_disabled"] = "#define AMBIENT_LIGHT_DISABLED\n"; - actions.render_mode_defines["shadow_to_opacity"] = "#define USE_SHADOW_TO_OPACITY\n"; - actions.render_mode_defines["unshaded"] = "#define MODE_UNSHADED\n"; - - actions.sampler_array_name = "material_samplers"; - actions.base_texture_binding_index = 1; - actions.texture_layout_set = MATERIAL_UNIFORM_SET; - actions.base_uniform_string = "material."; - actions.base_varying_index = 10; - - actions.default_filter = ShaderLanguage::FILTER_LINEAR_MIPMAP; - actions.default_repeat = ShaderLanguage::REPEAT_ENABLE; - actions.global_buffer_array_variable = "global_variables.data"; - actions.instance_uniform_index_variable = "instances.data[instance_index].instance_uniforms_ofs"; - - shader.compiler.initialize(actions); - } - - //render list - render_list.max_elements = GLOBAL_DEF_RST("rendering/limits/rendering/max_renderable_elements", (int)128000); - render_list.init(); - render_pass = 0; - - scene_state.uniform_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(SceneState::UBO)); - - { - //default material and shader - default_shader = storage->shader_create(); - storage->shader_set_code(default_shader, "shader_type spatial; void vertex() { ROUGHNESS = 0.8; } void fragment() { ALBEDO=vec3(0.6); ROUGHNESS=0.8; METALLIC=0.2; } \n"); - default_material = storage->material_create(); - storage->material_set_shader(default_material, default_shader); - - MaterialData *md = (MaterialData *)storage->material_get_data(default_material, RendererStorageRD::SHADER_TYPE_3D); - default_shader_rd = shader.scene_shader.version_get_shader(md->shader_data->version, SHADER_VERSION_COLOR_PASS); - if (!low_end) { - default_shader_sdfgi_rd = shader.scene_shader.version_get_shader(md->shader_data->version, SHADER_VERSION_DEPTH_PASS_WITH_SDF); - } - } - - { - overdraw_material_shader = storage->shader_create(); - storage->shader_set_code(overdraw_material_shader, "shader_type spatial;\nrender_mode blend_add,unshaded;\n void fragment() { ALBEDO=vec3(0.4,0.8,0.8); ALPHA=0.2; }"); - overdraw_material = storage->material_create(); - storage->material_set_shader(overdraw_material, overdraw_material_shader); - - wireframe_material_shader = storage->shader_create(); - storage->shader_set_code(wireframe_material_shader, "shader_type spatial;\nrender_mode wireframe,unshaded;\n void fragment() { ALBEDO=vec3(0.0,0.0,0.0); }"); - wireframe_material = storage->material_create(); - storage->material_set_shader(wireframe_material, wireframe_material_shader); - } - - { - default_vec4_xform_buffer = RD::get_singleton()->storage_buffer_create(256); - Vector<RD::Uniform> uniforms; - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(default_vec4_xform_buffer); - u.binding = 0; - uniforms.push_back(u); - - default_vec4_xform_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, default_shader_rd, TRANSFORMS_UNIFORM_SET); - } - { - RD::SamplerState sampler; - sampler.mag_filter = RD::SAMPLER_FILTER_LINEAR; - sampler.min_filter = RD::SAMPLER_FILTER_LINEAR; - sampler.enable_compare = true; - sampler.compare_op = RD::COMPARE_OP_LESS; - shadow_sampler = RD::get_singleton()->sampler_create(sampler); + scene_shader.init(p_storage, defines, low_end); } - render_list_thread_threshold = GLOBAL_GET("rendering/forward_renderer/threaded_render_minimum_instances"); + render_list_thread_threshold = GLOBAL_GET("rendering/limits/forward_renderer/threaded_render_minimum_instances"); } -RendererSceneRenderForward::~RendererSceneRenderForward() { +RenderForwardClustered::~RenderForwardClustered() { directional_shadow_atlas_set_size(0); //clear base uniform set if still valid - if (render_pass_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(render_pass_uniform_set)) { - RD::get_singleton()->free(render_pass_uniform_set); + for (uint32_t i = 0; i < render_pass_uniform_sets.size(); i++) { + if (render_pass_uniform_sets[i].is_valid() && RD::get_singleton()->uniform_set_is_valid(render_pass_uniform_sets[i])) { + RD::get_singleton()->free(render_pass_uniform_sets[i]); + } } if (sdfgi_pass_uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(sdfgi_pass_uniform_set)) { RD::get_singleton()->free(sdfgi_pass_uniform_set); } - RD::get_singleton()->free(default_vec4_xform_buffer); - RD::get_singleton()->free(shadow_sampler); - - storage->free(wireframe_material_shader); - storage->free(overdraw_material_shader); - storage->free(default_shader); - - storage->free(wireframe_material); - storage->free(overdraw_material); - storage->free(default_material); - { - RD::get_singleton()->free(scene_state.uniform_buffer); + for (uint32_t i = 0; i < scene_state.uniform_buffers.size(); i++) { + RD::get_singleton()->free(scene_state.uniform_buffers[i]); + } RD::get_singleton()->free(scene_state.lightmap_buffer); RD::get_singleton()->free(scene_state.lightmap_capture_buffer); + for (uint32_t i = 0; i < RENDER_LIST_MAX; i++) { + if (scene_state.instance_buffer[i] != RID()) { + RD::get_singleton()->free(scene_state.instance_buffer[i]); + } + } memdelete_arr(scene_state.lightmap_captures); } diff --git a/servers/rendering/renderer_rd/renderer_scene_render_forward.h b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h index 8a6f268c46..72e84a6f24 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_forward.h +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h @@ -1,5 +1,5 @@ /*************************************************************************/ -/* renderer_scene_render_forward.h */ +/* render_forward_clustered.h */ /*************************************************************************/ /* This file is part of: */ /* GODOT ENGINE */ @@ -28,16 +28,21 @@ /* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ /*************************************************************************/ -#ifndef RENDERING_SERVER_SCENE_RENDER_FORWARD_H -#define RENDERING_SERVER_SCENE_RENDER_FORWARD_H +#ifndef RENDERING_SERVER_SCENE_RENDER_FORWARD_CLUSTERED_H +#define RENDERING_SERVER_SCENE_RENDER_FORWARD_CLUSTERED_H #include "core/templates/paged_allocator.h" +#include "servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h" #include "servers/rendering/renderer_rd/pipeline_cache_rd.h" #include "servers/rendering/renderer_rd/renderer_scene_render_rd.h" #include "servers/rendering/renderer_rd/renderer_storage_rd.h" -#include "servers/rendering/renderer_rd/shaders/scene_forward.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl.gen.h" + +namespace RendererSceneRenderImplementation { + +class RenderForwardClustered : public RendererSceneRenderRD { + friend SceneShaderForwardClustered; -class RendererSceneRenderForward : public RendererSceneRenderRD { enum { SCENE_UNIFORM_SET = 0, RENDER_PASS_UNIFORM_SET = 1, @@ -50,159 +55,24 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { MAX_GI_PROBES = 8, MAX_LIGHTMAPS = 8, MAX_GI_PROBES_PER_INSTANCE = 2, + INSTANCE_DATA_BUFFER_MIN_SIZE = 4096 }; - /* Scene Shader */ + enum RenderListType { + RENDER_LIST_OPAQUE, //used for opaque objects + RENDER_LIST_ALPHA, //used for transparent objects + RENDER_LIST_SECONDARY, //used for shadows and other objects + RENDER_LIST_MAX - enum ShaderVersion { - SHADER_VERSION_DEPTH_PASS, - SHADER_VERSION_DEPTH_PASS_DP, - SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS, - SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE, - SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL, - SHADER_VERSION_DEPTH_PASS_WITH_SDF, - SHADER_VERSION_COLOR_PASS, - SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI, - SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR, - SHADER_VERSION_LIGHTMAP_COLOR_PASS, - SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR, - SHADER_VERSION_MAX }; - struct { - SceneForwardShaderRD scene_shader; - ShaderCompilerRD compiler; - } shader; - - RendererStorageRD *storage; - - /* Material */ - - struct ShaderData : public RendererStorageRD::ShaderData { - enum BlendMode { //used internally - BLEND_MODE_MIX, - BLEND_MODE_ADD, - BLEND_MODE_SUB, - BLEND_MODE_MUL, - BLEND_MODE_ALPHA_TO_COVERAGE - }; - - enum DepthDraw { - DEPTH_DRAW_DISABLED, - DEPTH_DRAW_OPAQUE, - DEPTH_DRAW_ALWAYS - }; - - enum DepthTest { - DEPTH_TEST_DISABLED, - DEPTH_TEST_ENABLED - }; - - enum Cull { - CULL_DISABLED, - CULL_FRONT, - CULL_BACK - }; - - enum CullVariant { - CULL_VARIANT_NORMAL, - CULL_VARIANT_REVERSED, - CULL_VARIANT_DOUBLE_SIDED, - CULL_VARIANT_MAX - - }; - - enum AlphaAntiAliasing { - ALPHA_ANTIALIASING_OFF, - ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE, - ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE - }; - - bool valid; - RID version; - uint32_t vertex_input_mask; - PipelineCacheRD pipelines[CULL_VARIANT_MAX][RS::PRIMITIVE_MAX][SHADER_VERSION_MAX]; - - String path; - - Map<StringName, ShaderLanguage::ShaderNode::Uniform> uniforms; - Vector<ShaderCompilerRD::GeneratedCode::Texture> texture_uniforms; - - Vector<uint32_t> ubo_offsets; - uint32_t ubo_size; - - String code; - Map<StringName, RID> default_texture_params; - - DepthDraw depth_draw; - DepthTest depth_test; - - bool uses_point_size; - bool uses_alpha; - bool uses_blend_alpha; - bool uses_alpha_clip; - bool uses_depth_pre_pass; - bool uses_discard; - bool uses_roughness; - bool uses_normal; - - bool unshaded; - bool uses_vertex; - bool uses_sss; - bool uses_transmittance; - bool uses_screen_texture; - bool uses_depth_texture; - bool uses_normal_texture; - bool uses_time; - bool writes_modelview_or_projection; - bool uses_world_coordinates; - - uint64_t last_pass = 0; - uint32_t index = 0; - - virtual void set_code(const String &p_Code); - virtual void set_default_texture_param(const StringName &p_name, RID p_texture); - virtual void get_param_list(List<PropertyInfo> *p_param_list) const; - void get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const; - - virtual bool is_param_texture(const StringName &p_param) const; - virtual bool is_animated() const; - virtual bool casts_shadows() const; - virtual Variant get_default_parameter(const StringName &p_parameter) const; - ShaderData(); - virtual ~ShaderData(); - }; - - RendererStorageRD::ShaderData *_create_shader_func(); - static RendererStorageRD::ShaderData *_create_shader_funcs() { - return static_cast<RendererSceneRenderForward *>(singleton)->_create_shader_func(); - } - - struct MaterialData : public RendererStorageRD::MaterialData { - uint64_t last_frame; - ShaderData *shader_data; - RID uniform_buffer; - RID uniform_set; - Vector<RID> texture_cache; - Vector<uint8_t> ubo_data; - uint64_t last_pass = 0; - uint32_t index = 0; - RID next_pass; - uint8_t priority; - virtual void set_render_priority(int p_priority); - virtual void set_next_pass(RID p_pass); - virtual void update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty); - virtual ~MaterialData(); - }; + /* Scene Shader */ - RendererStorageRD::MaterialData *_create_material_func(ShaderData *p_shader); - static RendererStorageRD::MaterialData *_create_material_funcs(RendererStorageRD::ShaderData *p_shader) { - return static_cast<RendererSceneRenderForward *>(singleton)->_create_material_func(static_cast<ShaderData *>(p_shader)); - } + SceneShaderForwardClustered scene_shader; /* Framebuffer */ - struct RenderBufferDataForward : public RenderBufferData { + struct RenderBufferDataForwardClustered : public RenderBufferData { //for rendering, may be MSAAd RID color; @@ -211,9 +81,6 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { RID normal_roughness_buffer; RID giprobe_buffer; - RID ambient_buffer; - RID reflection_buffer; - RS::ViewportMSAA msaa; RD::TextureSamples texture_samples; @@ -234,34 +101,81 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { RID render_sdfgi_uniform_set; void ensure_specular(); - void ensure_gi(); void ensure_giprobe(); void clear(); virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa); - ~RenderBufferDataForward(); + ~RenderBufferDataForwardClustered(); }; virtual RenderBufferData *_create_render_buffer_data(); - void _allocate_normal_roughness_texture(RenderBufferDataForward *rb); + void _allocate_normal_roughness_texture(RenderBufferDataForwardClustered *rb); - RID shadow_sampler; RID render_base_uniform_set; - RID render_pass_uniform_set; + LocalVector<RID> render_pass_uniform_sets; RID sdfgi_pass_uniform_set; uint64_t lightmap_texture_array_version = 0xFFFFFFFF; virtual void _base_uniforms_changed(); - void _render_buffers_clear_uniform_set(RenderBufferDataForward *rb); + void _render_buffers_clear_uniform_set(RenderBufferDataForwardClustered *rb); virtual void _render_buffers_uniform_set_changed(RID p_render_buffers); virtual RID _render_buffers_get_normal_texture(RID p_render_buffers); - virtual RID _render_buffers_get_ambient_texture(RID p_render_buffers); - virtual RID _render_buffers_get_reflection_texture(RID p_render_buffers); void _update_render_base_uniform_set(); RID _setup_sdfgi_render_pass_uniform_set(RID p_albedo_texture, RID p_emission_texture, RID p_emission_aniso_texture, RID p_geom_facing_texture); - RID _setup_render_pass_uniform_set(RID p_render_buffers, RID p_radiance_texture, RID p_shadow_atlas, RID p_reflection_atlas, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps); + RID _setup_render_pass_uniform_set(RenderListType p_render_list, RID p_render_buffers, RID p_radiance_texture, RID p_shadow_atlas, RID p_reflection_atlas, RID p_cluster_buffer, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, bool p_use_directional_shadow_atlas = false, int p_index = 0); + + enum PassMode { + PASS_MODE_COLOR, + PASS_MODE_COLOR_SPECULAR, + PASS_MODE_COLOR_TRANSPARENT, + PASS_MODE_SHADOW, + PASS_MODE_SHADOW_DP, + PASS_MODE_DEPTH, + PASS_MODE_DEPTH_NORMAL_ROUGHNESS, + PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE, + PASS_MODE_DEPTH_MATERIAL, + PASS_MODE_SDF, + }; + + struct GeometryInstanceSurfaceDataCache; + struct RenderElementInfo; + + struct RenderListParameters { + GeometryInstanceSurfaceDataCache **elements = nullptr; + RenderElementInfo *element_info = nullptr; + int element_count = 0; + bool reverse_cull = false; + PassMode pass_mode = PASS_MODE_COLOR; + bool no_gi = false; + RID render_pass_uniform_set; + bool force_wireframe = false; + Vector2 uv_offset; + Plane lod_plane; + float lod_distance_multiplier = 0.0; + float screen_lod_threshold = 0.0; + RD::FramebufferFormatID framebuffer_format = 0; + uint32_t element_offset = 0; + uint32_t barrier = RD::BARRIER_MASK_ALL; + + RenderListParameters(GeometryInstanceSurfaceDataCache **p_elements, RenderElementInfo *p_element_info, int p_element_count, bool p_reverse_cull, PassMode p_pass_mode, bool p_no_gi, RID p_render_pass_uniform_set, bool p_force_wireframe = false, const Vector2 &p_uv_offset = Vector2(), const Plane &p_lod_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, uint32_t p_element_offset = 0, uint32_t p_barrier = RD::BARRIER_MASK_ALL) { + elements = p_elements; + element_info = p_element_info; + element_count = p_element_count; + reverse_cull = p_reverse_cull; + pass_mode = p_pass_mode; + no_gi = p_no_gi; + render_pass_uniform_set = p_render_pass_uniform_set; + force_wireframe = p_force_wireframe; + uv_offset = p_uv_offset; + lod_plane = p_lod_plane; + lod_distance_multiplier = p_lod_distance_multiplier; + screen_lod_threshold = p_screen_lod_threshold; + element_offset = p_element_offset; + barrier = p_barrier; + } + }; struct LightmapData { float normal_xform[12]; @@ -298,6 +212,11 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { float viewport_size[2]; float screen_pixel_size[2]; + uint32_t cluster_shift; + uint32_t cluster_width; + uint32_t cluster_type_size; + uint32_t max_cluster_element_count_div_32; + float directional_penumbra_shadow_kernel[128]; //32 vec4s float directional_soft_shadow_kernel[128]; float penumbra_shadow_kernel[128]; @@ -366,9 +285,24 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { uint32_t pancake_shadows; }; + struct PushConstant { + uint32_t base_index; // + uint32_t uv_offset; //packed + uint32_t pad[2]; + }; + + struct InstanceData { + float transform[16]; + uint32_t flags; + uint32_t instance_uniforms_ofs; //base offset in global buffer for instance variables + uint32_t gi_offset; //GI information when using lightmapping (VCT or lightmap index) + uint32_t layer_mask; + float lightmap_uv_scale[4]; + }; + UBO ubo; - RID uniform_buffer; + LocalVector<RID> uniform_buffers; LightmapData lightmaps[MAX_LIGHTMAPS]; RID lightmap_ids[MAX_LIGHTMAPS]; @@ -377,6 +311,10 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { uint32_t max_lightmaps; RID lightmap_buffer; + RID instance_buffer[RENDER_LIST_MAX]; + uint32_t instance_buffer_size[RENDER_LIST_MAX] = { 0, 0, 0 }; + LocalVector<InstanceData> instance_data[RENDER_LIST_MAX]; + LightmapCaptureData *lightmap_captures; uint32_t max_lightmap_captures; RID lightmap_capture_buffer; @@ -389,68 +327,38 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { bool used_depth_texture = false; bool used_sss = false; - } scene_state; + struct ShadowPass { + uint32_t element_from; + uint32_t element_count; + bool flip_cull; + PassMode pass_mode; + + RID rp_uniform_set; + Plane camera_plane; + float lod_distance_multiplier; + float screen_lod_threshold; + + RID framebuffer; + RD::InitialAction initial_depth_action; + RD::FinalAction final_depth_action; + Rect2i rect; + }; - static RendererSceneRenderForward *singleton; - uint64_t render_pass; - double time; - RID default_shader; - RID default_material; - RID overdraw_material_shader; - RID overdraw_material; - RID wireframe_material_shader; - RID wireframe_material; - RID default_shader_rd; - RID default_shader_sdfgi_rd; - - RID default_vec4_xform_buffer; - RID default_vec4_xform_uniform_set; + LocalVector<ShadowPass> shadow_passes; - enum PassMode { - PASS_MODE_COLOR, - PASS_MODE_COLOR_SPECULAR, - PASS_MODE_COLOR_TRANSPARENT, - PASS_MODE_SHADOW, - PASS_MODE_SHADOW_DP, - PASS_MODE_DEPTH, - PASS_MODE_DEPTH_NORMAL_ROUGHNESS, - PASS_MODE_DEPTH_NORMAL_ROUGHNESS_GIPROBE, - PASS_MODE_DEPTH_MATERIAL, - PASS_MODE_SDF, - }; + } scene_state; + + static RenderForwardClustered *singleton; - void _setup_environment(RID p_environment, RID p_render_buffers, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_reflection_probe, bool p_no_fog, const Size2 &p_screen_pixel_size, RID p_shadow_atlas, bool p_flip_y, const Color &p_default_bg_color, float p_znear, float p_zfar, bool p_opaque_render_buffers = false, bool p_pancake_shadows = false); + void _setup_environment(RID p_environment, RID p_render_buffers, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_reflection_probe, bool p_no_fog, const Size2i &p_screen_size, uint32_t p_cluster_size, uint32_t p_max_cluster_elements, RID p_shadow_atlas, bool p_flip_y, const Color &p_default_bg_color, float p_znear, float p_zfar, bool p_opaque_render_buffers = false, bool p_pancake_shadows = false, int p_index = 0); void _setup_giprobes(const PagedArray<RID> &p_giprobes); void _setup_lightmaps(const PagedArray<RID> &p_lightmaps, const Transform &p_cam_transform); - struct GeometryInstanceSurfaceDataCache; - - struct RenderListParameters { - GeometryInstanceSurfaceDataCache **elements = nullptr; - int element_count = 0; - bool reverse_cull = false; - PassMode pass_mode = PASS_MODE_COLOR; - bool no_gi = false; - RID render_pass_uniform_set; - bool force_wireframe = false; - Vector2 uv_offset; - Plane lod_plane; - float lod_distance_multiplier = 0.0; - float screen_lod_threshold = 0.0; - RD::FramebufferFormatID framebuffer_format = 0; - RenderListParameters(GeometryInstanceSurfaceDataCache **p_elements, int p_element_count, bool p_reverse_cull, PassMode p_pass_mode, bool p_no_gi, RID p_render_pass_uniform_set, bool p_force_wireframe = false, const Vector2 &p_uv_offset = Vector2(), const Plane &p_lod_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0) { - elements = p_elements; - element_count = p_element_count; - reverse_cull = p_reverse_cull; - pass_mode = p_pass_mode; - no_gi = p_no_gi; - render_pass_uniform_set = p_render_pass_uniform_set; - force_wireframe = p_force_wireframe; - uv_offset = p_uv_offset; - lod_plane = p_lod_plane; - lod_distance_multiplier = p_lod_distance_multiplier; - screen_lod_threshold = p_screen_lod_threshold; - } + struct RenderElementInfo { + uint32_t repeat : 22; + uint32_t uses_lightmap : 1; + uint32_t uses_forward_gi : 1; + uint32_t lod_index : 8; }; template <PassMode p_pass_mode> @@ -464,12 +372,14 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { uint32_t render_list_thread_threshold = 500; - void _fill_render_list(const PagedArray<GeometryInstance *> &p_instances, PassMode p_pass_mode, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, bool p_using_sdfgi = false, bool p_using_opaque_gi = false); + void _update_instance_data_buffer(RenderListType p_render_list); + void _fill_instance_data(RenderListType p_render_list, uint32_t p_offset = 0, int32_t p_max_elements = -1, bool p_update_buffer = true); + void _fill_render_list(RenderListType p_render_list, const PagedArray<GeometryInstance *> &p_instances, PassMode p_pass_mode, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, bool p_using_sdfgi = false, bool p_using_opaque_gi = false, const Plane &p_lod_camera_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, bool p_append = false); Map<Size2i, RID> sdfgi_framebuffer_size_cache; struct GeometryInstanceData; - struct GeometryInstanceForward; + struct GeometryInstanceForwardClustered; struct GeometryInstanceLightmapSH { Color sh[9]; @@ -492,14 +402,17 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { union { struct { - uint32_t geometry_id; - uint32_t material_id; - uint32_t shader_id; - uint32_t surface_type : 4; - uint32_t uses_forward_gi : 1; //set during addition - uint32_t uses_lightmap : 1; //set during addition - uint32_t depth_layer : 4; //set during addition - uint32_t priority : 8; + uint64_t lod_index : 8; + uint64_t surface_index : 10; + uint64_t geometry_id : 32; + uint64_t material_id_low : 14; + + uint64_t material_id_hi : 18; + uint64_t shader_id : 32; + uint64_t uses_forward_gi : 1; + uint64_t uses_lightmap : 1; + uint64_t depth_layer : 4; + uint64_t priority : 8; }; struct { uint64_t sort_key1; @@ -513,17 +426,17 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { void *surface = nullptr; RID material_uniform_set; - ShaderData *shader = nullptr; + SceneShaderForwardClustered::ShaderData *shader = nullptr; void *surface_shadow = nullptr; RID material_uniform_set_shadow; - ShaderData *shader_shadow = nullptr; + SceneShaderForwardClustered::ShaderData *shader_shadow = nullptr; GeometryInstanceSurfaceDataCache *next = nullptr; - GeometryInstanceForward *owner = nullptr; + GeometryInstanceForwardClustered *owner = nullptr; }; - struct GeometryInstanceForward : public GeometryInstance { + struct GeometryInstanceForwardClustered : public GeometryInstance { //used during rendering bool mirror = false; bool non_uniform_scale = false; @@ -531,25 +444,25 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { float lod_model_scale = 1.0; AABB transformed_aabb; //needed for LOD float depth = 0; - struct PushConstant { - float transform[16]; - uint32_t flags; - uint32_t instance_uniforms_ofs; //base offset in global buffer for instance variables - uint32_t gi_offset; //GI information when using lightmapping (VCT or lightmap index) - uint32_t layer_mask; - float lightmap_uv_scale[4]; - } push_constant; + uint32_t gi_offset_cache = 0; + uint32_t flags_cache = 0; + bool store_transform_cache = true; + int32_t shader_parameters_offset = -1; + uint32_t lightmap_slice_index; + Rect2 lightmap_uv_scale; + uint32_t layer_mask = 1; RID transforms_uniform_set; uint32_t instance_count = 0; RID mesh_instance; bool can_sdfgi = false; //used during setup uint32_t base_flags = 0; + Transform transform; RID gi_probes[MAX_GI_PROBES_PER_INSTANCE]; RID lightmap_instance; GeometryInstanceLightmapSH *lightmap_sh = nullptr; GeometryInstanceSurfaceDataCache *surface_caches = nullptr; - SelfList<GeometryInstanceForward> dirty_list_element; + SelfList<GeometryInstanceForwardClustered> dirty_list_element; struct Data { //data used less often goes into regular heap @@ -557,21 +470,14 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { RS::InstanceType base_type; RID skeleton; - - uint32_t layer_mask = 1; - Vector<RID> surface_materials; RID material_override; - Transform transform; AABB aabb; - int32_t shader_parameters_offset = -1; bool use_dynamic_gi = false; bool use_baked_light = false; - bool cast_double_sided_shaodows = false; + bool cast_double_sided_shadows = false; bool mirror = false; - Rect2 lightmap_uv_scale; - uint32_t lightmap_slice_index = 0; bool dirty_dependencies = false; RendererStorage::DependencyTracker dependency_tracker; @@ -579,40 +485,34 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { Data *data = nullptr; - GeometryInstanceForward() : + GeometryInstanceForwardClustered() : dirty_list_element(this) {} }; static void _geometry_instance_dependency_changed(RendererStorage::DependencyChangedNotification p_notification, RendererStorage::DependencyTracker *p_tracker); static void _geometry_instance_dependency_deleted(const RID &p_dependency, RendererStorage::DependencyTracker *p_tracker); - SelfList<GeometryInstanceForward>::List geometry_instance_dirty_list; + SelfList<GeometryInstanceForwardClustered>::List geometry_instance_dirty_list; - PagedAllocator<GeometryInstanceForward> geometry_instance_alloc; + PagedAllocator<GeometryInstanceForwardClustered> geometry_instance_alloc; PagedAllocator<GeometryInstanceSurfaceDataCache> geometry_instance_surface_alloc; PagedAllocator<GeometryInstanceLightmapSH> geometry_instance_lightmap_sh; - void _geometry_instance_add_surface_with_material(GeometryInstanceForward *ginstance, uint32_t p_surface, MaterialData *p_material, uint32_t p_material_id, uint32_t p_shader_id, RID p_mesh); - void _geometry_instance_add_surface(GeometryInstanceForward *ginstance, uint32_t p_surface, RID p_material, RID p_mesh); + void _geometry_instance_add_surface_with_material(GeometryInstanceForwardClustered *ginstance, uint32_t p_surface, SceneShaderForwardClustered::MaterialData *p_material, uint32_t p_material_id, uint32_t p_shader_id, RID p_mesh); + void _geometry_instance_add_surface(GeometryInstanceForwardClustered *ginstance, uint32_t p_surface, RID p_material, RID p_mesh); void _geometry_instance_mark_dirty(GeometryInstance *p_geometry_instance); void _geometry_instance_update(GeometryInstance *p_geometry_instance); void _update_dirty_geometry_instances(); - bool low_end = false; - /* Render List */ struct RenderList { - int max_elements; - - GeometryInstanceSurfaceDataCache **elements = nullptr; - - int element_count; - int alpha_element_count; + LocalVector<GeometryInstanceSurfaceDataCache *> elements; + LocalVector<RenderElementInfo> element_info; void clear() { - element_count = 0; - alpha_element_count = 0; + elements.clear(); + element_info.clear(); } //should eventually be replaced by radix @@ -623,13 +523,14 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { } }; - void sort_by_key(bool p_alpha) { + void sort_by_key() { SortArray<GeometryInstanceSurfaceDataCache *, SortByKey> sorter; - if (p_alpha) { - sorter.sort(&elements[max_elements - alpha_element_count], alpha_element_count); - } else { - sorter.sort(elements, element_count); - } + sorter.sort(elements.ptr(), elements.size()); + } + + void sort_by_key_range(uint32_t p_from, uint32_t p_size) { + SortArray<GeometryInstanceSurfaceDataCache *, SortByKey> sorter; + sorter.sort(elements.ptr() + p_from, p_size); } struct SortByDepth { @@ -638,14 +539,10 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { } }; - void sort_by_depth(bool p_alpha) { //used for shadows + void sort_by_depth() { //used for shadows SortArray<GeometryInstanceSurfaceDataCache *, SortByDepth> sorter; - if (p_alpha) { - sorter.sort(&elements[max_elements - alpha_element_count], alpha_element_count); - } else { - sorter.sort(elements, element_count); - } + sorter.sort(elements.ptr(), elements.size()); } struct SortByReverseDepthAndPriority { @@ -657,50 +554,24 @@ class RendererSceneRenderForward : public RendererSceneRenderRD { void sort_by_reverse_depth_and_priority(bool p_alpha) { //used for alpha SortArray<GeometryInstanceSurfaceDataCache *, SortByReverseDepthAndPriority> sorter; - if (p_alpha) { - sorter.sort(&elements[max_elements - alpha_element_count], alpha_element_count); - } else { - sorter.sort(elements, element_count); - } + sorter.sort(elements.ptr(), elements.size()); } _FORCE_INLINE_ void add_element(GeometryInstanceSurfaceDataCache *p_element) { - if (element_count + alpha_element_count >= max_elements) { - return; - } - elements[element_count] = p_element; - element_count++; - } - - _FORCE_INLINE_ void add_alpha_element(GeometryInstanceSurfaceDataCache *p_element) { - if (element_count + alpha_element_count >= max_elements) { - return; - } - int idx = max_elements - alpha_element_count - 1; - elements[idx] = p_element; - alpha_element_count++; - } - - void init() { - element_count = 0; - alpha_element_count = 0; - elements = memnew_arr(GeometryInstanceSurfaceDataCache *, max_elements); - } - - RenderList() { - max_elements = 0; - } - - ~RenderList() { - memdelete_arr(elements); + elements.push_back(p_element); } }; - RenderList render_list; + RenderList render_list[RENDER_LIST_MAX]; protected: - virtual void _render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, int p_directional_light_count, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_bg_color, float p_lod_threshold); - virtual void _render_shadow(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool p_use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0); + virtual void _render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_cluster_buffer, uint32_t p_cluster_size, uint32_t p_max_cluster_elements, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_bg_color, float p_lod_threshold); + + virtual void _render_shadow_begin(); + virtual void _render_shadow_append(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool p_use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, const Rect2i &p_rect = Rect2i(), bool p_flip_y = false, bool p_clear_region = true, bool p_begin = true, bool p_end = true); + virtual void _render_shadow_process(); + virtual void _render_shadow_end(uint32_t p_barrier = RD::BARRIER_MASK_ALL); + virtual void _render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region); virtual void _render_uv2(const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region); virtual void _render_sdfgi(RID p_render_buffers, const Vector3i &p_from, const Vector3i &p_size, const AABB &p_bounds, const PagedArray<GeometryInstance *> &p_instances, const RID &p_albedo_texture, const RID &p_emission_texture, const RID &p_emission_aniso_texture, const RID &p_geom_facing_texture); @@ -733,11 +604,10 @@ public: virtual void geometry_instance_pair_decal_instances(GeometryInstance *p_geometry_instance, const RID *p_decal_instances, uint32_t p_decal_instance_count); virtual void geometry_instance_pair_gi_probe_instances(GeometryInstance *p_geometry_instance, const RID *p_gi_probe_instances, uint32_t p_gi_probe_instance_count); - virtual void set_time(double p_time, double p_step); - virtual bool free(RID p_rid); - RendererSceneRenderForward(RendererStorageRD *p_storage); - ~RendererSceneRenderForward(); + RenderForwardClustered(RendererStorageRD *p_storage); + ~RenderForwardClustered(); }; -#endif // RASTERIZER_SCENE_HIGHEND_RD_H +} // namespace RendererSceneRenderImplementation +#endif // !RENDERING_SERVER_SCENE_RENDER_FORWARD_CLUSTERED_H diff --git a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp new file mode 100644 index 0000000000..cb8c6e0cf3 --- /dev/null +++ b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.cpp @@ -0,0 +1,807 @@ +/*************************************************************************/ +/* scene_shader_forward_clustered.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "scene_shader_forward_clustered.h" +#include "core/config/project_settings.h" +#include "render_forward_clustered.h" + +using namespace RendererSceneRenderImplementation; + +void SceneShaderForwardClustered::ShaderData::set_code(const String &p_code) { + //compile + + code = p_code; + valid = false; + ubo_size = 0; + uniforms.clear(); + uses_screen_texture = false; + + if (code == String()) { + return; //just invalid, but no error + } + + ShaderCompilerRD::GeneratedCode gen_code; + + int blend_mode = BLEND_MODE_MIX; + int depth_testi = DEPTH_TEST_ENABLED; + int alpha_antialiasing_mode = ALPHA_ANTIALIASING_OFF; + int cull = CULL_BACK; + + uses_point_size = false; + uses_alpha = false; + uses_blend_alpha = false; + uses_depth_pre_pass = false; + uses_discard = false; + uses_roughness = false; + uses_normal = false; + bool wireframe = false; + + unshaded = false; + uses_vertex = false; + uses_sss = false; + uses_transmittance = false; + uses_screen_texture = false; + uses_depth_texture = false; + uses_normal_texture = false; + uses_time = false; + writes_modelview_or_projection = false; + uses_world_coordinates = false; + + int depth_drawi = DEPTH_DRAW_OPAQUE; + + ShaderCompilerRD::IdentifierActions actions; + + actions.render_mode_values["blend_add"] = Pair<int *, int>(&blend_mode, BLEND_MODE_ADD); + actions.render_mode_values["blend_mix"] = Pair<int *, int>(&blend_mode, BLEND_MODE_MIX); + actions.render_mode_values["blend_sub"] = Pair<int *, int>(&blend_mode, BLEND_MODE_SUB); + actions.render_mode_values["blend_mul"] = Pair<int *, int>(&blend_mode, BLEND_MODE_MUL); + + actions.render_mode_values["alpha_to_coverage"] = Pair<int *, int>(&alpha_antialiasing_mode, ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE); + actions.render_mode_values["alpha_to_coverage_and_one"] = Pair<int *, int>(&alpha_antialiasing_mode, ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE); + + actions.render_mode_values["depth_draw_never"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_DISABLED); + actions.render_mode_values["depth_draw_opaque"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_OPAQUE); + actions.render_mode_values["depth_draw_always"] = Pair<int *, int>(&depth_drawi, DEPTH_DRAW_ALWAYS); + + actions.render_mode_values["depth_test_disabled"] = Pair<int *, int>(&depth_testi, DEPTH_TEST_DISABLED); + + actions.render_mode_values["cull_disabled"] = Pair<int *, int>(&cull, CULL_DISABLED); + actions.render_mode_values["cull_front"] = Pair<int *, int>(&cull, CULL_FRONT); + actions.render_mode_values["cull_back"] = Pair<int *, int>(&cull, CULL_BACK); + + actions.render_mode_flags["unshaded"] = &unshaded; + actions.render_mode_flags["wireframe"] = &wireframe; + + actions.usage_flag_pointers["ALPHA"] = &uses_alpha; + actions.render_mode_flags["depth_prepass_alpha"] = &uses_depth_pre_pass; + + actions.usage_flag_pointers["SSS_STRENGTH"] = &uses_sss; + actions.usage_flag_pointers["SSS_TRANSMITTANCE_DEPTH"] = &uses_transmittance; + + actions.usage_flag_pointers["SCREEN_TEXTURE"] = &uses_screen_texture; + actions.usage_flag_pointers["DEPTH_TEXTURE"] = &uses_depth_texture; + actions.usage_flag_pointers["NORMAL_TEXTURE"] = &uses_normal_texture; + actions.usage_flag_pointers["DISCARD"] = &uses_discard; + actions.usage_flag_pointers["TIME"] = &uses_time; + actions.usage_flag_pointers["ROUGHNESS"] = &uses_roughness; + actions.usage_flag_pointers["NORMAL"] = &uses_normal; + actions.usage_flag_pointers["NORMAL_MAP"] = &uses_normal; + + actions.usage_flag_pointers["POINT_SIZE"] = &uses_point_size; + actions.usage_flag_pointers["POINT_COORD"] = &uses_point_size; + + actions.write_flag_pointers["MODELVIEW_MATRIX"] = &writes_modelview_or_projection; + actions.write_flag_pointers["PROJECTION_MATRIX"] = &writes_modelview_or_projection; + actions.write_flag_pointers["VERTEX"] = &uses_vertex; + + actions.uniforms = &uniforms; + + SceneShaderForwardClustered *shader_singleton = (SceneShaderForwardClustered *)SceneShaderForwardClustered::singleton; + Error err = shader_singleton->compiler.compile(RS::SHADER_SPATIAL, code, &actions, path, gen_code); + + ERR_FAIL_COND(err != OK); + + if (version.is_null()) { + version = shader_singleton->shader.version_create(); + } + + depth_draw = DepthDraw(depth_drawi); + depth_test = DepthTest(depth_testi); + +#if 0 + print_line("**compiling shader:"); + print_line("**defines:\n"); + for (int i = 0; i < gen_code.defines.size(); i++) { + print_line(gen_code.defines[i]); + } + print_line("\n**uniforms:\n" + gen_code.uniforms); + print_line("\n**vertex_globals:\n" + gen_code.vertex_global); + print_line("\n**vertex_code:\n" + gen_code.vertex); + print_line("\n**fragment_globals:\n" + gen_code.fragment_global); + print_line("\n**fragment_code:\n" + gen_code.fragment); + print_line("\n**light_code:\n" + gen_code.light); +#endif + shader_singleton->shader.version_set_code(version, gen_code.uniforms, gen_code.vertex_global, gen_code.vertex, gen_code.fragment_global, gen_code.light, gen_code.fragment, gen_code.defines); + ERR_FAIL_COND(!shader_singleton->shader.version_is_valid(version)); + + ubo_size = gen_code.uniform_total_size; + ubo_offsets = gen_code.uniform_offsets; + texture_uniforms = gen_code.texture_uniforms; + + //blend modes + + // if any form of Alpha Antialiasing is enabled, set the blend mode to alpha to coverage + if (alpha_antialiasing_mode != ALPHA_ANTIALIASING_OFF) { + blend_mode = BLEND_MODE_ALPHA_TO_COVERAGE; + } + + RD::PipelineColorBlendState::Attachment blend_attachment; + + switch (blend_mode) { + case BLEND_MODE_MIX: { + blend_attachment.enable_blend = true; + blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; + blend_attachment.color_blend_op = RD::BLEND_OP_ADD; + blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; + blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_ONE; + blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; + + } break; + case BLEND_MODE_ADD: { + blend_attachment.enable_blend = true; + blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; + blend_attachment.color_blend_op = RD::BLEND_OP_ADD; + blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE; + blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE; + uses_blend_alpha = true; //force alpha used because of blend + + } break; + case BLEND_MODE_SUB: { + blend_attachment.enable_blend = true; + blend_attachment.alpha_blend_op = RD::BLEND_OP_SUBTRACT; + blend_attachment.color_blend_op = RD::BLEND_OP_SUBTRACT; + blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE; + blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ONE; + uses_blend_alpha = true; //force alpha used because of blend + + } break; + case BLEND_MODE_MUL: { + blend_attachment.enable_blend = true; + blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; + blend_attachment.color_blend_op = RD::BLEND_OP_ADD; + blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_DST_COLOR; + blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ZERO; + blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_DST_ALPHA; + blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ZERO; + uses_blend_alpha = true; //force alpha used because of blend + } break; + case BLEND_MODE_ALPHA_TO_COVERAGE: { + blend_attachment.enable_blend = true; + blend_attachment.alpha_blend_op = RD::BLEND_OP_ADD; + blend_attachment.color_blend_op = RD::BLEND_OP_ADD; + blend_attachment.src_color_blend_factor = RD::BLEND_FACTOR_SRC_ALPHA; + blend_attachment.dst_color_blend_factor = RD::BLEND_FACTOR_ONE_MINUS_SRC_ALPHA; + blend_attachment.src_alpha_blend_factor = RD::BLEND_FACTOR_ONE; + blend_attachment.dst_alpha_blend_factor = RD::BLEND_FACTOR_ZERO; + } + } + + RD::PipelineColorBlendState blend_state_blend; + blend_state_blend.attachments.push_back(blend_attachment); + RD::PipelineColorBlendState blend_state_opaque = RD::PipelineColorBlendState::create_disabled(1); + RD::PipelineColorBlendState blend_state_opaque_specular = RD::PipelineColorBlendState::create_disabled(2); + RD::PipelineColorBlendState blend_state_depth_normal_roughness = RD::PipelineColorBlendState::create_disabled(1); + RD::PipelineColorBlendState blend_state_depth_normal_roughness_giprobe = RD::PipelineColorBlendState::create_disabled(2); + + //update pipelines + + RD::PipelineDepthStencilState depth_stencil_state; + + if (depth_test != DEPTH_TEST_DISABLED) { + depth_stencil_state.enable_depth_test = true; + depth_stencil_state.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; + depth_stencil_state.enable_depth_write = depth_draw != DEPTH_DRAW_DISABLED ? true : false; + } + + for (int i = 0; i < CULL_VARIANT_MAX; i++) { + RD::PolygonCullMode cull_mode_rd_table[CULL_VARIANT_MAX][3] = { + { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_FRONT, RD::POLYGON_CULL_BACK }, + { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_BACK, RD::POLYGON_CULL_FRONT }, + { RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_DISABLED, RD::POLYGON_CULL_DISABLED } + }; + + RD::PolygonCullMode cull_mode_rd = cull_mode_rd_table[i][cull]; + + for (int j = 0; j < RS::PRIMITIVE_MAX; j++) { + RD::RenderPrimitive primitive_rd_table[RS::PRIMITIVE_MAX] = { + RD::RENDER_PRIMITIVE_POINTS, + RD::RENDER_PRIMITIVE_LINES, + RD::RENDER_PRIMITIVE_LINESTRIPS, + RD::RENDER_PRIMITIVE_TRIANGLES, + RD::RENDER_PRIMITIVE_TRIANGLE_STRIPS, + }; + + RD::RenderPrimitive primitive_rd = uses_point_size ? RD::RENDER_PRIMITIVE_POINTS : primitive_rd_table[j]; + + for (int k = 0; k < SHADER_VERSION_MAX; k++) { + if (!static_cast<SceneShaderForwardClustered *>(singleton)->shader.is_variant_enabled(k)) { + continue; + } + RD::PipelineRasterizationState raster_state; + raster_state.cull_mode = cull_mode_rd; + raster_state.wireframe = wireframe; + + RD::PipelineColorBlendState blend_state; + RD::PipelineDepthStencilState depth_stencil = depth_stencil_state; + RD::PipelineMultisampleState multisample_state; + + if (uses_alpha || uses_blend_alpha) { + // only allow these flags to go through if we have some form of msaa + if (alpha_antialiasing_mode == ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE) { + multisample_state.enable_alpha_to_coverage = true; + } else if (alpha_antialiasing_mode == ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE) { + multisample_state.enable_alpha_to_coverage = true; + multisample_state.enable_alpha_to_one = true; + } + + if (k == SHADER_VERSION_COLOR_PASS || k == SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI || k == SHADER_VERSION_LIGHTMAP_COLOR_PASS) { + blend_state = blend_state_blend; + if (depth_draw == DEPTH_DRAW_OPAQUE) { + depth_stencil.enable_depth_write = false; //alpha does not draw depth + } + } else if (uses_depth_pre_pass && (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP || k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS || k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL)) { + if (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP) { + //none, blend state contains nothing + } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL) { + blend_state = RD::PipelineColorBlendState::create_disabled(5); //writes to normal and roughness in opaque way + } else { + blend_state = blend_state_opaque; //writes to normal and roughness in opaque way + } + } else { + pipelines[i][j][k].clear(); + continue; // do not use this version (will error if using it is attempted) + } + } else { + if (k == SHADER_VERSION_COLOR_PASS || k == SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI || k == SHADER_VERSION_LIGHTMAP_COLOR_PASS) { + blend_state = blend_state_opaque; + } else if (k == SHADER_VERSION_DEPTH_PASS || k == SHADER_VERSION_DEPTH_PASS_DP) { + //none, leave empty + } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS) { + blend_state = blend_state_depth_normal_roughness; + } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE) { + blend_state = blend_state_depth_normal_roughness_giprobe; + } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL) { + blend_state = RD::PipelineColorBlendState::create_disabled(5); //writes to normal and roughness in opaque way + } else if (k == SHADER_VERSION_DEPTH_PASS_WITH_SDF) { + blend_state = RD::PipelineColorBlendState(); //no color targets for SDF + } else { + //specular write + blend_state = blend_state_opaque_specular; + depth_stencil.enable_depth_test = false; + depth_stencil.enable_depth_write = false; + } + } + + RID shader_variant = shader_singleton->shader.version_get_shader(version, k); + pipelines[i][j][k].setup(shader_variant, primitive_rd, raster_state, multisample_state, depth_stencil, blend_state, 0); + } + } + } + + valid = true; +} + +void SceneShaderForwardClustered::ShaderData::set_default_texture_param(const StringName &p_name, RID p_texture) { + if (!p_texture.is_valid()) { + default_texture_params.erase(p_name); + } else { + default_texture_params[p_name] = p_texture; + } +} + +void SceneShaderForwardClustered::ShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { + Map<int, StringName> order; + + for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { + if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_LOCAL) { + continue; + } + + if (E->get().texture_order >= 0) { + order[E->get().texture_order + 100000] = E->key(); + } else { + order[E->get().order] = E->key(); + } + } + + for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); + pi.name = E->get(); + p_param_list->push_back(pi); + } +} + +void SceneShaderForwardClustered::ShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { + for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { + if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + continue; + } + + RendererStorage::InstanceShaderParam p; + p.info = ShaderLanguage::uniform_to_property_info(E->get()); + p.info.name = E->key(); //supply name + p.index = E->get().instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p_param_list->push_back(p); + } +} + +bool SceneShaderForwardClustered::ShaderData::is_param_texture(const StringName &p_param) const { + if (!uniforms.has(p_param)) { + return false; + } + + return uniforms[p_param].texture_order >= 0; +} + +bool SceneShaderForwardClustered::ShaderData::is_animated() const { + return false; +} + +bool SceneShaderForwardClustered::ShaderData::casts_shadows() const { + return false; +} + +Variant SceneShaderForwardClustered::ShaderData::get_default_parameter(const StringName &p_parameter) const { + if (uniforms.has(p_parameter)) { + ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; + Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + } + return Variant(); +} + +RS::ShaderNativeSourceCode SceneShaderForwardClustered::ShaderData::get_native_source_code() const { + SceneShaderForwardClustered *shader_singleton = (SceneShaderForwardClustered *)SceneShaderForwardClustered::singleton; + + return shader_singleton->shader.version_get_native_source_code(version); +} + +SceneShaderForwardClustered::ShaderData::ShaderData() { + valid = false; + uses_screen_texture = false; +} + +SceneShaderForwardClustered::ShaderData::~ShaderData() { + SceneShaderForwardClustered *shader_singleton = (SceneShaderForwardClustered *)SceneShaderForwardClustered::singleton; + ERR_FAIL_COND(!shader_singleton); + //pipeline variants will clear themselves if shader is gone + if (version.is_valid()) { + shader_singleton->shader.version_free(version); + } +} + +RendererStorageRD::ShaderData *SceneShaderForwardClustered::_create_shader_func() { + ShaderData *shader_data = memnew(ShaderData); + return shader_data; +} + +void SceneShaderForwardClustered::MaterialData::set_render_priority(int p_priority) { + priority = p_priority - RS::MATERIAL_RENDER_PRIORITY_MIN; //8 bits +} + +void SceneShaderForwardClustered::MaterialData::set_next_pass(RID p_pass) { + next_pass = p_pass; +} + +void SceneShaderForwardClustered::MaterialData::update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty) { + SceneShaderForwardClustered *shader_singleton = (SceneShaderForwardClustered *)SceneShaderForwardClustered::singleton; + + if ((uint32_t)ubo_data.size() != shader_data->ubo_size) { + p_uniform_dirty = true; + if (uniform_buffer.is_valid()) { + RD::get_singleton()->free(uniform_buffer); + uniform_buffer = RID(); + } + + ubo_data.resize(shader_data->ubo_size); + if (ubo_data.size()) { + uniform_buffer = RD::get_singleton()->uniform_buffer_create(ubo_data.size()); + memset(ubo_data.ptrw(), 0, ubo_data.size()); //clear + } + + //clear previous uniform set + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + uniform_set = RID(); + } + } + + //check whether buffer changed + if (p_uniform_dirty && ubo_data.size()) { + update_uniform_buffer(shader_data->uniforms, shader_data->ubo_offsets.ptr(), p_parameters, ubo_data.ptrw(), ubo_data.size(), false); + RD::get_singleton()->buffer_update(uniform_buffer, 0, ubo_data.size(), ubo_data.ptrw(), RD::BARRIER_MASK_RASTER); + } + + uint32_t tex_uniform_count = shader_data->texture_uniforms.size(); + + if ((uint32_t)texture_cache.size() != tex_uniform_count) { + texture_cache.resize(tex_uniform_count); + p_textures_dirty = true; + + //clear previous uniform set + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + uniform_set = RID(); + } + } + + if (p_textures_dirty && tex_uniform_count) { + update_textures(p_parameters, shader_data->default_texture_params, shader_data->texture_uniforms, texture_cache.ptrw(), true); + } + + if (shader_data->ubo_size == 0 && shader_data->texture_uniforms.size() == 0) { + // This material does not require an uniform set, so don't create it. + return; + } + + if (!p_textures_dirty && uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + //no reason to update uniform set, only UBO (or nothing) was needed to update + return; + } + + Vector<RD::Uniform> uniforms; + + { + if (shader_data->ubo_size) { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 0; + u.ids.push_back(uniform_buffer); + uniforms.push_back(u); + } + + const RID *textures = texture_cache.ptrw(); + for (uint32_t i = 0; i < tex_uniform_count; i++) { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 1 + i; + u.ids.push_back(textures[i]); + uniforms.push_back(u); + } + } + + uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shader_singleton->shader.version_get_shader(shader_data->version, 0), RenderForwardClustered::MATERIAL_UNIFORM_SET); +} + +SceneShaderForwardClustered::MaterialData::~MaterialData() { + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + } + + if (uniform_buffer.is_valid()) { + RD::get_singleton()->free(uniform_buffer); + } +} + +RendererStorageRD::MaterialData *SceneShaderForwardClustered::_create_material_func(ShaderData *p_shader) { + MaterialData *material_data = memnew(MaterialData); + material_data->shader_data = p_shader; + material_data->last_frame = false; + //update will happen later anyway so do nothing. + return material_data; +} + +SceneShaderForwardClustered *SceneShaderForwardClustered::singleton = nullptr; + +SceneShaderForwardClustered::SceneShaderForwardClustered() { + // there should be only one of these, contained within our RenderFM singleton. + singleton = this; +} + +SceneShaderForwardClustered::~SceneShaderForwardClustered() { + RD::get_singleton()->free(default_vec4_xform_buffer); + RD::get_singleton()->free(shadow_sampler); + + storage->free(wireframe_material_shader); + storage->free(overdraw_material_shader); + storage->free(default_shader); + + storage->free(wireframe_material); + storage->free(overdraw_material); + storage->free(default_material); +} + +void SceneShaderForwardClustered::init(RendererStorageRD *p_storage, const String p_defines, bool p_is_low_end) { + storage = p_storage; + + { + Vector<String> shader_versions; + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n"); + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_DUAL_PARABOLOID\n"); + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_NORMAL_ROUGHNESS\n"); + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_NORMAL_ROUGHNESS\n#define MODE_RENDER_GIPROBE\n"); + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_MATERIAL\n"); + shader_versions.push_back("\n#define MODE_RENDER_DEPTH\n#define MODE_RENDER_SDF\n"); + shader_versions.push_back(""); + shader_versions.push_back("\n#define USE_FORWARD_GI\n"); + shader_versions.push_back("\n#define MODE_MULTIPLE_RENDER_TARGETS\n"); + shader_versions.push_back("\n#define USE_LIGHTMAP\n"); + shader_versions.push_back("\n#define MODE_MULTIPLE_RENDER_TARGETS\n#define USE_LIGHTMAP\n"); + shader.initialize(shader_versions, p_defines); + + if (p_is_low_end) { + //disable the high end versions + shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS, false); + shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE, false); + shader.set_variant_enabled(SHADER_VERSION_DEPTH_PASS_WITH_SDF, false); + shader.set_variant_enabled(SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI, false); + shader.set_variant_enabled(SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR, false); + shader.set_variant_enabled(SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR, false); + } + } + + storage->shader_set_data_request_function(RendererStorageRD::SHADER_TYPE_3D, _create_shader_funcs); + storage->material_set_data_request_function(RendererStorageRD::SHADER_TYPE_3D, _create_material_funcs); + + { + //shader compiler + ShaderCompilerRD::DefaultIdentifierActions actions; + + actions.renames["WORLD_MATRIX"] = "world_matrix"; + actions.renames["WORLD_NORMAL_MATRIX"] = "world_normal_matrix"; + actions.renames["INV_CAMERA_MATRIX"] = "scene_data.inv_camera_matrix"; + actions.renames["CAMERA_MATRIX"] = "scene_data.camera_matrix"; + actions.renames["PROJECTION_MATRIX"] = "projection_matrix"; + actions.renames["INV_PROJECTION_MATRIX"] = "scene_data.inv_projection_matrix"; + actions.renames["MODELVIEW_MATRIX"] = "modelview"; + actions.renames["MODELVIEW_NORMAL_MATRIX"] = "modelview_normal"; + + actions.renames["VERTEX"] = "vertex"; + actions.renames["NORMAL"] = "normal"; + actions.renames["TANGENT"] = "tangent"; + actions.renames["BINORMAL"] = "binormal"; + actions.renames["POSITION"] = "position"; + actions.renames["UV"] = "uv_interp"; + actions.renames["UV2"] = "uv2_interp"; + actions.renames["COLOR"] = "color_interp"; + actions.renames["POINT_SIZE"] = "gl_PointSize"; + actions.renames["INSTANCE_ID"] = "gl_InstanceIndex"; + + actions.renames["ALPHA_SCISSOR_THRESHOLD"] = "alpha_scissor_threshold"; + actions.renames["ALPHA_HASH_SCALE"] = "alpha_hash_scale"; + actions.renames["ALPHA_ANTIALIASING_EDGE"] = "alpha_antialiasing_edge"; + actions.renames["ALPHA_TEXTURE_COORDINATE"] = "alpha_texture_coordinate"; + + //builtins + + actions.renames["TIME"] = "scene_data.time"; + actions.renames["VIEWPORT_SIZE"] = "scene_data.viewport_size"; + + actions.renames["FRAGCOORD"] = "gl_FragCoord"; + actions.renames["FRONT_FACING"] = "gl_FrontFacing"; + actions.renames["NORMAL_MAP"] = "normal_map"; + actions.renames["NORMAL_MAP_DEPTH"] = "normal_map_depth"; + actions.renames["ALBEDO"] = "albedo"; + actions.renames["ALPHA"] = "alpha"; + actions.renames["METALLIC"] = "metallic"; + actions.renames["SPECULAR"] = "specular"; + actions.renames["ROUGHNESS"] = "roughness"; + actions.renames["RIM"] = "rim"; + actions.renames["RIM_TINT"] = "rim_tint"; + actions.renames["CLEARCOAT"] = "clearcoat"; + actions.renames["CLEARCOAT_GLOSS"] = "clearcoat_gloss"; + actions.renames["ANISOTROPY"] = "anisotropy"; + actions.renames["ANISOTROPY_FLOW"] = "anisotropy_flow"; + actions.renames["SSS_STRENGTH"] = "sss_strength"; + actions.renames["SSS_TRANSMITTANCE_COLOR"] = "transmittance_color"; + actions.renames["SSS_TRANSMITTANCE_DEPTH"] = "transmittance_depth"; + actions.renames["SSS_TRANSMITTANCE_CURVE"] = "transmittance_curve"; + actions.renames["SSS_TRANSMITTANCE_BOOST"] = "transmittance_boost"; + actions.renames["BACKLIGHT"] = "backlight"; + actions.renames["AO"] = "ao"; + actions.renames["AO_LIGHT_AFFECT"] = "ao_light_affect"; + actions.renames["EMISSION"] = "emission"; + actions.renames["POINT_COORD"] = "gl_PointCoord"; + actions.renames["INSTANCE_CUSTOM"] = "instance_custom"; + actions.renames["SCREEN_UV"] = "screen_uv"; + actions.renames["SCREEN_TEXTURE"] = "color_buffer"; + actions.renames["DEPTH_TEXTURE"] = "depth_buffer"; + actions.renames["NORMAL_ROUGHNESS_TEXTURE"] = "normal_roughness_buffer"; + actions.renames["DEPTH"] = "gl_FragDepth"; + actions.renames["OUTPUT_IS_SRGB"] = "true"; + actions.renames["FOG"] = "custom_fog"; + actions.renames["RADIANCE"] = "custom_radiance"; + actions.renames["IRRADIANCE"] = "custom_irradiance"; + actions.renames["BONE_INDICES"] = "bone_attrib"; + actions.renames["BONE_WEIGHTS"] = "weight_attrib"; + actions.renames["CUSTOM0"] = "custom0_attrib"; + actions.renames["CUSTOM1"] = "custom1_attrib"; + actions.renames["CUSTOM2"] = "custom2_attrib"; + actions.renames["CUSTOM3"] = "custom3_attrib"; + + //for light + actions.renames["VIEW"] = "view"; + actions.renames["LIGHT_COLOR"] = "light_color"; + actions.renames["LIGHT"] = "light"; + actions.renames["ATTENUATION"] = "attenuation"; + actions.renames["SHADOW_ATTENUATION"] = "shadow_attenuation"; + actions.renames["DIFFUSE_LIGHT"] = "diffuse_light"; + actions.renames["SPECULAR_LIGHT"] = "specular_light"; + + actions.usage_defines["NORMAL"] = "#define NORMAL_USED\n"; + actions.usage_defines["TANGENT"] = "#define TANGENT_USED\n"; + actions.usage_defines["BINORMAL"] = "@TANGENT"; + actions.usage_defines["RIM"] = "#define LIGHT_RIM_USED\n"; + actions.usage_defines["RIM_TINT"] = "@RIM"; + actions.usage_defines["CLEARCOAT"] = "#define LIGHT_CLEARCOAT_USED\n"; + actions.usage_defines["CLEARCOAT_GLOSS"] = "@CLEARCOAT"; + actions.usage_defines["ANISOTROPY"] = "#define LIGHT_ANISOTROPY_USED\n"; + actions.usage_defines["ANISOTROPY_FLOW"] = "@ANISOTROPY"; + actions.usage_defines["AO"] = "#define AO_USED\n"; + actions.usage_defines["AO_LIGHT_AFFECT"] = "#define AO_USED\n"; + actions.usage_defines["UV"] = "#define UV_USED\n"; + actions.usage_defines["UV2"] = "#define UV2_USED\n"; + actions.usage_defines["BONE_INDICES"] = "#define BONES_USED\n"; + actions.usage_defines["BONE_WEIGHTS"] = "#define WEIGHTS_USED\n"; + actions.usage_defines["CUSTOM0"] = "#define CUSTOM0\n"; + actions.usage_defines["CUSTOM1"] = "#define CUSTOM1\n"; + actions.usage_defines["CUSTOM2"] = "#define CUSTOM2\n"; + actions.usage_defines["CUSTOM3"] = "#define CUSTOM3\n"; + actions.usage_defines["NORMAL_MAP"] = "#define NORMAL_MAP_USED\n"; + actions.usage_defines["NORMAL_MAP_DEPTH"] = "@NORMAL_MAP"; + actions.usage_defines["COLOR"] = "#define COLOR_USED\n"; + actions.usage_defines["INSTANCE_CUSTOM"] = "#define ENABLE_INSTANCE_CUSTOM\n"; + actions.usage_defines["POSITION"] = "#define OVERRIDE_POSITION\n"; + + actions.usage_defines["ALPHA_SCISSOR_THRESHOLD"] = "#define ALPHA_SCISSOR_USED\n"; + actions.usage_defines["ALPHA_HASH_SCALE"] = "#define ALPHA_HASH_USED\n"; + actions.usage_defines["ALPHA_ANTIALIASING_EDGE"] = "#define ALPHA_ANTIALIASING_EDGE_USED\n"; + actions.usage_defines["ALPHA_TEXTURE_COORDINATE"] = "@ALPHA_ANTIALIASING_EDGE"; + + actions.usage_defines["SSS_STRENGTH"] = "#define ENABLE_SSS\n"; + actions.usage_defines["SSS_TRANSMITTANCE_DEPTH"] = "#define ENABLE_TRANSMITTANCE\n"; + actions.usage_defines["BACKLIGHT"] = "#define LIGHT_BACKLIGHT_USED\n"; + actions.usage_defines["SCREEN_TEXTURE"] = "#define SCREEN_TEXTURE_USED\n"; + actions.usage_defines["SCREEN_UV"] = "#define SCREEN_UV_USED\n"; + + actions.usage_defines["DIFFUSE_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; + actions.usage_defines["SPECULAR_LIGHT"] = "#define USE_LIGHT_SHADER_CODE\n"; + + actions.usage_defines["FOG"] = "#define CUSTOM_FOG_USED\n"; + actions.usage_defines["RADIANCE"] = "#define CUSTOM_RADIANCE_USED\n"; + actions.usage_defines["IRRADIANCE"] = "#define CUSTOM_IRRADIANCE_USED\n"; + + actions.render_mode_defines["skip_vertex_transform"] = "#define SKIP_TRANSFORM_USED\n"; + actions.render_mode_defines["world_vertex_coords"] = "#define VERTEX_WORLD_COORDS_USED\n"; + actions.render_mode_defines["ensure_correct_normals"] = "#define ENSURE_CORRECT_NORMALS\n"; + actions.render_mode_defines["cull_front"] = "#define DO_SIDE_CHECK\n"; + actions.render_mode_defines["cull_disabled"] = "#define DO_SIDE_CHECK\n"; + + bool force_lambert = GLOBAL_GET("rendering/shading/overrides/force_lambert_over_burley"); + + if (!force_lambert) { + actions.render_mode_defines["diffuse_burley"] = "#define DIFFUSE_BURLEY\n"; + } + + actions.render_mode_defines["diffuse_oren_nayar"] = "#define DIFFUSE_OREN_NAYAR\n"; + actions.render_mode_defines["diffuse_lambert_wrap"] = "#define DIFFUSE_LAMBERT_WRAP\n"; + actions.render_mode_defines["diffuse_toon"] = "#define DIFFUSE_TOON\n"; + + actions.render_mode_defines["sss_mode_skin"] = "#define SSS_MODE_SKIN\n"; + + bool force_blinn = GLOBAL_GET("rendering/shading/overrides/force_blinn_over_ggx"); + + if (!force_blinn) { + actions.render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n"; + } else { + actions.render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_BLINN\n"; + } + + actions.render_mode_defines["specular_blinn"] = "#define SPECULAR_BLINN\n"; + actions.render_mode_defines["specular_phong"] = "#define SPECULAR_PHONG\n"; + actions.render_mode_defines["specular_toon"] = "#define SPECULAR_TOON\n"; + actions.render_mode_defines["specular_disabled"] = "#define SPECULAR_DISABLED\n"; + actions.render_mode_defines["shadows_disabled"] = "#define SHADOWS_DISABLED\n"; + actions.render_mode_defines["ambient_light_disabled"] = "#define AMBIENT_LIGHT_DISABLED\n"; + actions.render_mode_defines["shadow_to_opacity"] = "#define USE_SHADOW_TO_OPACITY\n"; + actions.render_mode_defines["unshaded"] = "#define MODE_UNSHADED\n"; + + actions.sampler_array_name = "material_samplers"; + actions.base_texture_binding_index = 1; + actions.texture_layout_set = RenderForwardClustered::MATERIAL_UNIFORM_SET; + actions.base_uniform_string = "material."; + actions.base_varying_index = 10; + + actions.default_filter = ShaderLanguage::FILTER_LINEAR_MIPMAP; + actions.default_repeat = ShaderLanguage::REPEAT_ENABLE; + actions.global_buffer_array_variable = "global_variables.data"; + actions.instance_uniform_index_variable = "draw_call.instance_uniforms_ofs"; + + compiler.initialize(actions); + } + + { + //default material and shader + default_shader = storage->shader_allocate(); + storage->shader_initialize(default_shader); + storage->shader_set_code(default_shader, "shader_type spatial; void vertex() { ROUGHNESS = 0.8; } void fragment() { ALBEDO=vec3(0.6); ROUGHNESS=0.8; METALLIC=0.2; } \n"); + default_material = storage->material_allocate(); + storage->material_initialize(default_material); + storage->material_set_shader(default_material, default_shader); + + MaterialData *md = (MaterialData *)storage->material_get_data(default_material, RendererStorageRD::SHADER_TYPE_3D); + default_shader_rd = shader.version_get_shader(md->shader_data->version, SHADER_VERSION_COLOR_PASS); + if (!p_is_low_end) { + default_shader_sdfgi_rd = shader.version_get_shader(md->shader_data->version, SHADER_VERSION_DEPTH_PASS_WITH_SDF); + } + } + + { + overdraw_material_shader = storage->shader_allocate(); + storage->shader_initialize(overdraw_material_shader); + storage->shader_set_code(overdraw_material_shader, "shader_type spatial;\nrender_mode blend_add,unshaded;\n void fragment() { ALBEDO=vec3(0.4,0.8,0.8); ALPHA=0.2; }"); + overdraw_material = storage->material_allocate(); + storage->material_initialize(overdraw_material); + storage->material_set_shader(overdraw_material, overdraw_material_shader); + + wireframe_material_shader = storage->shader_allocate(); + storage->shader_initialize(wireframe_material_shader); + storage->shader_set_code(wireframe_material_shader, "shader_type spatial;\nrender_mode wireframe,unshaded;\n void fragment() { ALBEDO=vec3(0.0,0.0,0.0); }"); + wireframe_material = storage->material_allocate(); + storage->material_initialize(wireframe_material); + storage->material_set_shader(wireframe_material, wireframe_material_shader); + } + + { + default_vec4_xform_buffer = RD::get_singleton()->storage_buffer_create(256); + Vector<RD::Uniform> uniforms; + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(default_vec4_xform_buffer); + u.binding = 0; + uniforms.push_back(u); + + default_vec4_xform_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, default_shader_rd, RenderForwardClustered::TRANSFORMS_UNIFORM_SET); + } + { + RD::SamplerState sampler; + sampler.mag_filter = RD::SAMPLER_FILTER_LINEAR; + sampler.min_filter = RD::SAMPLER_FILTER_LINEAR; + sampler.enable_compare = true; + sampler.compare_op = RD::COMPARE_OP_LESS; + shadow_sampler = RD::get_singleton()->sampler_create(sampler); + } +} diff --git a/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h new file mode 100644 index 0000000000..368340e258 --- /dev/null +++ b/servers/rendering/renderer_rd/forward_clustered/scene_shader_forward_clustered.h @@ -0,0 +1,210 @@ +/*************************************************************************/ +/* scene_shader_forward_clustered.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef RSSR_SCENE_SHADER_FC_H +#define RSSR_SCENE_SHADER_FC_H + +#include "servers/rendering/renderer_rd/renderer_scene_render_rd.h" +#include "servers/rendering/renderer_rd/renderer_storage_rd.h" +#include "servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl.gen.h" + +namespace RendererSceneRenderImplementation { + +class SceneShaderForwardClustered { +private: + static SceneShaderForwardClustered *singleton; + +public: + RendererStorageRD *storage; + + enum ShaderVersion { + SHADER_VERSION_DEPTH_PASS, + SHADER_VERSION_DEPTH_PASS_DP, + SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS, + SHADER_VERSION_DEPTH_PASS_WITH_NORMAL_AND_ROUGHNESS_AND_GIPROBE, + SHADER_VERSION_DEPTH_PASS_WITH_MATERIAL, + SHADER_VERSION_DEPTH_PASS_WITH_SDF, + SHADER_VERSION_COLOR_PASS, + SHADER_VERSION_COLOR_PASS_WITH_FORWARD_GI, + SHADER_VERSION_COLOR_PASS_WITH_SEPARATE_SPECULAR, + SHADER_VERSION_LIGHTMAP_COLOR_PASS, + SHADER_VERSION_LIGHTMAP_COLOR_PASS_WITH_SEPARATE_SPECULAR, + SHADER_VERSION_MAX + }; + + struct ShaderData : public RendererStorageRD::ShaderData { + enum BlendMode { //used internally + BLEND_MODE_MIX, + BLEND_MODE_ADD, + BLEND_MODE_SUB, + BLEND_MODE_MUL, + BLEND_MODE_ALPHA_TO_COVERAGE + }; + + enum DepthDraw { + DEPTH_DRAW_DISABLED, + DEPTH_DRAW_OPAQUE, + DEPTH_DRAW_ALWAYS + }; + + enum DepthTest { + DEPTH_TEST_DISABLED, + DEPTH_TEST_ENABLED + }; + + enum Cull { + CULL_DISABLED, + CULL_FRONT, + CULL_BACK + }; + + enum CullVariant { + CULL_VARIANT_NORMAL, + CULL_VARIANT_REVERSED, + CULL_VARIANT_DOUBLE_SIDED, + CULL_VARIANT_MAX + + }; + + enum AlphaAntiAliasing { + ALPHA_ANTIALIASING_OFF, + ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE, + ALPHA_ANTIALIASING_ALPHA_TO_COVERAGE_AND_TO_ONE + }; + + bool valid; + RID version; + uint32_t vertex_input_mask; + PipelineCacheRD pipelines[CULL_VARIANT_MAX][RS::PRIMITIVE_MAX][SHADER_VERSION_MAX]; + + String path; + + Map<StringName, ShaderLanguage::ShaderNode::Uniform> uniforms; + Vector<ShaderCompilerRD::GeneratedCode::Texture> texture_uniforms; + + Vector<uint32_t> ubo_offsets; + uint32_t ubo_size; + + String code; + Map<StringName, RID> default_texture_params; + + DepthDraw depth_draw; + DepthTest depth_test; + + bool uses_point_size; + bool uses_alpha; + bool uses_blend_alpha; + bool uses_alpha_clip; + bool uses_depth_pre_pass; + bool uses_discard; + bool uses_roughness; + bool uses_normal; + + bool unshaded; + bool uses_vertex; + bool uses_sss; + bool uses_transmittance; + bool uses_screen_texture; + bool uses_depth_texture; + bool uses_normal_texture; + bool uses_time; + bool writes_modelview_or_projection; + bool uses_world_coordinates; + + uint64_t last_pass = 0; + uint32_t index = 0; + + virtual void set_code(const String &p_Code); + virtual void set_default_texture_param(const StringName &p_name, RID p_texture); + virtual void get_param_list(List<PropertyInfo> *p_param_list) const; + void get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const; + + virtual bool is_param_texture(const StringName &p_param) const; + virtual bool is_animated() const; + virtual bool casts_shadows() const; + virtual Variant get_default_parameter(const StringName &p_parameter) const; + virtual RS::ShaderNativeSourceCode get_native_source_code() const; + + ShaderData(); + virtual ~ShaderData(); + }; + + RendererStorageRD::ShaderData *_create_shader_func(); + static RendererStorageRD::ShaderData *_create_shader_funcs() { + return static_cast<SceneShaderForwardClustered *>(singleton)->_create_shader_func(); + } + + struct MaterialData : public RendererStorageRD::MaterialData { + uint64_t last_frame; + ShaderData *shader_data; + RID uniform_buffer; + RID uniform_set; + Vector<RID> texture_cache; + Vector<uint8_t> ubo_data; + uint64_t last_pass = 0; + uint32_t index = 0; + RID next_pass; + uint8_t priority; + virtual void set_render_priority(int p_priority); + virtual void set_next_pass(RID p_pass); + virtual void update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty); + virtual ~MaterialData(); + }; + + RendererStorageRD::MaterialData *_create_material_func(ShaderData *p_shader); + static RendererStorageRD::MaterialData *_create_material_funcs(RendererStorageRD::ShaderData *p_shader) { + return static_cast<SceneShaderForwardClustered *>(singleton)->_create_material_func(static_cast<ShaderData *>(p_shader)); + } + + SceneForwardClusteredShaderRD shader; + ShaderCompilerRD compiler; + + RID default_shader; + RID default_material; + RID overdraw_material_shader; + RID overdraw_material; + RID wireframe_material_shader; + RID wireframe_material; + RID default_shader_rd; + RID default_shader_sdfgi_rd; + + RID default_vec4_xform_buffer; + RID default_vec4_xform_uniform_set; + + RID shadow_sampler; + + SceneShaderForwardClustered(); + ~SceneShaderForwardClustered(); + + void init(RendererStorageRD *p_storage, const String p_defines, bool p_is_low_end); +}; + +} // namespace RendererSceneRenderImplementation +#endif // !RSSR_SCENE_SHADER_FM_H diff --git a/servers/rendering/renderer_rd/light_cluster_builder.cpp b/servers/rendering/renderer_rd/light_cluster_builder.cpp deleted file mode 100644 index bb807ca4ca..0000000000 --- a/servers/rendering/renderer_rd/light_cluster_builder.cpp +++ /dev/null @@ -1,252 +0,0 @@ -/*************************************************************************/ -/* light_cluster_builder.cpp */ -/*************************************************************************/ -/* This file is part of: */ -/* GODOT ENGINE */ -/* https://godotengine.org */ -/*************************************************************************/ -/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ -/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ -/* */ -/* Permission is hereby granted, free of charge, to any person obtaining */ -/* a copy of this software and associated documentation files (the */ -/* "Software"), to deal in the Software without restriction, including */ -/* without limitation the rights to use, copy, modify, merge, publish, */ -/* distribute, sublicense, and/or sell copies of the Software, and to */ -/* permit persons to whom the Software is furnished to do so, subject to */ -/* the following conditions: */ -/* */ -/* The above copyright notice and this permission notice shall be */ -/* included in all copies or substantial portions of the Software. */ -/* */ -/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ -/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ -/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ -/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ -/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ -/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ -/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ -/*************************************************************************/ - -#include "light_cluster_builder.h" - -void LightClusterBuilder::begin(const Transform &p_view_transform, const CameraMatrix &p_cam_projection) { - view_xform = p_view_transform; - projection = p_cam_projection; - z_near = -projection.get_z_near(); - z_far = -projection.get_z_far(); - - //reset counts - light_count = 0; - refprobe_count = 0; - decal_count = 0; - item_count = 0; - sort_id_count = 0; -} - -void LightClusterBuilder::bake_cluster() { - float slice_depth = (z_near - z_far) / depth; - - uint8_t *cluster_dataw = cluster_data.ptrw(); - Cell *cluster_data_ptr = (Cell *)cluster_dataw; - //clear the cluster - zeromem(cluster_data_ptr, (width * height * depth * sizeof(Cell))); - - /* Step 1, create cell positions and count them */ - - for (uint32_t i = 0; i < item_count; i++) { - const Item &item = items[i]; - - int from_slice = Math::floor((z_near - (item.aabb.position.z + item.aabb.size.z)) / slice_depth); - int to_slice = Math::floor((z_near - item.aabb.position.z) / slice_depth); - - if (from_slice >= (int)depth || to_slice < 0) { - continue; //sorry no go - } - - from_slice = MAX(0, from_slice); - to_slice = MIN((int)depth - 1, to_slice); - - for (int j = from_slice; j <= to_slice; j++) { - Vector3 min = item.aabb.position; - Vector3 max = item.aabb.position + item.aabb.size; - - float limit_near = MIN((z_near - slice_depth * j), max.z); - float limit_far = MAX((z_near - slice_depth * (j + 1)), min.z); - - max.z = limit_near; - min.z = limit_near; - - Vector3 proj_min = projection.xform(min); - Vector3 proj_max = projection.xform(max); - - int near_from_x = int(Math::floor((proj_min.x * 0.5 + 0.5) * width)); - int near_from_y = int(Math::floor((-proj_max.y * 0.5 + 0.5) * height)); - int near_to_x = int(Math::floor((proj_max.x * 0.5 + 0.5) * width)); - int near_to_y = int(Math::floor((-proj_min.y * 0.5 + 0.5) * height)); - - max.z = limit_far; - min.z = limit_far; - - proj_min = projection.xform(min); - proj_max = projection.xform(max); - - int far_from_x = int(Math::floor((proj_min.x * 0.5 + 0.5) * width)); - int far_from_y = int(Math::floor((-proj_max.y * 0.5 + 0.5) * height)); - int far_to_x = int(Math::floor((proj_max.x * 0.5 + 0.5) * width)); - int far_to_y = int(Math::floor((-proj_min.y * 0.5 + 0.5) * height)); - - //print_line(itos(j) + " near - " + Vector2i(near_from_x, near_from_y) + " -> " + Vector2i(near_to_x, near_to_y)); - //print_line(itos(j) + " far - " + Vector2i(far_from_x, far_from_y) + " -> " + Vector2i(far_to_x, far_to_y)); - - int from_x = MIN(near_from_x, far_from_x); - int from_y = MIN(near_from_y, far_from_y); - int to_x = MAX(near_to_x, far_to_x); - int to_y = MAX(near_to_y, far_to_y); - - if (from_x >= (int)width || to_x < 0 || from_y >= (int)height || to_y < 0) { - continue; - } - - int sx = MAX(0, from_x); - int sy = MAX(0, from_y); - int dx = MIN((int)width - 1, to_x); - int dy = MIN((int)height - 1, to_y); - - //print_line(itos(j) + " - " + Vector2i(sx, sy) + " -> " + Vector2i(dx, dy)); - - for (int x = sx; x <= dx; x++) { - for (int y = sy; y <= dy; y++) { - uint32_t offset = j * (width * height) + y * width + x; - - if (unlikely(sort_id_count == sort_id_max)) { - sort_id_max = nearest_power_of_2_templated(sort_id_max + 1); - sort_ids = (SortID *)memrealloc(sort_ids, sizeof(SortID) * sort_id_max); - if (ids.size()) { - ids.resize(sort_id_max); - RD::get_singleton()->free(items_buffer); - items_buffer = RD::get_singleton()->storage_buffer_create(sizeof(uint32_t) * sort_id_max); - } - } - - sort_ids[sort_id_count].cell_index = offset; - sort_ids[sort_id_count].item_index = item.index; - sort_ids[sort_id_count].item_type = item.type; - - sort_id_count++; - - //for now, only count - cluster_data_ptr[offset].item_pointers[item.type]++; - //print_line("at offset " + itos(offset) + " value: " + itos(cluster_data_ptr[offset].item_pointers[item.type])); - } - } - } - } - - /* Step 2, Assign pointers (and reset counters) */ - - uint32_t offset = 0; - for (uint32_t i = 0; i < (width * height * depth); i++) { - for (int j = 0; j < ITEM_TYPE_MAX; j++) { - uint32_t count = cluster_data_ptr[i].item_pointers[j]; //save count - cluster_data_ptr[i].item_pointers[j] = offset; //replace count by pointer - offset += count; //increase offset by count; - } - } - - //print_line("offset: " + itos(offset)); - /* Step 3, Place item lists */ - - uint32_t *ids_ptr = ids.ptrw(); - - for (uint32_t i = 0; i < sort_id_count; i++) { - const SortID &id = sort_ids[i]; - Cell &cell = cluster_data_ptr[id.cell_index]; - uint32_t pointer = cell.item_pointers[id.item_type] & POINTER_MASK; - uint32_t counter = cell.item_pointers[id.item_type] >> COUNTER_SHIFT; - ids_ptr[pointer + counter] = id.item_index; - - cell.item_pointers[id.item_type] = pointer | ((counter + 1) << COUNTER_SHIFT); - } - - RD::get_singleton()->texture_update(cluster_texture, 0, cluster_data, true); - RD::get_singleton()->buffer_update(items_buffer, 0, offset * sizeof(uint32_t), ids_ptr, true); -} - -void LightClusterBuilder::setup(uint32_t p_width, uint32_t p_height, uint32_t p_depth) { - if (width == p_width && height == p_height && depth == p_depth) { - return; - } - if (cluster_texture.is_valid()) { - RD::get_singleton()->free(cluster_texture); - } - - width = p_width; - height = p_height; - depth = p_depth; - - cluster_data.resize(width * height * depth * sizeof(Cell)); - - { - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R32G32B32A32_UINT; - tf.texture_type = RD::TEXTURE_TYPE_3D; - tf.width = width; - tf.height = height; - tf.depth = depth; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_UPDATE_BIT; - - cluster_texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); - } -} - -RID LightClusterBuilder::get_cluster_texture() const { - return cluster_texture; -} - -RID LightClusterBuilder::get_cluster_indices_buffer() const { - return items_buffer; -} - -LightClusterBuilder::LightClusterBuilder() { - //initialize accumulators to something - lights = (LightData *)memalloc(sizeof(LightData) * 1024); - light_max = 1024; - - refprobes = (OrientedBoxData *)memalloc(sizeof(OrientedBoxData) * 1024); - refprobe_max = 1024; - - decals = (OrientedBoxData *)memalloc(sizeof(OrientedBoxData) * 1024); - decal_max = 1024; - - items = (Item *)memalloc(sizeof(Item) * 1024); - item_max = 1024; - - sort_ids = (SortID *)memalloc(sizeof(SortID) * 1024); - ids.resize(2014); - items_buffer = RD::get_singleton()->storage_buffer_create(sizeof(uint32_t) * 1024); - item_max = 1024; -} - -LightClusterBuilder::~LightClusterBuilder() { - if (cluster_data.size()) { - RD::get_singleton()->free(cluster_texture); - } - - if (lights) { - memfree(lights); - } - if (refprobes) { - memfree(refprobes); - } - if (decals) { - memfree(decals); - } - if (items) { - memfree(items); - } - if (sort_ids) { - memfree(sort_ids); - RD::get_singleton()->free(items_buffer); - } -} diff --git a/servers/rendering/renderer_rd/light_cluster_builder.h b/servers/rendering/renderer_rd/light_cluster_builder.h deleted file mode 100644 index 8f77ece6f5..0000000000 --- a/servers/rendering/renderer_rd/light_cluster_builder.h +++ /dev/null @@ -1,290 +0,0 @@ -/*************************************************************************/ -/* light_cluster_builder.h */ -/*************************************************************************/ -/* This file is part of: */ -/* GODOT ENGINE */ -/* https://godotengine.org */ -/*************************************************************************/ -/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ -/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ -/* */ -/* Permission is hereby granted, free of charge, to any person obtaining */ -/* a copy of this software and associated documentation files (the */ -/* "Software"), to deal in the Software without restriction, including */ -/* without limitation the rights to use, copy, modify, merge, publish, */ -/* distribute, sublicense, and/or sell copies of the Software, and to */ -/* permit persons to whom the Software is furnished to do so, subject to */ -/* the following conditions: */ -/* */ -/* The above copyright notice and this permission notice shall be */ -/* included in all copies or substantial portions of the Software. */ -/* */ -/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ -/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ -/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ -/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ -/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ -/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ -/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ -/*************************************************************************/ - -#ifndef LIGHT_CLUSTER_BUILDER_H -#define LIGHT_CLUSTER_BUILDER_H - -#include "servers/rendering/renderer_rd/renderer_storage_rd.h" - -class LightClusterBuilder { -public: - enum LightType { - LIGHT_TYPE_OMNI, - LIGHT_TYPE_SPOT - }; - - enum ItemType { - ITEM_TYPE_OMNI_LIGHT, - ITEM_TYPE_SPOT_LIGHT, - ITEM_TYPE_REFLECTION_PROBE, - ITEM_TYPE_DECAL, - ITEM_TYPE_MAX //should always be 4 - }; - - enum { - COUNTER_SHIFT = 20, //one million total ids - POINTER_MASK = (1 << COUNTER_SHIFT) - 1, - COUNTER_MASK = 0xfff // 4096 items per cell - }; - -private: - struct LightData { - float position[3]; - uint32_t type; - float radius; - float spot_aperture; - uint32_t pad[2]; - }; - - uint32_t light_count = 0; - uint32_t light_max = 0; - LightData *lights = nullptr; - - struct OrientedBoxData { - float position[3]; - uint32_t pad; - float x_axis[3]; - uint32_t pad2; - float y_axis[3]; - uint32_t pad3; - float z_axis[3]; - uint32_t pad4; - }; - - uint32_t refprobe_count = 0; - uint32_t refprobe_max = 0; - OrientedBoxData *refprobes = nullptr; - - uint32_t decal_count = 0; - uint32_t decal_max = 0; - OrientedBoxData *decals = nullptr; - - struct Item { - AABB aabb; - ItemType type; - uint32_t index; - }; - - Item *items = nullptr; - uint32_t item_count = 0; - uint32_t item_max = 0; - - uint32_t width = 0; - uint32_t height = 0; - uint32_t depth = 0; - - struct Cell { - uint32_t item_pointers[ITEM_TYPE_MAX]; - }; - - Vector<uint8_t> cluster_data; - RID cluster_texture; - - struct SortID { - uint32_t cell_index; - uint32_t item_index; - ItemType item_type; - }; - - SortID *sort_ids = nullptr; - Vector<uint32_t> ids; - uint32_t sort_id_count = 0; - uint32_t sort_id_max = 0; - RID items_buffer; - - Transform view_xform; - CameraMatrix projection; - float z_far = 0; - float z_near = 0; - - _FORCE_INLINE_ void _add_item(const AABB &p_aabb, ItemType p_type, uint32_t p_index) { - if (unlikely(item_count == item_max)) { - item_max = nearest_power_of_2_templated(item_max + 1); - items = (Item *)memrealloc(items, sizeof(Item) * item_max); - } - - Item &item = items[item_count]; - item.aabb = p_aabb; - item.index = p_index; - item.type = p_type; - item_count++; - } - -public: - void begin(const Transform &p_view_transform, const CameraMatrix &p_cam_projection); - - _FORCE_INLINE_ void add_light(LightType p_type, const Transform &p_transform, float p_radius, float p_spot_aperture) { - if (unlikely(light_count == light_max)) { - light_max = nearest_power_of_2_templated(light_max + 1); - lights = (LightData *)memrealloc(lights, sizeof(LightData) * light_max); - } - - LightData &ld = lights[light_count]; - ld.type = p_type; - ld.position[0] = p_transform.origin.x; - ld.position[1] = p_transform.origin.y; - ld.position[2] = p_transform.origin.z; - ld.radius = p_radius; - ld.spot_aperture = p_spot_aperture; - - Transform xform = view_xform * p_transform; - - ld.radius *= xform.basis.get_uniform_scale(); - - AABB aabb; - - switch (p_type) { - case LIGHT_TYPE_OMNI: { - aabb.position = xform.origin; - aabb.size = Vector3(ld.radius, ld.radius, ld.radius); - aabb.position -= aabb.size; - aabb.size *= 2.0; - - _add_item(aabb, ITEM_TYPE_OMNI_LIGHT, light_count); - } break; - case LIGHT_TYPE_SPOT: { - float r = ld.radius; - real_t len = Math::tan(Math::deg2rad(ld.spot_aperture)) * r; - - aabb.position = xform.origin; - aabb.expand_to(xform.xform(Vector3(len, len, -r))); - aabb.expand_to(xform.xform(Vector3(-len, len, -r))); - aabb.expand_to(xform.xform(Vector3(-len, -len, -r))); - aabb.expand_to(xform.xform(Vector3(len, -len, -r))); - _add_item(aabb, ITEM_TYPE_SPOT_LIGHT, light_count); - } break; - } - - light_count++; - } - - _FORCE_INLINE_ void add_reflection_probe(const Transform &p_transform, const Vector3 &p_half_extents) { - if (unlikely(refprobe_count == refprobe_max)) { - refprobe_max = nearest_power_of_2_templated(refprobe_max + 1); - refprobes = (OrientedBoxData *)memrealloc(refprobes, sizeof(OrientedBoxData) * refprobe_max); - } - - Transform xform = view_xform * p_transform; - - OrientedBoxData &rp = refprobes[refprobe_count]; - Vector3 origin = xform.origin; - rp.position[0] = origin.x; - rp.position[1] = origin.y; - rp.position[2] = origin.z; - - Vector3 x_axis = xform.basis.get_axis(0) * p_half_extents.x; - rp.x_axis[0] = x_axis.x; - rp.x_axis[1] = x_axis.y; - rp.x_axis[2] = x_axis.z; - - Vector3 y_axis = xform.basis.get_axis(1) * p_half_extents.y; - rp.y_axis[0] = y_axis.x; - rp.y_axis[1] = y_axis.y; - rp.y_axis[2] = y_axis.z; - - Vector3 z_axis = xform.basis.get_axis(2) * p_half_extents.z; - rp.z_axis[0] = z_axis.x; - rp.z_axis[1] = z_axis.y; - rp.z_axis[2] = z_axis.z; - - AABB aabb; - - aabb.position = origin + x_axis + y_axis + z_axis; - aabb.expand_to(origin + x_axis + y_axis - z_axis); - aabb.expand_to(origin + x_axis - y_axis + z_axis); - aabb.expand_to(origin + x_axis - y_axis - z_axis); - aabb.expand_to(origin - x_axis + y_axis + z_axis); - aabb.expand_to(origin - x_axis + y_axis - z_axis); - aabb.expand_to(origin - x_axis - y_axis + z_axis); - aabb.expand_to(origin - x_axis - y_axis - z_axis); - - _add_item(aabb, ITEM_TYPE_REFLECTION_PROBE, refprobe_count); - - refprobe_count++; - } - - _FORCE_INLINE_ void add_decal(const Transform &p_transform, const Vector3 &p_half_extents) { - if (unlikely(decal_count == decal_max)) { - decal_max = nearest_power_of_2_templated(decal_max + 1); - decals = (OrientedBoxData *)memrealloc(decals, sizeof(OrientedBoxData) * decal_max); - } - - Transform xform = view_xform * p_transform; - - OrientedBoxData &dc = decals[decal_count]; - - Vector3 origin = xform.origin; - dc.position[0] = origin.x; - dc.position[1] = origin.y; - dc.position[2] = origin.z; - - Vector3 x_axis = xform.basis.get_axis(0) * p_half_extents.x; - dc.x_axis[0] = x_axis.x; - dc.x_axis[1] = x_axis.y; - dc.x_axis[2] = x_axis.z; - - Vector3 y_axis = xform.basis.get_axis(1) * p_half_extents.y; - dc.y_axis[0] = y_axis.x; - dc.y_axis[1] = y_axis.y; - dc.y_axis[2] = y_axis.z; - - Vector3 z_axis = xform.basis.get_axis(2) * p_half_extents.z; - dc.z_axis[0] = z_axis.x; - dc.z_axis[1] = z_axis.y; - dc.z_axis[2] = z_axis.z; - - AABB aabb; - - aabb.position = origin + x_axis + y_axis + z_axis; - aabb.expand_to(origin + x_axis + y_axis - z_axis); - aabb.expand_to(origin + x_axis - y_axis + z_axis); - aabb.expand_to(origin + x_axis - y_axis - z_axis); - aabb.expand_to(origin - x_axis + y_axis + z_axis); - aabb.expand_to(origin - x_axis + y_axis - z_axis); - aabb.expand_to(origin - x_axis - y_axis + z_axis); - aabb.expand_to(origin - x_axis - y_axis - z_axis); - - _add_item(aabb, ITEM_TYPE_DECAL, decal_count); - - decal_count++; - } - - void bake_cluster(); - - void setup(uint32_t p_width, uint32_t p_height, uint32_t p_depth); - - RID get_cluster_texture() const; - RID get_cluster_indices_buffer() const; - - LightClusterBuilder(); - ~LightClusterBuilder(); -}; - -#endif // LIGHT_CLUSTER_BUILDER_H diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index 05ffc0086d..7d6e2fa8e4 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -1367,7 +1367,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p } if (light_count > 0) { - RD::get_singleton()->buffer_update(state.lights_uniform_buffer, 0, sizeof(LightUniform) * light_count, &state.light_uniforms[0], true); + RD::get_singleton()->buffer_update(state.lights_uniform_buffer, 0, sizeof(LightUniform) * light_count, &state.light_uniforms[0]); } { @@ -1421,7 +1421,7 @@ void RendererCanvasRenderRD::canvas_render_items(RID p_to_render_target, Item *p //print_line("w: " + itos(ssize.width) + " s: " + rtos(canvas_scale)); state_buffer.tex_to_sdf = 1.0 / ((canvas_scale.x + canvas_scale.y) * 0.5); - RD::get_singleton()->buffer_update(state.canvas_state_buffer, 0, sizeof(State::Buffer), &state_buffer, true); + RD::get_singleton()->buffer_update(state.canvas_state_buffer, 0, sizeof(State::Buffer), &state_buffer); } { //default filter/repeat @@ -1622,7 +1622,7 @@ void RendererCanvasRenderRD::light_update_shadow(RID p_rid, int p_shadow_index, projection.set_frustum(xmin, xmax, ymin, ymax, nearp, farp); } - Vector3 cam_target = Basis(Vector3(0, 0, Math_PI * 2 * ((i + 3) / 4.0))).xform(Vector3(0, 1, 0)); + Vector3 cam_target = Basis(Vector3(0, 0, Math_TAU * ((i + 3) / 4.0))).xform(Vector3(0, 1, 0)); projection = projection * CameraMatrix(Transform().looking_at(cam_target, Vector3(0, 0, -1)).affine_inverse()); ShadowRenderPushConstant push_constant; @@ -2239,6 +2239,11 @@ Variant RendererCanvasRenderRD::ShaderData::get_default_parameter(const StringNa return Variant(); } +RS::ShaderNativeSourceCode RendererCanvasRenderRD::ShaderData::get_native_source_code() const { + RendererCanvasRenderRD *canvas_singleton = (RendererCanvasRenderRD *)RendererCanvasRender::singleton; + return canvas_singleton->shader.canvas_shader.version_get_native_source_code(version); +} + RendererCanvasRenderRD::ShaderData::ShaderData() { valid = false; uses_screen_texture = false; @@ -2501,7 +2506,7 @@ RendererCanvasRenderRD::RendererCanvasRenderRD(RendererStorageRD *p_storage) { actions.renames["FRAGCOORD"] = "gl_FragCoord"; actions.renames["POINT_COORD"] = "gl_PointCoord"; - actions.renames["LIGHT_POSITION"] = "light_pos"; + actions.renames["LIGHT_POSITION"] = "light_position"; actions.renames["LIGHT_COLOR"] = "light_color"; actions.renames["LIGHT_ENERGY"] = "light_energy"; actions.renames["LIGHT"] = "light"; @@ -2690,9 +2695,10 @@ RendererCanvasRenderRD::RendererCanvasRenderRD(RendererStorageRD *p_storage) { state.default_transforms_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shader.default_version_rd_shader, TRANSFORMS_UNIFORM_SET); } - default_canvas_texture = storage->canvas_texture_create(); + default_canvas_texture = storage->canvas_texture_allocate(); + storage->canvas_texture_initialize(default_canvas_texture); - state.shadow_texture_size = GLOBAL_GET("rendering/quality/2d_shadow_atlas/size"); + state.shadow_texture_size = GLOBAL_GET("rendering/2d/shadow_atlas/size"); //create functions for shader and material storage->shader_set_data_request_function(RendererStorageRD::SHADER_TYPE_2D, _create_shader_funcs); @@ -2701,9 +2707,14 @@ RendererCanvasRenderRD::RendererCanvasRenderRD(RendererStorageRD *p_storage) { state.time = 0; { - default_canvas_group_shader = storage->shader_create(); + default_canvas_group_shader = storage->shader_allocate(); + storage->shader_initialize(default_canvas_group_shader); + storage->shader_set_code(default_canvas_group_shader, "shader_type canvas_item; \nvoid fragment() {\n\tvec4 c = textureLod(SCREEN_TEXTURE,SCREEN_UV,0.0); if (c.a > 0.0001) c.rgb/=c.a; COLOR *= c; \n}\n"); - default_canvas_group_material = storage->material_create(); + + default_canvas_group_material = storage->material_allocate(); + storage->material_initialize(default_canvas_group_material); + storage->material_set_shader(default_canvas_group_material, default_canvas_group_shader); } diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h index 545eeaa106..cb947d7180 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.h @@ -188,6 +188,8 @@ class RendererCanvasRenderRD : public RendererCanvasRender { virtual bool is_animated() const; virtual bool casts_shadows() const; virtual Variant get_default_parameter(const StringName &p_parameter) const; + virtual RS::ShaderNativeSourceCode get_native_source_code() const; + ShaderData(); virtual ~ShaderData(); }; diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp index be2552bd32..2247b841c9 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.cpp @@ -175,5 +175,5 @@ RendererCompositorRD::RendererCompositorRD() { storage = memnew(RendererStorageRD); canvas = memnew(RendererCanvasRenderRD(storage)); - scene = memnew(RendererSceneRenderForward(storage)); + scene = memnew(RendererSceneRenderImplementation::RenderForwardClustered(storage)); } diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.h b/servers/rendering/renderer_rd/renderer_compositor_rd.h index cb85fc79e0..5b5f3ad0cb 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.h +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.h @@ -34,15 +34,15 @@ #include "core/os/os.h" #include "core/templates/thread_work_pool.h" #include "servers/rendering/renderer_compositor.h" +#include "servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h" #include "servers/rendering/renderer_rd/renderer_canvas_render_rd.h" -#include "servers/rendering/renderer_rd/renderer_scene_render_forward.h" #include "servers/rendering/renderer_rd/renderer_storage_rd.h" class RendererCompositorRD : public RendererCompositor { protected: RendererCanvasRenderRD *canvas; RendererStorageRD *storage; - RendererSceneRenderForward *scene; + RendererSceneRenderRD *scene; RID copy_viewports_rd_shader; RID copy_viewports_rd_pipeline; diff --git a/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp new file mode 100644 index 0000000000..d631cb4bac --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_environment_rd.cpp @@ -0,0 +1,126 @@ +/*************************************************************************/ +/* renderer_scene_environment_rd.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "servers/rendering/renderer_rd/renderer_scene_environment_rd.h" + +uint64_t RendererSceneEnvironmentRD::auto_exposure_counter = 2; + +void RendererSceneEnvironmentRD::set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color) { + ambient_light = p_color; + ambient_source = p_ambient; + ambient_light_energy = p_energy; + ambient_sky_contribution = p_sky_contribution; + reflection_source = p_reflection_source; + ao_color = p_ao_color; +} + +void RendererSceneEnvironmentRD::set_tonemap(RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) { + exposure = p_exposure; + tone_mapper = p_tone_mapper; + if (!auto_exposure && p_auto_exposure) { + auto_exposure_version = ++auto_exposure_counter; + } + auto_exposure = p_auto_exposure; + white = p_white; + min_luminance = p_min_luminance; + max_luminance = p_max_luminance; + auto_exp_speed = p_auto_exp_speed; + auto_exp_scale = p_auto_exp_scale; +} + +void RendererSceneEnvironmentRD::set_glow(bool p_enable, Vector<float> p_levels, float p_intensity, float p_strength, float p_mix, float p_bloom_threshold, RS::EnvironmentGlowBlendMode p_blend_mode, float p_hdr_bleed_threshold, float p_hdr_bleed_scale, float p_hdr_luminance_cap) { + ERR_FAIL_COND_MSG(p_levels.size() != 7, "Size of array of glow levels must be 7"); + glow_enabled = p_enable; + glow_levels = p_levels; + glow_intensity = p_intensity; + glow_strength = p_strength; + glow_mix = p_mix; + glow_bloom = p_bloom_threshold; + glow_blend_mode = p_blend_mode; + glow_hdr_bleed_threshold = p_hdr_bleed_threshold; + glow_hdr_bleed_scale = p_hdr_bleed_scale; + glow_hdr_luminance_cap = p_hdr_luminance_cap; +} + +void RendererSceneEnvironmentRD::set_sdfgi(bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) { + sdfgi_enabled = p_enable; + sdfgi_cascades = p_cascades; + sdfgi_min_cell_size = p_min_cell_size; + sdfgi_use_occlusion = p_use_occlusion; + sdfgi_bounce_feedback = p_bounce_feedback; + sdfgi_read_sky_light = p_read_sky; + sdfgi_energy = p_energy; + sdfgi_normal_bias = p_normal_bias; + sdfgi_probe_bias = p_probe_bias; + sdfgi_y_scale = p_y_scale; +} + +void RendererSceneEnvironmentRD::set_fog(bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_fog_aerial_perspective) { + fog_enabled = p_enable; + fog_light_color = p_light_color; + fog_light_energy = p_light_energy; + fog_sun_scatter = p_sun_scatter; + fog_density = p_density; + fog_height = p_height; + fog_height_density = p_height_density; + fog_aerial_perspective = p_fog_aerial_perspective; +} + +void RendererSceneEnvironmentRD::set_volumetric_fog(bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) { + volumetric_fog_enabled = p_enable; + volumetric_fog_density = p_density; + volumetric_fog_light = p_light; + volumetric_fog_light_energy = p_light_energy; + volumetric_fog_length = p_length; + volumetric_fog_detail_spread = p_detail_spread; + volumetric_fog_gi_inject = p_gi_inject; + volumetric_fog_temporal_reprojection = p_temporal_reprojection; + volumetric_fog_temporal_reprojection_amount = p_temporal_reprojection_amount; +} + +void RendererSceneEnvironmentRD::set_ssr(bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance) { + ssr_enabled = p_enable; + ssr_max_steps = p_max_steps; + ssr_fade_in = p_fade_int; + ssr_fade_out = p_fade_out; + ssr_depth_tolerance = p_depth_tolerance; +} + +void RendererSceneEnvironmentRD::set_ssao(bool p_enable, float p_radius, float p_intensity, float p_power, float p_detail, float p_horizon, float p_sharpness, float p_light_affect, float p_ao_channel_affect) { + ssao_enabled = p_enable; + ssao_radius = p_radius; + ssao_intensity = p_intensity; + ssao_power = p_power; + ssao_detail = p_detail; + ssao_horizon = p_horizon; + ssao_sharpness = p_sharpness; + ssao_direct_light_affect = p_light_affect; + ssao_ao_channel_affect = p_ao_channel_affect; +} diff --git a/servers/rendering/renderer_rd/renderer_scene_environment_rd.h b/servers/rendering/renderer_rd/renderer_scene_environment_rd.h new file mode 100644 index 0000000000..992c4bf471 --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_environment_rd.h @@ -0,0 +1,155 @@ +/*************************************************************************/ +/* renderer_scene_environment_rd.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef RENDERING_SERVER_SCENE_ENVIRONMENT_RD_H +#define RENDERING_SERVER_SCENE_ENVIRONMENT_RD_H + +#include "servers/rendering/renderer_scene_render.h" +#include "servers/rendering/rendering_device.h" + +class RendererSceneEnvironmentRD { +private: + static uint64_t auto_exposure_counter; + +public: + // BG + RS::EnvironmentBG background = RS::ENV_BG_CLEAR_COLOR; + RID sky; + float sky_custom_fov = 0.0; + Basis sky_orientation; + Color bg_color; + float bg_energy = 1.0; + int canvas_max_layer = 0; + RS::EnvironmentAmbientSource ambient_source = RS::ENV_AMBIENT_SOURCE_BG; + Color ambient_light; + float ambient_light_energy = 1.0; + float ambient_sky_contribution = 1.0; + RS::EnvironmentReflectionSource reflection_source = RS::ENV_REFLECTION_SOURCE_BG; + Color ao_color; + + /// Tonemap + + RS::EnvironmentToneMapper tone_mapper; + float exposure = 1.0; + float white = 1.0; + bool auto_exposure = false; + float min_luminance = 0.2; + float max_luminance = 8.0; + float auto_exp_speed = 0.2; + float auto_exp_scale = 0.5; + uint64_t auto_exposure_version = 0; + + // Fog + bool fog_enabled = false; + Color fog_light_color = Color(0.5, 0.6, 0.7); + float fog_light_energy = 1.0; + float fog_sun_scatter = 0.0; + float fog_density = 0.001; + float fog_height = 0.0; + float fog_height_density = 0.0; //can be negative to invert effect + float fog_aerial_perspective = 0.0; + + /// Volumetric Fog + /// + bool volumetric_fog_enabled = false; + float volumetric_fog_density = 0.01; + Color volumetric_fog_light = Color(0, 0, 0); + float volumetric_fog_light_energy = 0.0; + float volumetric_fog_length = 64.0; + float volumetric_fog_detail_spread = 2.0; + float volumetric_fog_gi_inject = 0.0; + bool volumetric_fog_temporal_reprojection = true; + float volumetric_fog_temporal_reprojection_amount = 0.9; + + /// Glow + + bool glow_enabled = false; + Vector<float> glow_levels; + float glow_intensity = 0.8; + float glow_strength = 1.0; + float glow_bloom = 0.0; + float glow_mix = 0.01; + RS::EnvironmentGlowBlendMode glow_blend_mode = RS::ENV_GLOW_BLEND_MODE_SOFTLIGHT; + float glow_hdr_bleed_threshold = 1.0; + float glow_hdr_luminance_cap = 12.0; + float glow_hdr_bleed_scale = 2.0; + + /// SSAO + + bool ssao_enabled = false; + float ssao_radius = 1.0; + float ssao_intensity = 2.0; + float ssao_power = 1.5; + float ssao_detail = 0.5; + float ssao_horizon = 0.06; + float ssao_sharpness = 0.98; + float ssao_direct_light_affect = 0.0; + float ssao_ao_channel_affect = 0.0; + + /// SSR + /// + bool ssr_enabled = false; + int ssr_max_steps = 64; + float ssr_fade_in = 0.15; + float ssr_fade_out = 2.0; + float ssr_depth_tolerance = 0.2; + + /// SDFGI + bool sdfgi_enabled = false; + RS::EnvironmentSDFGICascades sdfgi_cascades; + float sdfgi_min_cell_size = 0.2; + bool sdfgi_use_occlusion = false; + float sdfgi_bounce_feedback = 0.0; + bool sdfgi_read_sky_light = false; + float sdfgi_energy = 1.0; + float sdfgi_normal_bias = 1.1; + float sdfgi_probe_bias = 1.1; + RS::EnvironmentSDFGIYScale sdfgi_y_scale = RS::ENV_SDFGI_Y_SCALE_DISABLED; + + /// Adjustments + + bool adjustments_enabled = false; + float adjustments_brightness = 1.0f; + float adjustments_contrast = 1.0f; + float adjustments_saturation = 1.0f; + bool use_1d_color_correction = false; + RID color_correction = RID(); + + void set_ambient_light(const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color); + void set_tonemap(RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale); + void set_glow(bool p_enable, Vector<float> p_levels, float p_intensity, float p_strength, float p_mix, float p_bloom_threshold, RS::EnvironmentGlowBlendMode p_blend_mode, float p_hdr_bleed_threshold, float p_hdr_bleed_scale, float p_hdr_luminance_cap); + void set_sdfgi(bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias); + void set_fog(bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_fog_aerial_perspective); + void set_volumetric_fog(bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount); + void set_ssr(bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance); + void set_ssao(bool p_enable, float p_radius, float p_intensity, float p_power, float p_detail, float p_horizon, float p_sharpness, float p_light_affect, float p_ao_channel_affect); +}; + +#endif /* !RENDERING_SERVER_SCENE_ENVIRONMENT_RD_H */ diff --git a/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp new file mode 100644 index 0000000000..3856f38457 --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_gi_rd.cpp @@ -0,0 +1,3404 @@ +/*************************************************************************/ +/* renderer_scene_gi_rd.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "renderer_scene_gi_rd.h" + +#include "core/config/project_settings.h" +#include "servers/rendering/renderer_rd/renderer_scene_render_rd.h" +#include "servers/rendering/rendering_server_default.h" + +const Vector3i RendererSceneGIRD::SDFGI::Cascade::DIRTY_ALL = Vector3i(0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF); + +//////////////////////////////////////////////////////////////////////////////// +// SDFGI + +void RendererSceneGIRD::SDFGI::create(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size, RendererSceneGIRD *p_gi) { + storage = p_gi->storage; + gi = p_gi; + cascade_mode = p_env->sdfgi_cascades; + min_cell_size = p_env->sdfgi_min_cell_size; + uses_occlusion = p_env->sdfgi_use_occlusion; + y_scale_mode = p_env->sdfgi_y_scale; + static const float y_scale[3] = { 1.0, 1.5, 2.0 }; + y_mult = y_scale[y_scale_mode]; + static const int cascasde_size[3] = { 4, 6, 8 }; + cascades.resize(cascasde_size[cascade_mode]); + probe_axis_count = SDFGI::PROBE_DIVISOR + 1; + solid_cell_ratio = gi->sdfgi_solid_cell_ratio; + solid_cell_count = uint32_t(float(cascade_size * cascade_size * cascade_size) * solid_cell_ratio); + + float base_cell_size = min_cell_size; + + RD::TextureFormat tf_sdf; + tf_sdf.format = RD::DATA_FORMAT_R8_UNORM; + tf_sdf.width = cascade_size; // Always 64x64 + tf_sdf.height = cascade_size; + tf_sdf.depth = cascade_size; + tf_sdf.texture_type = RD::TEXTURE_TYPE_3D; + tf_sdf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; + + { + RD::TextureFormat tf_render = tf_sdf; + tf_render.format = RD::DATA_FORMAT_R16_UINT; + render_albedo = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + tf_render.format = RD::DATA_FORMAT_R32_UINT; + render_emission = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + render_emission_aniso = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + + tf_render.format = RD::DATA_FORMAT_R8_UNORM; //at least its easy to visualize + + for (int i = 0; i < 8; i++) { + render_occlusion[i] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + } + + tf_render.format = RD::DATA_FORMAT_R32_UINT; + render_geom_facing = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + + tf_render.format = RD::DATA_FORMAT_R8G8B8A8_UINT; + render_sdf[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + render_sdf[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + + tf_render.width /= 2; + tf_render.height /= 2; + tf_render.depth /= 2; + + render_sdf_half[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + render_sdf_half[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); + } + + RD::TextureFormat tf_occlusion = tf_sdf; + tf_occlusion.format = RD::DATA_FORMAT_R16_UINT; + tf_occlusion.shareable_formats.push_back(RD::DATA_FORMAT_R16_UINT); + tf_occlusion.shareable_formats.push_back(RD::DATA_FORMAT_R4G4B4A4_UNORM_PACK16); + tf_occlusion.depth *= cascades.size(); //use depth for occlusion slices + tf_occlusion.width *= 2; //use width for the other half + + RD::TextureFormat tf_light = tf_sdf; + tf_light.format = RD::DATA_FORMAT_R32_UINT; + tf_light.shareable_formats.push_back(RD::DATA_FORMAT_R32_UINT); + tf_light.shareable_formats.push_back(RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32); + + RD::TextureFormat tf_aniso0 = tf_sdf; + tf_aniso0.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; + RD::TextureFormat tf_aniso1 = tf_sdf; + tf_aniso1.format = RD::DATA_FORMAT_R8G8_UNORM; + + int passes = nearest_shift(cascade_size) - 1; + + //store lightprobe SH + RD::TextureFormat tf_probes; + tf_probes.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf_probes.width = probe_axis_count * probe_axis_count; + tf_probes.height = probe_axis_count * SDFGI::SH_SIZE; + tf_probes.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; + tf_probes.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; + + history_size = p_requested_history_size; + + RD::TextureFormat tf_probe_history = tf_probes; + tf_probe_history.format = RD::DATA_FORMAT_R16G16B16A16_SINT; //signed integer because SH are signed + tf_probe_history.array_layers = history_size; + + RD::TextureFormat tf_probe_average = tf_probes; + tf_probe_average.format = RD::DATA_FORMAT_R32G32B32A32_SINT; //signed integer because SH are signed + tf_probe_average.texture_type = RD::TEXTURE_TYPE_2D; + + lightprobe_history_scroll = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); + lightprobe_average_scroll = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); + + { + //octahedral lightprobes + RD::TextureFormat tf_octprobes = tf_probes; + tf_octprobes.array_layers = cascades.size() * 2; + tf_octprobes.format = RD::DATA_FORMAT_R32_UINT; //pack well with RGBE + tf_octprobes.width = probe_axis_count * probe_axis_count * (SDFGI::LIGHTPROBE_OCT_SIZE + 2); + tf_octprobes.height = probe_axis_count * (SDFGI::LIGHTPROBE_OCT_SIZE + 2); + tf_octprobes.shareable_formats.push_back(RD::DATA_FORMAT_R32_UINT); + tf_octprobes.shareable_formats.push_back(RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32); + //lightprobe texture is an octahedral texture + + lightprobe_data = RD::get_singleton()->texture_create(tf_octprobes, RD::TextureView()); + RD::TextureView tv; + tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; + lightprobe_texture = RD::get_singleton()->texture_create_shared(tv, lightprobe_data); + + //texture handling ambient data, to integrate with volumetric foc + RD::TextureFormat tf_ambient = tf_probes; + tf_ambient.array_layers = cascades.size(); + tf_ambient.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; //pack well with RGBE + tf_ambient.width = probe_axis_count * probe_axis_count; + tf_ambient.height = probe_axis_count; + tf_ambient.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; + //lightprobe texture is an octahedral texture + ambient_texture = RD::get_singleton()->texture_create(tf_ambient, RD::TextureView()); + } + + cascades_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(SDFGI::Cascade::UBO) * SDFGI::MAX_CASCADES); + + occlusion_data = RD::get_singleton()->texture_create(tf_occlusion, RD::TextureView()); + { + RD::TextureView tv; + tv.format_override = RD::DATA_FORMAT_R4G4B4A4_UNORM_PACK16; + occlusion_texture = RD::get_singleton()->texture_create_shared(tv, occlusion_data); + } + + for (uint32_t i = 0; i < cascades.size(); i++) { + SDFGI::Cascade &cascade = cascades[i]; + + /* 3D Textures */ + + cascade.sdf_tex = RD::get_singleton()->texture_create(tf_sdf, RD::TextureView()); + + cascade.light_data = RD::get_singleton()->texture_create(tf_light, RD::TextureView()); + + cascade.light_aniso_0_tex = RD::get_singleton()->texture_create(tf_aniso0, RD::TextureView()); + cascade.light_aniso_1_tex = RD::get_singleton()->texture_create(tf_aniso1, RD::TextureView()); + + { + RD::TextureView tv; + tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; + cascade.light_tex = RD::get_singleton()->texture_create_shared(tv, cascade.light_data); + + RD::get_singleton()->texture_clear(cascade.light_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(cascade.light_aniso_0_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(cascade.light_aniso_1_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + } + + cascade.cell_size = base_cell_size; + Vector3 world_position = p_world_position; + world_position.y *= y_mult; + int32_t probe_cells = cascade_size / SDFGI::PROBE_DIVISOR; + Vector3 probe_size = Vector3(1, 1, 1) * cascade.cell_size * probe_cells; + Vector3i probe_pos = Vector3i((world_position / probe_size + Vector3(0.5, 0.5, 0.5)).floor()); + cascade.position = probe_pos * probe_cells; + + cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; + + base_cell_size *= 2.0; + + /* Probe History */ + + cascade.lightprobe_history_tex = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); + RD::get_singleton()->texture_clear(cascade.lightprobe_history_tex, Color(0, 0, 0, 0), 0, 1, 0, tf_probe_history.array_layers); //needs to be cleared for average to work + + cascade.lightprobe_average_tex = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); + RD::get_singleton()->texture_clear(cascade.lightprobe_average_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); //needs to be cleared for average to work + + /* Buffers */ + + cascade.solid_cell_buffer = RD::get_singleton()->storage_buffer_create(sizeof(SDFGI::Cascade::SolidCell) * solid_cell_count); + cascade.solid_cell_dispatch_buffer = RD::get_singleton()->storage_buffer_create(sizeof(uint32_t) * 4, Vector<uint8_t>(), RD::STORAGE_BUFFER_USAGE_DISPATCH_INDIRECT); + cascade.lights_buffer = RD::get_singleton()->storage_buffer_create(sizeof(SDFGIShader::Light) * MAX(SDFGI::MAX_STATIC_LIGHTS, SDFGI::MAX_DYNAMIC_LIGHTS)); + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_sdf[(passes & 1) ? 1 : 0]); //if passes are even, we read from buffer 0, else we read from buffer 1 + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 3; + for (int j = 0; j < 8; j++) { + u.ids.push_back(render_occlusion[j]); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 4; + u.ids.push_back(render_emission); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 5; + u.ids.push_back(render_emission_aniso); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 6; + u.ids.push_back(render_geom_facing); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 7; + u.ids.push_back(cascade.sdf_tex); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 8; + u.ids.push_back(occlusion_data); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 10; + u.ids.push_back(cascade.solid_cell_dispatch_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 11; + u.ids.push_back(cascade.solid_cell_buffer); + uniforms.push_back(u); + } + + cascade.sdf_store_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_STORE), 0); + } + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_geom_facing); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 3; + u.ids.push_back(render_emission); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 4; + u.ids.push_back(render_emission_aniso); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 5; + u.ids.push_back(cascade.solid_cell_dispatch_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 6; + u.ids.push_back(cascade.solid_cell_buffer); + uniforms.push_back(u); + } + + cascade.scroll_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_SCROLL), 0); + } + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + for (int j = 0; j < 8; j++) { + u.ids.push_back(render_occlusion[j]); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(occlusion_data); + uniforms.push_back(u); + } + + cascade.scroll_occlusion_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_SCROLL_OCCLUSION), 0); + } + } + + //direct light + for (uint32_t i = 0; i < cascades.size(); i++) { + SDFGI::Cascade &cascade = cascades[i]; + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (j < cascades.size()) { + u.ids.push_back(cascades[j].sdf_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(cascade.solid_cell_dispatch_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(cascade.solid_cell_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 5; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.ids.push_back(cascade.light_data); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 6; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.ids.push_back(cascade.light_aniso_0_tex); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 7; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.ids.push_back(cascade.light_aniso_1_tex); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 8; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(cascades_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 9; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.ids.push_back(cascade.lights_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 10; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(lightprobe_texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 11; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(occlusion_texture); + uniforms.push_back(u); + } + + cascade.sdf_direct_light_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.direct_light.version_get_shader(gi->sdfgi_shader.direct_light_shader, 0), 0); + } + + //preprocess initialize uniform set + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_sdf[0]); + uniforms.push_back(u); + } + + sdf_initialize_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE), 0); + } + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_sdf_half[0]); + uniforms.push_back(u); + } + + sdf_initialize_half_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF), 0); + } + + //jump flood uniform set + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_sdf[0]); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_sdf[1]); + uniforms.push_back(u); + } + + jump_flood_uniform_set[0] = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD), 0); + SWAP(uniforms.write[0].ids.write[0], uniforms.write[1].ids.write[0]); + jump_flood_uniform_set[1] = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD), 0); + } + //jump flood half uniform set + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_sdf_half[0]); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_sdf_half[1]); + uniforms.push_back(u); + } + + jump_flood_half_uniform_set[0] = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD), 0); + SWAP(uniforms.write[0].ids.write[0], uniforms.write[1].ids.write[0]); + jump_flood_half_uniform_set[1] = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD), 0); + } + + //upscale half size sdf + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + u.ids.push_back(render_sdf_half[(passes & 1) ? 0 : 1]); //reverse pass order because half size + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 3; + u.ids.push_back(render_sdf[(passes & 1) ? 0 : 1]); //reverse pass order because it needs an extra JFA pass + uniforms.push_back(u); + } + + upscale_jfa_uniform_set_index = (passes & 1) ? 0 : 1; + sdf_upscale_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_JUMP_FLOOD_UPSCALE), 0); + } + + //occlusion uniform set + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 1; + u.ids.push_back(render_albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 2; + for (int i = 0; i < 8; i++) { + u.ids.push_back(render_occlusion[i]); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 3; + u.ids.push_back(render_geom_facing); + uniforms.push_back(u); + } + + occlusion_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.preprocess.version_get_shader(gi->sdfgi_shader.preprocess_shader, SDFGIShader::PRE_PROCESS_OCCLUSION), 0); + } + + for (uint32_t i = 0; i < cascades.size(); i++) { + //integrate uniform + + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (j < cascades.size()) { + u.ids.push_back(cascades[j].sdf_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (j < cascades.size()) { + u.ids.push_back(cascades[j].light_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (j < cascades.size()) { + u.ids.push_back(cascades[j].light_aniso_0_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (j < cascades.size()) { + u.ids.push_back(cascades[j].light_aniso_1_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 6; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 7; + u.ids.push_back(cascades_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 8; + u.ids.push_back(lightprobe_data); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 9; + u.ids.push_back(cascades[i].lightprobe_history_tex); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 10; + u.ids.push_back(cascades[i].lightprobe_average_tex); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 11; + u.ids.push_back(lightprobe_history_scroll); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 12; + u.ids.push_back(lightprobe_average_scroll); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 13; + RID parent_average; + if (i < cascades.size() - 1) { + parent_average = cascades[i + 1].lightprobe_average_tex; + } else { + parent_average = cascades[i - 1].lightprobe_average_tex; //to use something, but it won't be used + } + u.ids.push_back(parent_average); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 14; + u.ids.push_back(ambient_texture); + uniforms.push_back(u); + } + + cascades[i].integrate_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.integrate.version_get_shader(gi->sdfgi_shader.integrate_shader, 0), 0); + } + + bounce_feedback = p_env->sdfgi_bounce_feedback; + energy = p_env->sdfgi_energy; + normal_bias = p_env->sdfgi_normal_bias; + probe_bias = p_env->sdfgi_probe_bias; + reads_sky = p_env->sdfgi_read_sky_light; +} + +void RendererSceneGIRD::SDFGI::erase() { + for (uint32_t i = 0; i < cascades.size(); i++) { + const SDFGI::Cascade &c = cascades[i]; + RD::get_singleton()->free(c.light_data); + RD::get_singleton()->free(c.light_aniso_0_tex); + RD::get_singleton()->free(c.light_aniso_1_tex); + RD::get_singleton()->free(c.sdf_tex); + RD::get_singleton()->free(c.solid_cell_dispatch_buffer); + RD::get_singleton()->free(c.solid_cell_buffer); + RD::get_singleton()->free(c.lightprobe_history_tex); + RD::get_singleton()->free(c.lightprobe_average_tex); + RD::get_singleton()->free(c.lights_buffer); + } + + RD::get_singleton()->free(render_albedo); + RD::get_singleton()->free(render_emission); + RD::get_singleton()->free(render_emission_aniso); + + RD::get_singleton()->free(render_sdf[0]); + RD::get_singleton()->free(render_sdf[1]); + + RD::get_singleton()->free(render_sdf_half[0]); + RD::get_singleton()->free(render_sdf_half[1]); + + for (int i = 0; i < 8; i++) { + RD::get_singleton()->free(render_occlusion[i]); + } + + RD::get_singleton()->free(render_geom_facing); + + RD::get_singleton()->free(lightprobe_data); + RD::get_singleton()->free(lightprobe_history_scroll); + RD::get_singleton()->free(occlusion_data); + RD::get_singleton()->free(ambient_texture); + + RD::get_singleton()->free(cascades_ubo); +} + +void RendererSceneGIRD::SDFGI::update(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position) { + bounce_feedback = p_env->sdfgi_bounce_feedback; + energy = p_env->sdfgi_energy; + normal_bias = p_env->sdfgi_normal_bias; + probe_bias = p_env->sdfgi_probe_bias; + reads_sky = p_env->sdfgi_read_sky_light; + + int32_t drag_margin = (cascade_size / SDFGI::PROBE_DIVISOR) / 2; + + for (uint32_t i = 0; i < cascades.size(); i++) { + SDFGI::Cascade &cascade = cascades[i]; + cascade.dirty_regions = Vector3i(); + + Vector3 probe_half_size = Vector3(1, 1, 1) * cascade.cell_size * float(cascade_size / SDFGI::PROBE_DIVISOR) * 0.5; + probe_half_size = Vector3(0, 0, 0); + + Vector3 world_position = p_world_position; + world_position.y *= y_mult; + Vector3i pos_in_cascade = Vector3i((world_position + probe_half_size) / cascade.cell_size); + + for (int j = 0; j < 3; j++) { + if (pos_in_cascade[j] < cascade.position[j]) { + while (pos_in_cascade[j] < (cascade.position[j] - drag_margin)) { + cascade.position[j] -= drag_margin * 2; + cascade.dirty_regions[j] += drag_margin * 2; + } + } else if (pos_in_cascade[j] > cascade.position[j]) { + while (pos_in_cascade[j] > (cascade.position[j] + drag_margin)) { + cascade.position[j] += drag_margin * 2; + cascade.dirty_regions[j] -= drag_margin * 2; + } + } + + if (cascade.dirty_regions[j] == 0) { + continue; // not dirty + } else if (uint32_t(ABS(cascade.dirty_regions[j])) >= cascade_size) { + //moved too much, just redraw everything (make all dirty) + cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; + break; + } + } + + if (cascade.dirty_regions != Vector3i() && cascade.dirty_regions != SDFGI::Cascade::DIRTY_ALL) { + //see how much the total dirty volume represents from the total volume + uint32_t total_volume = cascade_size * cascade_size * cascade_size; + uint32_t safe_volume = 1; + for (int j = 0; j < 3; j++) { + safe_volume *= cascade_size - ABS(cascade.dirty_regions[j]); + } + uint32_t dirty_volume = total_volume - safe_volume; + if (dirty_volume > (safe_volume / 2)) { + //more than half the volume is dirty, make all dirty so its only rendered once + cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; + } + } + } +} + +void RendererSceneGIRD::SDFGI::update_light() { + RD::get_singleton()->draw_command_begin_label("SDFGI Update dynamic Light"); + + /* Update dynamic light */ + + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.direct_light_pipeline[SDFGIShader::DIRECT_LIGHT_MODE_DYNAMIC]); + + SDFGIShader::DirectLightPushConstant push_constant; + + push_constant.grid_size[0] = cascade_size; + push_constant.grid_size[1] = cascade_size; + push_constant.grid_size[2] = cascade_size; + push_constant.max_cascades = cascades.size(); + push_constant.probe_axis_size = probe_axis_count; + push_constant.bounce_feedback = bounce_feedback; + push_constant.y_mult = y_mult; + push_constant.use_occlusion = uses_occlusion; + + for (uint32_t i = 0; i < cascades.size(); i++) { + SDFGI::Cascade &cascade = cascades[i]; + push_constant.light_count = cascade_dynamic_light_count[i]; + push_constant.cascade = i; + + if (cascades[i].all_dynamic_lights_dirty || gi->sdfgi_frames_to_update_light == RS::ENV_SDFGI_UPDATE_LIGHT_IN_1_FRAME) { + push_constant.process_offset = 0; + push_constant.process_increment = 1; + } else { + static uint32_t frames_to_update_table[RS::ENV_SDFGI_UPDATE_LIGHT_MAX] = { + 1, 2, 4, 8, 16 + }; + + uint32_t frames_to_update = frames_to_update_table[gi->sdfgi_frames_to_update_light]; + + push_constant.process_offset = RSG::rasterizer->get_frame_number() % frames_to_update; + push_constant.process_increment = frames_to_update; + } + cascades[i].all_dynamic_lights_dirty = false; + + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascade.sdf_direct_light_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::DirectLightPushConstant)); + RD::get_singleton()->compute_list_dispatch_indirect(compute_list, cascade.solid_cell_dispatch_buffer, 0); + } + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_COMPUTE); + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneGIRD::SDFGI::update_probes(RendererSceneEnvironmentRD *p_env, RendererSceneSkyRD::Sky *p_sky) { + RD::get_singleton()->draw_command_begin_label("SDFGI Update Probes"); + + SDFGIShader::IntegratePushConstant push_constant; + push_constant.grid_size[1] = cascade_size; + push_constant.grid_size[2] = cascade_size; + push_constant.grid_size[0] = cascade_size; + push_constant.max_cascades = cascades.size(); + push_constant.probe_axis_size = probe_axis_count; + push_constant.history_index = render_pass % history_size; + push_constant.history_size = history_size; + static const uint32_t ray_count[RS::ENV_SDFGI_RAY_COUNT_MAX] = { 4, 8, 16, 32, 64, 96, 128 }; + push_constant.ray_count = ray_count[gi->sdfgi_ray_count]; + push_constant.ray_bias = probe_bias; + push_constant.image_size[0] = probe_axis_count * probe_axis_count; + push_constant.image_size[1] = probe_axis_count; + push_constant.store_ambient_texture = p_env->volumetric_fog_enabled; + + RID sky_uniform_set = gi->sdfgi_shader.integrate_default_sky_uniform_set; + push_constant.sky_mode = SDFGIShader::IntegratePushConstant::SKY_MODE_DISABLED; + push_constant.y_mult = y_mult; + + if (reads_sky && p_env) { + push_constant.sky_energy = p_env->bg_energy; + + if (p_env->background == RS::ENV_BG_CLEAR_COLOR) { + push_constant.sky_mode = SDFGIShader::IntegratePushConstant::SKY_MODE_COLOR; + Color c = storage->get_default_clear_color().to_linear(); + push_constant.sky_color[0] = c.r; + push_constant.sky_color[1] = c.g; + push_constant.sky_color[2] = c.b; + } else if (p_env->background == RS::ENV_BG_COLOR) { + push_constant.sky_mode = SDFGIShader::IntegratePushConstant::SKY_MODE_COLOR; + Color c = p_env->bg_color; + push_constant.sky_color[0] = c.r; + push_constant.sky_color[1] = c.g; + push_constant.sky_color[2] = c.b; + + } else if (p_env->background == RS::ENV_BG_SKY) { + if (p_sky && p_sky->radiance.is_valid()) { + if (integrate_sky_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(integrate_sky_uniform_set)) { + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 0; + u.ids.push_back(p_sky->radiance); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 1; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + integrate_sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.integrate.version_get_shader(gi->sdfgi_shader.integrate_shader, 0), 1); + } + sky_uniform_set = integrate_sky_uniform_set; + push_constant.sky_mode = SDFGIShader::IntegratePushConstant::SKY_MODE_SKY; + } + } + } + + render_pass++; + + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(true); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.integrate_pipeline[SDFGIShader::INTEGRATE_MODE_PROCESS]); + + int32_t probe_divisor = cascade_size / SDFGI::PROBE_DIVISOR; + for (uint32_t i = 0; i < cascades.size(); i++) { + push_constant.cascade = i; + push_constant.world_offset[0] = cascades[i].position.x / probe_divisor; + push_constant.world_offset[1] = cascades[i].position.y / probe_divisor; + push_constant.world_offset[2] = cascades[i].position.z / probe_divisor; + + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[i].integrate_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sky_uniform_set, 1); + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::IntegratePushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, probe_axis_count * probe_axis_count, probe_axis_count, 1); + } + + //end later after raster to avoid barriering on layout changes + //RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_NO_BARRIER); + + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneGIRD::SDFGI::store_probes() { + RD::get_singleton()->barrier(RD::BARRIER_MASK_COMPUTE, RD::BARRIER_MASK_COMPUTE); + RD::get_singleton()->draw_command_begin_label("SDFGI Store Probes"); + + SDFGIShader::IntegratePushConstant push_constant; + push_constant.grid_size[1] = cascade_size; + push_constant.grid_size[2] = cascade_size; + push_constant.grid_size[0] = cascade_size; + push_constant.max_cascades = cascades.size(); + push_constant.probe_axis_size = probe_axis_count; + push_constant.history_index = render_pass % history_size; + push_constant.history_size = history_size; + static const uint32_t ray_count[RS::ENV_SDFGI_RAY_COUNT_MAX] = { 4, 8, 16, 32, 64, 96, 128 }; + push_constant.ray_count = ray_count[gi->sdfgi_ray_count]; + push_constant.ray_bias = probe_bias; + push_constant.image_size[0] = probe_axis_count * probe_axis_count; + push_constant.image_size[1] = probe_axis_count; + push_constant.store_ambient_texture = false; + + push_constant.sky_mode = 0; + push_constant.y_mult = y_mult; + + // Then store values into the lightprobe texture. Separating these steps has a small performance hit, but it allows for multiple bounces + RENDER_TIMESTAMP("Average Probes"); + + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.integrate_pipeline[SDFGIShader::INTEGRATE_MODE_STORE]); + + //convert to octahedral to store + push_constant.image_size[0] *= SDFGI::LIGHTPROBE_OCT_SIZE; + push_constant.image_size[1] *= SDFGI::LIGHTPROBE_OCT_SIZE; + + for (uint32_t i = 0; i < cascades.size(); i++) { + push_constant.cascade = i; + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[i].integrate_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi->sdfgi_shader.integrate_default_sky_uniform_set, 1); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::IntegratePushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, probe_axis_count * probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, 1); + } + + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_COMPUTE); + + RD::get_singleton()->draw_command_end_label(); +} + +int RendererSceneGIRD::SDFGI::get_pending_region_data(int p_region, Vector3i &r_local_offset, Vector3i &r_local_size, AABB &r_bounds) const { + int dirty_count = 0; + for (uint32_t i = 0; i < cascades.size(); i++) { + const SDFGI::Cascade &c = cascades[i]; + + if (c.dirty_regions == SDFGI::Cascade::DIRTY_ALL) { + if (dirty_count == p_region) { + r_local_offset = Vector3i(); + r_local_size = Vector3i(1, 1, 1) * cascade_size; + + r_bounds.position = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + c.position)) * c.cell_size * Vector3(1, 1.0 / y_mult, 1); + r_bounds.size = Vector3(r_local_size) * c.cell_size * Vector3(1, 1.0 / y_mult, 1); + return i; + } + dirty_count++; + } else { + for (int j = 0; j < 3; j++) { + if (c.dirty_regions[j] != 0) { + if (dirty_count == p_region) { + Vector3i from = Vector3i(0, 0, 0); + Vector3i to = Vector3i(1, 1, 1) * cascade_size; + + if (c.dirty_regions[j] > 0) { + //fill from the beginning + to[j] = c.dirty_regions[j]; + } else { + //fill from the end + from[j] = to[j] + c.dirty_regions[j]; + } + + for (int k = 0; k < j; k++) { + // "chip" away previous regions to avoid re-voxelizing the same thing + if (c.dirty_regions[k] > 0) { + from[k] += c.dirty_regions[k]; + } else if (c.dirty_regions[k] < 0) { + to[k] += c.dirty_regions[k]; + } + } + + r_local_offset = from; + r_local_size = to - from; + + r_bounds.position = Vector3(from + Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + c.position) * c.cell_size * Vector3(1, 1.0 / y_mult, 1); + r_bounds.size = Vector3(r_local_size) * c.cell_size * Vector3(1, 1.0 / y_mult, 1); + + return i; + } + + dirty_count++; + } + } + } + } + return -1; +} + +void RendererSceneGIRD::SDFGI::update_cascades() { + //update cascades + SDFGI::Cascade::UBO cascade_data[SDFGI::MAX_CASCADES]; + int32_t probe_divisor = cascade_size / SDFGI::PROBE_DIVISOR; + + for (uint32_t i = 0; i < cascades.size(); i++) { + Vector3 pos = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + cascades[i].position)) * cascades[i].cell_size; + + cascade_data[i].offset[0] = pos.x; + cascade_data[i].offset[1] = pos.y; + cascade_data[i].offset[2] = pos.z; + cascade_data[i].to_cell = 1.0 / cascades[i].cell_size; + cascade_data[i].probe_offset[0] = cascades[i].position.x / probe_divisor; + cascade_data[i].probe_offset[1] = cascades[i].position.y / probe_divisor; + cascade_data[i].probe_offset[2] = cascades[i].position.z / probe_divisor; + cascade_data[i].pad = 0; + } + + RD::get_singleton()->buffer_update(cascades_ubo, 0, sizeof(SDFGI::Cascade::UBO) * SDFGI::MAX_CASCADES, cascade_data, RD::BARRIER_MASK_COMPUTE); +} + +void RendererSceneGIRD::SDFGI::debug_draw(const CameraMatrix &p_projection, const Transform &p_transform, int p_width, int p_height, RID p_render_target, RID p_texture) { + if (!debug_uniform_set.is_valid() || !RD::get_singleton()->uniform_set_is_valid(debug_uniform_set)) { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { + if (i < cascades.size()) { + u.ids.push_back(cascades[i].sdf_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { + if (i < cascades.size()) { + u.ids.push_back(cascades[i].light_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { + if (i < cascades.size()) { + u.ids.push_back(cascades[i].light_aniso_0_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { + if (i < cascades.size()) { + u.ids.push_back(cascades[i].light_aniso_1_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 5; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(occlusion_texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 8; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 9; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(cascades_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 10; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.ids.push_back(p_texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 11; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(lightprobe_texture); + uniforms.push_back(u); + } + debug_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.debug_shader_version, 0); + } + + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.debug_pipeline); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, debug_uniform_set, 0); + + SDFGIShader::DebugPushConstant push_constant; + push_constant.grid_size[0] = cascade_size; + push_constant.grid_size[1] = cascade_size; + push_constant.grid_size[2] = cascade_size; + push_constant.max_cascades = cascades.size(); + push_constant.screen_size[0] = p_width; + push_constant.screen_size[1] = p_height; + push_constant.probe_axis_size = probe_axis_count; + push_constant.use_occlusion = uses_occlusion; + push_constant.y_mult = y_mult; + + Vector2 vp_half = p_projection.get_viewport_half_extents(); + push_constant.cam_extent[0] = vp_half.x; + push_constant.cam_extent[1] = vp_half.y; + push_constant.cam_extent[2] = -p_projection.get_z_near(); + + push_constant.cam_transform[0] = p_transform.basis.elements[0][0]; + push_constant.cam_transform[1] = p_transform.basis.elements[1][0]; + push_constant.cam_transform[2] = p_transform.basis.elements[2][0]; + push_constant.cam_transform[3] = 0; + push_constant.cam_transform[4] = p_transform.basis.elements[0][1]; + push_constant.cam_transform[5] = p_transform.basis.elements[1][1]; + push_constant.cam_transform[6] = p_transform.basis.elements[2][1]; + push_constant.cam_transform[7] = 0; + push_constant.cam_transform[8] = p_transform.basis.elements[0][2]; + push_constant.cam_transform[9] = p_transform.basis.elements[1][2]; + push_constant.cam_transform[10] = p_transform.basis.elements[2][2]; + push_constant.cam_transform[11] = 0; + push_constant.cam_transform[12] = p_transform.origin.x; + push_constant.cam_transform[13] = p_transform.origin.y; + push_constant.cam_transform[14] = p_transform.origin.z; + push_constant.cam_transform[15] = 1; + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::DebugPushConstant)); + + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_width, p_height, 1); + RD::get_singleton()->compute_list_end(); + + Size2 rtsize = storage->render_target_get_size(p_render_target); + storage->get_effects()->copy_to_fb_rect(p_texture, storage->render_target_get_rd_framebuffer(p_render_target), Rect2(Vector2(), rtsize), true); +} + +void RendererSceneGIRD::SDFGI::debug_probes(RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform) { + SDFGIShader::DebugProbesPushConstant push_constant; + + for (int i = 0; i < 4; i++) { + for (int j = 0; j < 4; j++) { + push_constant.projection[i * 4 + j] = p_camera_with_transform.matrix[i][j]; + } + } + + //gen spheres from strips + uint32_t band_points = 16; + push_constant.band_power = 4; + push_constant.sections_in_band = ((band_points / 2) - 1); + push_constant.band_mask = band_points - 2; + push_constant.section_arc = Math_TAU / float(push_constant.sections_in_band); + push_constant.y_mult = y_mult; + + uint32_t total_points = push_constant.sections_in_band * band_points; + uint32_t total_probes = probe_axis_count * probe_axis_count * probe_axis_count; + + push_constant.grid_size[0] = cascade_size; + push_constant.grid_size[1] = cascade_size; + push_constant.grid_size[2] = cascade_size; + push_constant.cascade = 0; + + push_constant.probe_axis_size = probe_axis_count; + + if (!debug_probes_uniform_set.is_valid() || !RD::get_singleton()->uniform_set_is_valid(debug_probes_uniform_set)) { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(cascades_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(lightprobe_texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.ids.push_back(occlusion_texture); + uniforms.push_back(u); + } + + debug_probes_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->sdfgi_shader.debug_probes.version_get_shader(gi->sdfgi_shader.debug_probes_shader, 0), 0); + } + + RD::get_singleton()->draw_list_bind_render_pipeline(p_draw_list, gi->sdfgi_shader.debug_probes_pipeline[SDFGIShader::PROBE_DEBUG_PROBES].get_render_pipeline(RD::INVALID_FORMAT_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, debug_probes_uniform_set, 0); + RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(SDFGIShader::DebugProbesPushConstant)); + RD::get_singleton()->draw_list_draw(p_draw_list, false, total_probes, total_points); + + if (gi->sdfgi_debug_probe_dir != Vector3()) { + print_line("CLICK DEBUG ME?"); + uint32_t cascade = 0; + Vector3 offset = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + cascades[cascade].position)) * cascades[cascade].cell_size * Vector3(1.0, 1.0 / y_mult, 1.0); + Vector3 probe_size = cascades[cascade].cell_size * (cascade_size / SDFGI::PROBE_DIVISOR) * Vector3(1.0, 1.0 / y_mult, 1.0); + Vector3 ray_from = gi->sdfgi_debug_probe_pos; + Vector3 ray_to = gi->sdfgi_debug_probe_pos + gi->sdfgi_debug_probe_dir * cascades[cascade].cell_size * Math::sqrt(3.0) * cascade_size; + float sphere_radius = 0.2; + float closest_dist = 1e20; + gi->sdfgi_debug_probe_enabled = false; + + Vector3i probe_from = cascades[cascade].position / (cascade_size / SDFGI::PROBE_DIVISOR); + for (int i = 0; i < (SDFGI::PROBE_DIVISOR + 1); i++) { + for (int j = 0; j < (SDFGI::PROBE_DIVISOR + 1); j++) { + for (int k = 0; k < (SDFGI::PROBE_DIVISOR + 1); k++) { + Vector3 pos = offset + probe_size * Vector3(i, j, k); + Vector3 res; + if (Geometry3D::segment_intersects_sphere(ray_from, ray_to, pos, sphere_radius, &res)) { + float d = ray_from.distance_to(res); + if (d < closest_dist) { + closest_dist = d; + gi->sdfgi_debug_probe_enabled = true; + gi->sdfgi_debug_probe_index = probe_from + Vector3i(i, j, k); + } + } + } + } + } + + if (gi->sdfgi_debug_probe_enabled) { + print_line("found: " + gi->sdfgi_debug_probe_index); + } else { + print_line("no found"); + } + gi->sdfgi_debug_probe_dir = Vector3(); + } + + if (gi->sdfgi_debug_probe_enabled) { + uint32_t cascade = 0; + uint32_t probe_cells = (cascade_size / SDFGI::PROBE_DIVISOR); + Vector3i probe_from = cascades[cascade].position / probe_cells; + Vector3i ofs = gi->sdfgi_debug_probe_index - probe_from; + if (ofs.x < 0 || ofs.y < 0 || ofs.z < 0) { + return; + } + if (ofs.x > SDFGI::PROBE_DIVISOR || ofs.y > SDFGI::PROBE_DIVISOR || ofs.z > SDFGI::PROBE_DIVISOR) { + return; + } + + uint32_t mult = (SDFGI::PROBE_DIVISOR + 1); + uint32_t index = ofs.z * mult * mult + ofs.y * mult + ofs.x; + + push_constant.probe_debug_index = index; + + uint32_t cell_count = probe_cells * 2 * probe_cells * 2 * probe_cells * 2; + + RD::get_singleton()->draw_list_bind_render_pipeline(p_draw_list, gi->sdfgi_shader.debug_probes_pipeline[SDFGIShader::PROBE_DEBUG_VISIBILITY].get_render_pipeline(RD::INVALID_FORMAT_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, debug_probes_uniform_set, 0); + RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(SDFGIShader::DebugProbesPushConstant)); + RD::get_singleton()->draw_list_draw(p_draw_list, false, cell_count, total_points); + } +} + +void RendererSceneGIRD::SDFGI::pre_process_gi(const Transform &p_transform, RendererSceneRenderRD *p_scene_render) { + /* Update general SDFGI Buffer */ + + SDFGIData sdfgi_data; + + sdfgi_data.grid_size[0] = cascade_size; + sdfgi_data.grid_size[1] = cascade_size; + sdfgi_data.grid_size[2] = cascade_size; + + sdfgi_data.max_cascades = cascades.size(); + sdfgi_data.probe_axis_size = probe_axis_count; + sdfgi_data.cascade_probe_size[0] = sdfgi_data.probe_axis_size - 1; //float version for performance + sdfgi_data.cascade_probe_size[1] = sdfgi_data.probe_axis_size - 1; + sdfgi_data.cascade_probe_size[2] = sdfgi_data.probe_axis_size - 1; + + float csize = cascade_size; + sdfgi_data.probe_to_uvw = 1.0 / float(sdfgi_data.cascade_probe_size[0]); + sdfgi_data.use_occlusion = uses_occlusion; + //sdfgi_data.energy = energy; + + sdfgi_data.y_mult = y_mult; + + float cascade_voxel_size = (csize / sdfgi_data.cascade_probe_size[0]); + float occlusion_clamp = (cascade_voxel_size - 0.5) / cascade_voxel_size; + sdfgi_data.occlusion_clamp[0] = occlusion_clamp; + sdfgi_data.occlusion_clamp[1] = occlusion_clamp; + sdfgi_data.occlusion_clamp[2] = occlusion_clamp; + sdfgi_data.normal_bias = (normal_bias / csize) * sdfgi_data.cascade_probe_size[0]; + + //vec2 tex_pixel_size = 1.0 / vec2(ivec2( (OCT_SIZE+2) * params.probe_axis_size * params.probe_axis_size, (OCT_SIZE+2) * params.probe_axis_size ) ); + //vec3 probe_uv_offset = (ivec3(OCT_SIZE+2,OCT_SIZE+2,(OCT_SIZE+2) * params.probe_axis_size)) * tex_pixel_size.xyx; + + uint32_t oct_size = SDFGI::LIGHTPROBE_OCT_SIZE; + + sdfgi_data.lightprobe_tex_pixel_size[0] = 1.0 / ((oct_size + 2) * sdfgi_data.probe_axis_size * sdfgi_data.probe_axis_size); + sdfgi_data.lightprobe_tex_pixel_size[1] = 1.0 / ((oct_size + 2) * sdfgi_data.probe_axis_size); + sdfgi_data.lightprobe_tex_pixel_size[2] = 1.0; + + sdfgi_data.energy = energy; + + sdfgi_data.lightprobe_uv_offset[0] = float(oct_size + 2) * sdfgi_data.lightprobe_tex_pixel_size[0]; + sdfgi_data.lightprobe_uv_offset[1] = float(oct_size + 2) * sdfgi_data.lightprobe_tex_pixel_size[1]; + sdfgi_data.lightprobe_uv_offset[2] = float((oct_size + 2) * sdfgi_data.probe_axis_size) * sdfgi_data.lightprobe_tex_pixel_size[0]; + + sdfgi_data.occlusion_renormalize[0] = 0.5; + sdfgi_data.occlusion_renormalize[1] = 1.0; + sdfgi_data.occlusion_renormalize[2] = 1.0 / float(sdfgi_data.max_cascades); + + int32_t probe_divisor = cascade_size / SDFGI::PROBE_DIVISOR; + + for (uint32_t i = 0; i < sdfgi_data.max_cascades; i++) { + SDFGIData::ProbeCascadeData &c = sdfgi_data.cascades[i]; + Vector3 pos = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + cascades[i].position)) * cascades[i].cell_size; + Vector3 cam_origin = p_transform.origin; + cam_origin.y *= y_mult; + pos -= cam_origin; //make pos local to camera, to reduce numerical error + c.position[0] = pos.x; + c.position[1] = pos.y; + c.position[2] = pos.z; + c.to_probe = 1.0 / (float(cascade_size) * cascades[i].cell_size / float(probe_axis_count - 1)); + + Vector3i probe_ofs = cascades[i].position / probe_divisor; + c.probe_world_offset[0] = probe_ofs.x; + c.probe_world_offset[1] = probe_ofs.y; + c.probe_world_offset[2] = probe_ofs.z; + + c.to_cell = 1.0 / cascades[i].cell_size; + } + + RD::get_singleton()->buffer_update(gi->sdfgi_ubo, 0, sizeof(SDFGIData), &sdfgi_data, RD::BARRIER_MASK_COMPUTE); + + /* Update dynamic lights in SDFGI cascades */ + + for (uint32_t i = 0; i < cascades.size(); i++) { + SDFGI::Cascade &cascade = cascades[i]; + + SDFGIShader::Light lights[SDFGI::MAX_DYNAMIC_LIGHTS]; + uint32_t idx = 0; + for (uint32_t j = 0; j < (uint32_t)p_scene_render->render_state.sdfgi_update_data->directional_lights->size(); j++) { + if (idx == SDFGI::MAX_DYNAMIC_LIGHTS) { + break; + } + + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_scene_render->render_state.sdfgi_update_data->directional_lights->get(j)); + ERR_CONTINUE(!li); + + if (storage->light_directional_is_sky_only(li->light)) { + continue; + } + + Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); + dir.y *= y_mult; + dir.normalize(); + lights[idx].direction[0] = dir.x; + lights[idx].direction[1] = dir.y; + lights[idx].direction[2] = dir.z; + Color color = storage->light_get_color(li->light); + color = color.to_linear(); + lights[idx].color[0] = color.r; + lights[idx].color[1] = color.g; + lights[idx].color[2] = color.b; + lights[idx].type = RS::LIGHT_DIRECTIONAL; + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].has_shadow = storage->light_has_shadow(li->light); + + idx++; + } + + AABB cascade_aabb; + cascade_aabb.position = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + cascade.position)) * cascade.cell_size; + cascade_aabb.size = Vector3(1, 1, 1) * cascade_size * cascade.cell_size; + + for (uint32_t j = 0; j < p_scene_render->render_state.sdfgi_update_data->positional_light_count; j++) { + if (idx == SDFGI::MAX_DYNAMIC_LIGHTS) { + break; + } + + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_scene_render->render_state.sdfgi_update_data->positional_light_instances[j]); + ERR_CONTINUE(!li); + + uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); + if (i > max_sdfgi_cascade) { + continue; + } + + if (!cascade_aabb.intersects(li->aabb)) { + continue; + } + + Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); + //faster to not do this here + //dir.y *= y_mult; + //dir.normalize(); + lights[idx].direction[0] = dir.x; + lights[idx].direction[1] = dir.y; + lights[idx].direction[2] = dir.z; + Vector3 pos = li->transform.origin; + pos.y *= y_mult; + lights[idx].position[0] = pos.x; + lights[idx].position[1] = pos.y; + lights[idx].position[2] = pos.z; + Color color = storage->light_get_color(li->light); + color = color.to_linear(); + lights[idx].color[0] = color.r; + lights[idx].color[1] = color.g; + lights[idx].color[2] = color.b; + lights[idx].type = storage->light_get_type(li->light); + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].has_shadow = storage->light_has_shadow(li->light); + lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); + lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); + lights[idx].cos_spot_angle = Math::cos(Math::deg2rad(storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ANGLE))); + lights[idx].inv_spot_attenuation = 1.0f / storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ATTENUATION); + + idx++; + } + + if (idx > 0) { + RD::get_singleton()->buffer_update(cascade.lights_buffer, 0, idx * sizeof(SDFGIShader::Light), lights, RD::BARRIER_MASK_COMPUTE); + } + + cascade_dynamic_light_count[i] = idx; + } +} + +void RendererSceneGIRD::SDFGI::render_region(RID p_render_buffers, int p_region, const PagedArray<RendererSceneRender::GeometryInstance *> &p_instances, RendererSceneRenderRD *p_scene_render) { + //print_line("rendering region " + itos(p_region)); + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND(!rb); // we wouldn't be here if this failed but... + AABB bounds; + Vector3i from; + Vector3i size; + + int cascade_prev = get_pending_region_data(p_region - 1, from, size, bounds); + int cascade_next = get_pending_region_data(p_region + 1, from, size, bounds); + int cascade = get_pending_region_data(p_region, from, size, bounds); + ERR_FAIL_COND(cascade < 0); + + if (cascade_prev != cascade) { + //initialize render + RD::get_singleton()->texture_clear(render_albedo, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(render_emission, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(render_emission_aniso, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(render_geom_facing, Color(0, 0, 0, 0), 0, 1, 0, 1); + } + + //print_line("rendering cascade " + itos(p_region) + " objects: " + itos(p_cull_count) + " bounds: " + bounds + " from: " + from + " size: " + size + " cell size: " + rtos(cascades[cascade].cell_size)); + p_scene_render->_render_sdfgi(p_render_buffers, from, size, bounds, p_instances, render_albedo, render_emission, render_emission_aniso, render_geom_facing); + + if (cascade_next != cascade) { + RD::get_singleton()->draw_command_begin_label("SDFGI Pre-Process Cascade"); + + RENDER_TIMESTAMP(">SDFGI Update SDF"); + //done rendering! must update SDF + //clear dispatch indirect data + + SDFGIShader::PreprocessPushConstant push_constant; + zeromem(&push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + + RENDER_TIMESTAMP("Scroll SDF"); + + //scroll + if (cascades[cascade].dirty_regions != SDFGI::Cascade::DIRTY_ALL) { + //for scroll + Vector3i dirty = cascades[cascade].dirty_regions; + push_constant.scroll[0] = dirty.x; + push_constant.scroll[1] = dirty.y; + push_constant.scroll[2] = dirty.z; + } else { + //for no scroll + push_constant.scroll[0] = 0; + push_constant.scroll[1] = 0; + push_constant.scroll[2] = 0; + } + + cascades[cascade].all_dynamic_lights_dirty = true; + + push_constant.grid_size = cascade_size; + push_constant.cascade = cascade; + + if (cascades[cascade].dirty_regions != SDFGI::Cascade::DIRTY_ALL) { + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + + //must pre scroll existing data because not all is dirty + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_SCROLL]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].scroll_uniform_set, 0); + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_indirect(compute_list, cascades[cascade].solid_cell_dispatch_buffer, 0); + // no barrier do all together + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_SCROLL_OCCLUSION]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].scroll_occlusion_uniform_set, 0); + + Vector3i dirty = cascades[cascade].dirty_regions; + Vector3i groups; + groups.x = cascade_size - ABS(dirty.x); + groups.y = cascade_size - ABS(dirty.y); + groups.z = cascade_size - ABS(dirty.z); + + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, groups.x, groups.y, groups.z); + + //no barrier, continue together + + { + //scroll probes and their history also + + SDFGIShader::IntegratePushConstant ipush_constant; + ipush_constant.grid_size[1] = cascade_size; + ipush_constant.grid_size[2] = cascade_size; + ipush_constant.grid_size[0] = cascade_size; + ipush_constant.max_cascades = cascades.size(); + ipush_constant.probe_axis_size = probe_axis_count; + ipush_constant.history_index = 0; + ipush_constant.history_size = history_size; + ipush_constant.ray_count = 0; + ipush_constant.ray_bias = 0; + ipush_constant.sky_mode = 0; + ipush_constant.sky_energy = 0; + ipush_constant.sky_color[0] = 0; + ipush_constant.sky_color[1] = 0; + ipush_constant.sky_color[2] = 0; + ipush_constant.y_mult = y_mult; + ipush_constant.store_ambient_texture = false; + + ipush_constant.image_size[0] = probe_axis_count * probe_axis_count; + ipush_constant.image_size[1] = probe_axis_count; + + int32_t probe_divisor = cascade_size / SDFGI::PROBE_DIVISOR; + ipush_constant.cascade = cascade; + ipush_constant.world_offset[0] = cascades[cascade].position.x / probe_divisor; + ipush_constant.world_offset[1] = cascades[cascade].position.y / probe_divisor; + ipush_constant.world_offset[2] = cascades[cascade].position.z / probe_divisor; + + ipush_constant.scroll[0] = dirty.x / probe_divisor; + ipush_constant.scroll[1] = dirty.y / probe_divisor; + ipush_constant.scroll[2] = dirty.z / probe_divisor; + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.integrate_pipeline[SDFGIShader::INTEGRATE_MODE_SCROLL]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].integrate_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi->sdfgi_shader.integrate_default_sky_uniform_set, 1); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &ipush_constant, sizeof(SDFGIShader::IntegratePushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, probe_axis_count * probe_axis_count, probe_axis_count, 1); + + RD::get_singleton()->compute_list_add_barrier(compute_list); + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.integrate_pipeline[SDFGIShader::INTEGRATE_MODE_SCROLL_STORE]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].integrate_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi->sdfgi_shader.integrate_default_sky_uniform_set, 1); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &ipush_constant, sizeof(SDFGIShader::IntegratePushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, probe_axis_count * probe_axis_count, probe_axis_count, 1); + + RD::get_singleton()->compute_list_add_barrier(compute_list); + + if (bounce_feedback > 0.0) { + //multibounce requires this to be stored so direct light can read from it + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.integrate_pipeline[SDFGIShader::INTEGRATE_MODE_STORE]); + + //convert to octahedral to store + ipush_constant.image_size[0] *= SDFGI::LIGHTPROBE_OCT_SIZE; + ipush_constant.image_size[1] *= SDFGI::LIGHTPROBE_OCT_SIZE; + + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].integrate_uniform_set, 0); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi->sdfgi_shader.integrate_default_sky_uniform_set, 1); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &ipush_constant, sizeof(SDFGIShader::IntegratePushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, probe_axis_count * probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, 1); + } + } + + //ok finally barrier + RD::get_singleton()->compute_list_end(); + } + + //clear dispatch indirect data + uint32_t dispatch_indirct_data[4] = { 0, 0, 0, 0 }; + RD::get_singleton()->buffer_update(cascades[cascade].solid_cell_dispatch_buffer, 0, sizeof(uint32_t) * 4, dispatch_indirct_data); + + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + + bool half_size = true; //much faster, very little difference + static const int optimized_jf_group_size = 8; + + if (half_size) { + push_constant.grid_size >>= 1; + + uint32_t cascade_half_size = cascade_size >> 1; + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sdf_initialize_half_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + + //must start with regular jumpflood + + push_constant.half_size = true; + { + RENDER_TIMESTAMP("SDFGI Jump Flood (Half Size)"); + + uint32_t s = cascade_half_size; + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD]); + + int jf_us = 0; + //start with regular jump flood for very coarse reads, as this is impossible to optimize + while (s > 1) { + s /= 2; + push_constant.step_size = s; + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, jump_flood_half_uniform_set[jf_us], 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + jf_us = jf_us == 0 ? 1 : 0; + + if (cascade_half_size / (s / 2) >= optimized_jf_group_size) { + break; + } + } + + RENDER_TIMESTAMP("SDFGI Jump Flood Optimized (Half Size)"); + + //continue with optimized jump flood for smaller reads + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); + while (s > 1) { + s /= 2; + push_constant.step_size = s; + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, jump_flood_half_uniform_set[jf_us], 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + jf_us = jf_us == 0 ? 1 : 0; + } + } + + // restore grid size for last passes + push_constant.grid_size = cascade_size; + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_UPSCALE]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sdf_upscale_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + + //run one pass of fullsize jumpflood to fix up half size arctifacts + + push_constant.half_size = false; + push_constant.step_size = 1; + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, jump_flood_uniform_set[upscale_jfa_uniform_set_index], 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + + } else { + //full size jumpflood + RENDER_TIMESTAMP("SDFGI Jump Flood"); + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sdf_initialize_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + + RD::get_singleton()->compute_list_add_barrier(compute_list); + + push_constant.half_size = false; + { + uint32_t s = cascade_size; + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD]); + + int jf_us = 0; + //start with regular jump flood for very coarse reads, as this is impossible to optimize + while (s > 1) { + s /= 2; + push_constant.step_size = s; + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, jump_flood_uniform_set[jf_us], 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + jf_us = jf_us == 0 ? 1 : 0; + + if (cascade_size / (s / 2) >= optimized_jf_group_size) { + break; + } + } + + RENDER_TIMESTAMP("SDFGI Jump Flood Optimized"); + + //continue with optimized jump flood for smaller reads + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); + while (s > 1) { + s /= 2; + push_constant.step_size = s; + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, jump_flood_uniform_set[jf_us], 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + RD::get_singleton()->compute_list_add_barrier(compute_list); + jf_us = jf_us == 0 ? 1 : 0; + } + } + } + + RENDER_TIMESTAMP("SDFGI Occlusion"); + + // occlusion + { + uint32_t probe_size = cascade_size / SDFGI::PROBE_DIVISOR; + Vector3i probe_global_pos = cascades[cascade].position / probe_size; + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_OCCLUSION]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, occlusion_uniform_set, 0); + for (int i = 0; i < 8; i++) { + //dispatch all at once for performance + Vector3i offset(i & 1, (i >> 1) & 1, (i >> 2) & 1); + + if ((probe_global_pos.x & 1) != 0) { + offset.x = (offset.x + 1) & 1; + } + if ((probe_global_pos.y & 1) != 0) { + offset.y = (offset.y + 1) & 1; + } + if ((probe_global_pos.z & 1) != 0) { + offset.z = (offset.z + 1) & 1; + } + push_constant.probe_offset[0] = offset.x; + push_constant.probe_offset[1] = offset.y; + push_constant.probe_offset[2] = offset.z; + push_constant.occlusion_index = i; + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + + Vector3i groups = Vector3i(probe_size + 1, probe_size + 1, probe_size + 1) - offset; //if offset, it's one less probe per axis to compute + RD::get_singleton()->compute_list_dispatch(compute_list, groups.x, groups.y, groups.z); + } + RD::get_singleton()->compute_list_add_barrier(compute_list); + } + + RENDER_TIMESTAMP("SDFGI Store"); + + // store + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.preprocess_pipeline[SDFGIShader::PRE_PROCESS_STORE]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascades[cascade].sdf_store_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDFGIShader::PreprocessPushConstant)); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_size, cascade_size, cascade_size); + + RD::get_singleton()->compute_list_end(); + + //clear these textures, as they will have previous garbage on next draw + RD::get_singleton()->texture_clear(cascades[cascade].light_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(cascades[cascade].light_aniso_0_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + RD::get_singleton()->texture_clear(cascades[cascade].light_aniso_1_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); + +#if 0 + Vector<uint8_t> data = RD::get_singleton()->texture_get_data(cascades[cascade].sdf, 0); + Ref<Image> img; + img.instance(); + for (uint32_t i = 0; i < cascade_size; i++) { + Vector<uint8_t> subarr = data.subarray(128 * 128 * i, 128 * 128 * (i + 1) - 1); + img->create(cascade_size, cascade_size, false, Image::FORMAT_L8, subarr); + img->save_png("res://cascade_sdf_" + itos(cascade) + "_" + itos(i) + ".png"); + } + + //finalize render and update sdf +#endif + +#if 0 + Vector<uint8_t> data = RD::get_singleton()->texture_get_data(render_albedo, 0); + Ref<Image> img; + img.instance(); + for (uint32_t i = 0; i < cascade_size; i++) { + Vector<uint8_t> subarr = data.subarray(128 * 128 * i * 2, 128 * 128 * (i + 1) * 2 - 1); + img->createcascade_size, cascade_size, false, Image::FORMAT_RGB565, subarr); + img->convert(Image::FORMAT_RGBA8); + img->save_png("res://cascade_" + itos(cascade) + "_" + itos(i) + ".png"); + } + + //finalize render and update sdf +#endif + + RENDER_TIMESTAMP("<SDFGI Update SDF"); + RD::get_singleton()->draw_command_end_label(); + } +} + +void RendererSceneGIRD::SDFGI::render_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_light_cull_result, RendererSceneRenderRD *p_scene_render) { + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND(!rb); // we wouldn't be here if this failed but... + + RD::get_singleton()->draw_command_begin_label("SDFGI Render Static Lighs"); + + update_cascades(); + ; //need cascades updated for this + + SDFGIShader::Light lights[SDFGI::MAX_STATIC_LIGHTS]; + uint32_t light_count[SDFGI::MAX_STATIC_LIGHTS]; + + for (uint32_t i = 0; i < p_cascade_count; i++) { + ERR_CONTINUE(p_cascade_indices[i] >= cascades.size()); + + SDFGI::Cascade &cc = cascades[p_cascade_indices[i]]; + + { //fill light buffer + + AABB cascade_aabb; + cascade_aabb.position = Vector3((Vector3i(1, 1, 1) * -int32_t(cascade_size >> 1) + cc.position)) * cc.cell_size; + cascade_aabb.size = Vector3(1, 1, 1) * cascade_size * cc.cell_size; + + int idx = 0; + + for (uint32_t j = 0; j < (uint32_t)p_positional_light_cull_result[i].size(); j++) { + if (idx == SDFGI::MAX_STATIC_LIGHTS) { + break; + } + + RendererSceneRenderRD::LightInstance *li = p_scene_render->light_instance_owner.getornull(p_positional_light_cull_result[i][j]); + ERR_CONTINUE(!li); + + uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); + if (p_cascade_indices[i] > max_sdfgi_cascade) { + continue; + } + + if (!cascade_aabb.intersects(li->aabb)) { + continue; + } + + lights[idx].type = storage->light_get_type(li->light); + + Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); + if (lights[idx].type == RS::LIGHT_DIRECTIONAL) { + dir.y *= y_mult; //only makes sense for directional + dir.normalize(); + } + lights[idx].direction[0] = dir.x; + lights[idx].direction[1] = dir.y; + lights[idx].direction[2] = dir.z; + Vector3 pos = li->transform.origin; + pos.y *= y_mult; + lights[idx].position[0] = pos.x; + lights[idx].position[1] = pos.y; + lights[idx].position[2] = pos.z; + Color color = storage->light_get_color(li->light); + color = color.to_linear(); + lights[idx].color[0] = color.r; + lights[idx].color[1] = color.g; + lights[idx].color[2] = color.b; + lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); + lights[idx].has_shadow = storage->light_has_shadow(li->light); + lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); + lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); + lights[idx].cos_spot_angle = Math::cos(Math::deg2rad(storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ANGLE))); + lights[idx].inv_spot_attenuation = 1.0f / storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ATTENUATION); + + idx++; + } + + if (idx > 0) { + RD::get_singleton()->buffer_update(cc.lights_buffer, 0, idx * sizeof(SDFGIShader::Light), lights); + } + + light_count[i] = idx; + } + } + + /* Static Lights */ + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->sdfgi_shader.direct_light_pipeline[SDFGIShader::DIRECT_LIGHT_MODE_STATIC]); + + SDFGIShader::DirectLightPushConstant dl_push_constant; + + dl_push_constant.grid_size[0] = cascade_size; + dl_push_constant.grid_size[1] = cascade_size; + dl_push_constant.grid_size[2] = cascade_size; + dl_push_constant.max_cascades = cascades.size(); + dl_push_constant.probe_axis_size = probe_axis_count; + dl_push_constant.bounce_feedback = 0.0; // this is static light, do not multibounce yet + dl_push_constant.y_mult = y_mult; + dl_push_constant.use_occlusion = uses_occlusion; + + //all must be processed + dl_push_constant.process_offset = 0; + dl_push_constant.process_increment = 1; + + for (uint32_t i = 0; i < p_cascade_count; i++) { + ERR_CONTINUE(p_cascade_indices[i] >= cascades.size()); + + SDFGI::Cascade &cc = cascades[p_cascade_indices[i]]; + + dl_push_constant.light_count = light_count[i]; + dl_push_constant.cascade = p_cascade_indices[i]; + + if (dl_push_constant.light_count > 0) { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cc.sdf_direct_light_uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &dl_push_constant, sizeof(SDFGIShader::DirectLightPushConstant)); + RD::get_singleton()->compute_list_dispatch_indirect(compute_list, cc.solid_cell_dispatch_buffer, 0); + } + } + + RD::get_singleton()->compute_list_end(); + + RD::get_singleton()->draw_command_end_label(); +} + +//////////////////////////////////////////////////////////////////////////////// +// GIProbeInstance + +void RendererSceneGIRD::GIProbeInstance::update(bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<RendererSceneRender::GeometryInstance *> &p_dynamic_objects, RendererSceneRenderRD *p_scene_render) { + uint32_t data_version = storage->gi_probe_get_data_version(probe); + + // (RE)CREATE IF NEEDED + + if (last_probe_data_version != data_version) { + //need to re-create everything + if (texture.is_valid()) { + RD::get_singleton()->free(texture); + RD::get_singleton()->free(write_buffer); + mipmaps.clear(); + } + + for (int i = 0; i < dynamic_maps.size(); i++) { + RD::get_singleton()->free(dynamic_maps[i].texture); + RD::get_singleton()->free(dynamic_maps[i].depth); + } + + dynamic_maps.clear(); + + Vector3i octree_size = storage->gi_probe_get_octree_size(probe); + + if (octree_size != Vector3i()) { + //can create a 3D texture + Vector<int> levels = storage->gi_probe_get_level_counts(probe); + + RD::TextureFormat tf; + tf.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; + tf.width = octree_size.x; + tf.height = octree_size.y; + tf.depth = octree_size.z; + tf.texture_type = RD::TEXTURE_TYPE_3D; + tf.mipmaps = levels.size(); + + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; + + texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + RD::get_singleton()->texture_clear(texture, Color(0, 0, 0, 0), 0, levels.size(), 0, 1); + + { + int total_elements = 0; + for (int i = 0; i < levels.size(); i++) { + total_elements += levels[i]; + } + + write_buffer = RD::get_singleton()->storage_buffer_create(total_elements * 16); + } + + for (int i = 0; i < levels.size(); i++) { + GIProbeInstance::Mipmap mipmap; + mipmap.texture = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), texture, 0, i, RD::TEXTURE_SLICE_3D); + mipmap.level = levels.size() - i - 1; + mipmap.cell_offset = 0; + for (uint32_t j = 0; j < mipmap.level; j++) { + mipmap.cell_offset += levels[j]; + } + mipmap.cell_count = levels[mipmap.level]; + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(storage->gi_probe_get_octree_buffer(probe)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 2; + u.ids.push_back(storage->gi_probe_get_data_buffer(probe)); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 4; + u.ids.push_back(write_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 9; + u.ids.push_back(storage->gi_probe_get_sdf_texture(probe)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 10; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + { + Vector<RD::Uniform> copy_uniforms = uniforms; + if (i == 0) { + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 3; + u.ids.push_back(gi->gi_probe_lights_uniform); + copy_uniforms.push_back(u); + } + + mipmap.uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, gi->giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT], 0); + + copy_uniforms = uniforms; //restore + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 5; + u.ids.push_back(texture); + copy_uniforms.push_back(u); + } + mipmap.second_bounce_uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, gi->giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE], 0); + } else { + mipmap.uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, gi->giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP], 0); + } + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 5; + u.ids.push_back(mipmap.texture); + uniforms.push_back(u); + } + + mipmap.write_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_WRITE_TEXTURE], 0); + + mipmaps.push_back(mipmap); + } + + { + uint32_t dynamic_map_size = MAX(MAX(octree_size.x, octree_size.y), octree_size.z); + uint32_t oversample = nearest_power_of_2_templated(4); + int mipmap_index = 0; + + while (mipmap_index < mipmaps.size()) { + GIProbeInstance::DynamicMap dmap; + + if (oversample > 0) { + dmap.size = dynamic_map_size * (1 << oversample); + dmap.mipmap = -1; + oversample--; + } else { + dmap.size = dynamic_map_size >> mipmap_index; + dmap.mipmap = mipmap_index; + mipmap_index++; + } + + RD::TextureFormat dtf; + dtf.width = dmap.size; + dtf.height = dmap.size; + dtf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; + + if (dynamic_maps.size() == 0) { + dtf.usage_bits |= RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + } + dmap.texture = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + + if (dynamic_maps.size() == 0) { + //render depth for first one + dtf.format = RD::get_singleton()->texture_is_format_supported_for_usage(RD::DATA_FORMAT_D32_SFLOAT, RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT) ? RD::DATA_FORMAT_D32_SFLOAT : RD::DATA_FORMAT_X8_D24_UNORM_PACK32; + dtf.usage_bits = RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT; + dmap.fb_depth = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + } + + //just use depth as-is + dtf.format = RD::DATA_FORMAT_R32_SFLOAT; + dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + + dmap.depth = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + + if (dynamic_maps.size() == 0) { + dtf.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; + dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + dmap.albedo = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + dmap.normal = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + dmap.orm = RD::get_singleton()->texture_create(dtf, RD::TextureView()); + + Vector<RID> fb; + fb.push_back(dmap.albedo); + fb.push_back(dmap.normal); + fb.push_back(dmap.orm); + fb.push_back(dmap.texture); //emission + fb.push_back(dmap.depth); + fb.push_back(dmap.fb_depth); + + dmap.fb = RD::get_singleton()->framebuffer_create(fb); + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 3; + u.ids.push_back(gi->gi_probe_lights_uniform); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 5; + u.ids.push_back(dmap.albedo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 6; + u.ids.push_back(dmap.normal); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 7; + u.ids.push_back(dmap.orm); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 8; + u.ids.push_back(dmap.fb_depth); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 9; + u.ids.push_back(storage->gi_probe_get_sdf_texture(probe)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 10; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 11; + u.ids.push_back(dmap.texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 12; + u.ids.push_back(dmap.depth); + uniforms.push_back(u); + } + + dmap.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING], 0); + } + } else { + bool plot = dmap.mipmap >= 0; + bool write = dmap.mipmap < (mipmaps.size() - 1); + + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 5; + u.ids.push_back(dynamic_maps[dynamic_maps.size() - 1].texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 6; + u.ids.push_back(dynamic_maps[dynamic_maps.size() - 1].depth); + uniforms.push_back(u); + } + + if (write) { + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 7; + u.ids.push_back(dmap.texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 8; + u.ids.push_back(dmap.depth); + uniforms.push_back(u); + } + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 9; + u.ids.push_back(storage->gi_probe_get_sdf_texture(probe)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 10; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + if (plot) { + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 11; + u.ids.push_back(mipmaps[dmap.mipmap].texture); + uniforms.push_back(u); + } + } + + dmap.uniform_set = RD::get_singleton()->uniform_set_create( + uniforms, + gi->giprobe_lighting_shader_version_shaders[(write && plot) ? GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT : (write ? GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE : GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT)], + 0); + } + + dynamic_maps.push_back(dmap); + } + } + } + + last_probe_data_version = data_version; + p_update_light_instances = true; //just in case + + p_scene_render->_base_uniforms_changed(); + } + + // UDPDATE TIME + + if (has_dynamic_object_data) { + //if it has dynamic object data, it needs to be cleared + RD::get_singleton()->texture_clear(texture, Color(0, 0, 0, 0), 0, mipmaps.size(), 0, 1); + } + + uint32_t light_count = 0; + + if (p_update_light_instances || p_dynamic_objects.size() > 0) { + light_count = MIN(gi->gi_probe_max_lights, (uint32_t)p_light_instances.size()); + + { + Transform to_cell = storage->gi_probe_get_to_cell_xform(probe); + Transform to_probe_xform = (transform * to_cell.affine_inverse()).affine_inverse(); + //update lights + + for (uint32_t i = 0; i < light_count; i++) { + GIProbeLight &l = gi->gi_probe_lights[i]; + RID light_instance = p_light_instances[i]; + RID light = p_scene_render->light_instance_get_base_light(light_instance); + + l.type = storage->light_get_type(light); + if (l.type == RS::LIGHT_DIRECTIONAL && storage->light_directional_is_sky_only(light)) { + light_count--; + continue; + } + + l.attenuation = storage->light_get_param(light, RS::LIGHT_PARAM_ATTENUATION); + l.energy = storage->light_get_param(light, RS::LIGHT_PARAM_ENERGY) * storage->light_get_param(light, RS::LIGHT_PARAM_INDIRECT_ENERGY); + l.radius = to_cell.basis.xform(Vector3(storage->light_get_param(light, RS::LIGHT_PARAM_RANGE), 0, 0)).length(); + Color color = storage->light_get_color(light).to_linear(); + l.color[0] = color.r; + l.color[1] = color.g; + l.color[2] = color.b; + + l.cos_spot_angle = Math::cos(Math::deg2rad(storage->light_get_param(light, RS::LIGHT_PARAM_SPOT_ANGLE))); + l.inv_spot_attenuation = 1.0f / storage->light_get_param(light, RS::LIGHT_PARAM_SPOT_ATTENUATION); + + Transform xform = p_scene_render->light_instance_get_base_transform(light_instance); + + Vector3 pos = to_probe_xform.xform(xform.origin); + Vector3 dir = to_probe_xform.basis.xform(-xform.basis.get_axis(2)).normalized(); + + l.position[0] = pos.x; + l.position[1] = pos.y; + l.position[2] = pos.z; + + l.direction[0] = dir.x; + l.direction[1] = dir.y; + l.direction[2] = dir.z; + + l.has_shadow = storage->light_has_shadow(light); + } + + RD::get_singleton()->buffer_update(gi->gi_probe_lights_uniform, 0, sizeof(GIProbeLight) * light_count, gi->gi_probe_lights); + } + } + + if (has_dynamic_object_data || p_update_light_instances || p_dynamic_objects.size()) { + // PROCESS MIPMAPS + if (mipmaps.size()) { + //can update mipmaps + + Vector3i probe_size = storage->gi_probe_get_octree_size(probe); + + GIProbePushConstant push_constant; + + push_constant.limits[0] = probe_size.x; + push_constant.limits[1] = probe_size.y; + push_constant.limits[2] = probe_size.z; + push_constant.stack_size = mipmaps.size(); + push_constant.emission_scale = 1.0; + push_constant.propagation = storage->gi_probe_get_propagation(probe); + push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(probe); + push_constant.light_count = light_count; + push_constant.aniso_strength = 0; + + /* print_line("probe update to version " + itos(last_probe_version)); + print_line("propagation " + rtos(push_constant.propagation)); + print_line("dynrange " + rtos(push_constant.dynamic_range)); + */ + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + + int passes; + if (p_update_light_instances) { + passes = storage->gi_probe_is_using_two_bounces(probe) ? 2 : 1; + } else { + passes = 1; //only re-blitting is necessary + } + int wg_size = 64; + int wg_limit_x = RD::get_singleton()->limit_get(RD::LIMIT_MAX_COMPUTE_WORKGROUP_COUNT_X); + + for (int pass = 0; pass < passes; pass++) { + if (p_update_light_instances) { + for (int i = 0; i < mipmaps.size(); i++) { + if (i == 0) { + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[pass == 0 ? GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT : GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE]); + } else if (i == 1) { + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP]); + } + + if (pass == 1 || i > 0) { + RD::get_singleton()->compute_list_add_barrier(compute_list); //wait til previous step is done + } + if (pass == 0 || i > 0) { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, mipmaps[i].uniform_set, 0); + } else { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, mipmaps[i].second_bounce_uniform_set, 0); + } + + push_constant.cell_offset = mipmaps[i].cell_offset; + push_constant.cell_count = mipmaps[i].cell_count; + + int wg_todo = (mipmaps[i].cell_count - 1) / wg_size + 1; + while (wg_todo) { + int wg_count = MIN(wg_todo, wg_limit_x); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbePushConstant)); + RD::get_singleton()->compute_list_dispatch(compute_list, wg_count, 1, 1); + wg_todo -= wg_count; + push_constant.cell_offset += wg_count * wg_size; + } + } + + RD::get_singleton()->compute_list_add_barrier(compute_list); //wait til previous step is done + } + + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_WRITE_TEXTURE]); + + for (int i = 0; i < mipmaps.size(); i++) { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, mipmaps[i].write_uniform_set, 0); + + push_constant.cell_offset = mipmaps[i].cell_offset; + push_constant.cell_count = mipmaps[i].cell_count; + + int wg_todo = (mipmaps[i].cell_count - 1) / wg_size + 1; + while (wg_todo) { + int wg_count = MIN(wg_todo, wg_limit_x); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbePushConstant)); + RD::get_singleton()->compute_list_dispatch(compute_list, wg_count, 1, 1); + wg_todo -= wg_count; + push_constant.cell_offset += wg_count * wg_size; + } + } + } + + RD::get_singleton()->compute_list_end(); + } + } + + has_dynamic_object_data = false; //clear until dynamic object data is used again + + if (p_dynamic_objects.size() && dynamic_maps.size()) { + Vector3i octree_size = storage->gi_probe_get_octree_size(probe); + int multiplier = dynamic_maps[0].size / MAX(MAX(octree_size.x, octree_size.y), octree_size.z); + + Transform oversample_scale; + oversample_scale.basis.scale(Vector3(multiplier, multiplier, multiplier)); + + Transform to_cell = oversample_scale * storage->gi_probe_get_to_cell_xform(probe); + Transform to_world_xform = transform * to_cell.affine_inverse(); + Transform to_probe_xform = to_world_xform.affine_inverse(); + + AABB probe_aabb(Vector3(), octree_size); + + //this could probably be better parallelized in compute.. + for (int i = 0; i < (int)p_dynamic_objects.size(); i++) { + RendererSceneRender::GeometryInstance *instance = p_dynamic_objects[i]; + + //transform aabb to giprobe + AABB aabb = (to_probe_xform * p_scene_render->geometry_instance_get_transform(instance)).xform(p_scene_render->geometry_instance_get_aabb(instance)); + + //this needs to wrap to grid resolution to avoid jitter + //also extend margin a bit just in case + Vector3i begin = aabb.position - Vector3i(1, 1, 1); + Vector3i end = aabb.position + aabb.size + Vector3i(1, 1, 1); + + for (int j = 0; j < 3; j++) { + if ((end[j] - begin[j]) & 1) { + end[j]++; //for half extents split, it needs to be even + } + begin[j] = MAX(begin[j], 0); + end[j] = MIN(end[j], octree_size[j] * multiplier); + } + + //aabb = aabb.intersection(probe_aabb); //intersect + aabb.position = begin; + aabb.size = end - begin; + + //print_line("aabb: " + aabb); + + for (int j = 0; j < 6; j++) { + //if (j != 0 && j != 3) { + // continue; + //} + static const Vector3 render_z[6] = { + Vector3(1, 0, 0), + Vector3(0, 1, 0), + Vector3(0, 0, 1), + Vector3(-1, 0, 0), + Vector3(0, -1, 0), + Vector3(0, 0, -1), + }; + static const Vector3 render_up[6] = { + Vector3(0, 1, 0), + Vector3(0, 0, 1), + Vector3(0, 1, 0), + Vector3(0, 1, 0), + Vector3(0, 0, 1), + Vector3(0, 1, 0), + }; + + Vector3 render_dir = render_z[j]; + Vector3 up_dir = render_up[j]; + + Vector3 center = aabb.position + aabb.size * 0.5; + Transform xform; + xform.set_look_at(center - aabb.size * 0.5 * render_dir, center, up_dir); + + Vector3 x_dir = xform.basis.get_axis(0).abs(); + int x_axis = int(Vector3(0, 1, 2).dot(x_dir)); + Vector3 y_dir = xform.basis.get_axis(1).abs(); + int y_axis = int(Vector3(0, 1, 2).dot(y_dir)); + Vector3 z_dir = -xform.basis.get_axis(2); + int z_axis = int(Vector3(0, 1, 2).dot(z_dir.abs())); + + Rect2i rect(aabb.position[x_axis], aabb.position[y_axis], aabb.size[x_axis], aabb.size[y_axis]); + bool x_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(0)) < 0); + bool y_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(1)) < 0); + bool z_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(2)) > 0); + + CameraMatrix cm; + cm.set_orthogonal(-rect.size.width / 2, rect.size.width / 2, -rect.size.height / 2, rect.size.height / 2, 0.0001, aabb.size[z_axis]); + + if (p_scene_render->cull_argument.size() == 0) { + p_scene_render->cull_argument.push_back(nullptr); + } + p_scene_render->cull_argument[0] = instance; + + p_scene_render->_render_material(to_world_xform * xform, cm, true, p_scene_render->cull_argument, dynamic_maps[0].fb, Rect2i(Vector2i(), rect.size)); + + GIProbeDynamicPushConstant push_constant; + zeromem(&push_constant, sizeof(GIProbeDynamicPushConstant)); + push_constant.limits[0] = octree_size.x; + push_constant.limits[1] = octree_size.y; + push_constant.limits[2] = octree_size.z; + push_constant.light_count = p_light_instances.size(); + push_constant.x_dir[0] = x_dir[0]; + push_constant.x_dir[1] = x_dir[1]; + push_constant.x_dir[2] = x_dir[2]; + push_constant.y_dir[0] = y_dir[0]; + push_constant.y_dir[1] = y_dir[1]; + push_constant.y_dir[2] = y_dir[2]; + push_constant.z_dir[0] = z_dir[0]; + push_constant.z_dir[1] = z_dir[1]; + push_constant.z_dir[2] = z_dir[2]; + push_constant.z_base = xform.origin[z_axis]; + push_constant.z_sign = (z_flip ? -1.0 : 1.0); + push_constant.pos_multiplier = float(1.0) / multiplier; + push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(probe); + push_constant.flip_x = x_flip; + push_constant.flip_y = y_flip; + push_constant.rect_pos[0] = rect.position[0]; + push_constant.rect_pos[1] = rect.position[1]; + push_constant.rect_size[0] = rect.size[0]; + push_constant.rect_size[1] = rect.size[1]; + push_constant.prev_rect_ofs[0] = 0; + push_constant.prev_rect_ofs[1] = 0; + push_constant.prev_rect_size[0] = 0; + push_constant.prev_rect_size[1] = 0; + push_constant.on_mipmap = false; + push_constant.propagation = storage->gi_probe_get_propagation(probe); + push_constant.pad[0] = 0; + push_constant.pad[1] = 0; + push_constant.pad[2] = 0; + + //process lighting + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, dynamic_maps[0].uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbeDynamicPushConstant)); + RD::get_singleton()->compute_list_dispatch(compute_list, (rect.size.x - 1) / 8 + 1, (rect.size.y - 1) / 8 + 1, 1); + //print_line("rect: " + itos(i) + ": " + rect); + + for (int k = 1; k < dynamic_maps.size(); k++) { + // enlarge the rect if needed so all pixels fit when downscaled, + // this ensures downsampling is smooth and optimal because no pixels are left behind + + //x + if (rect.position.x & 1) { + rect.size.x++; + push_constant.prev_rect_ofs[0] = 1; //this is used to ensure reading is also optimal + } else { + push_constant.prev_rect_ofs[0] = 0; + } + if (rect.size.x & 1) { + rect.size.x++; + } + + rect.position.x >>= 1; + rect.size.x = MAX(1, rect.size.x >> 1); + + //y + if (rect.position.y & 1) { + rect.size.y++; + push_constant.prev_rect_ofs[1] = 1; + } else { + push_constant.prev_rect_ofs[1] = 0; + } + if (rect.size.y & 1) { + rect.size.y++; + } + + rect.position.y >>= 1; + rect.size.y = MAX(1, rect.size.y >> 1); + + //shrink limits to ensure plot does not go outside map + if (dynamic_maps[k].mipmap > 0) { + for (int l = 0; l < 3; l++) { + push_constant.limits[l] = MAX(1, push_constant.limits[l] >> 1); + } + } + + //print_line("rect: " + itos(i) + ": " + rect); + push_constant.rect_pos[0] = rect.position[0]; + push_constant.rect_pos[1] = rect.position[1]; + push_constant.prev_rect_size[0] = push_constant.rect_size[0]; + push_constant.prev_rect_size[1] = push_constant.rect_size[1]; + push_constant.rect_size[0] = rect.size[0]; + push_constant.rect_size[1] = rect.size[1]; + push_constant.on_mipmap = dynamic_maps[k].mipmap > 0; + + RD::get_singleton()->compute_list_add_barrier(compute_list); + + if (dynamic_maps[k].mipmap < 0) { + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE]); + } else if (k < dynamic_maps.size() - 1) { + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT]); + } else { + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi->giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT]); + } + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, dynamic_maps[k].uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbeDynamicPushConstant)); + RD::get_singleton()->compute_list_dispatch(compute_list, (rect.size.x - 1) / 8 + 1, (rect.size.y - 1) / 8 + 1, 1); + } + + RD::get_singleton()->compute_list_end(); + } + } + + has_dynamic_object_data = true; //clear until dynamic object data is used again + } + + last_probe_version = storage->gi_probe_get_version(probe); +} + +void RendererSceneGIRD::GIProbeInstance::debug(RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha) { + if (mipmaps.size() == 0) { + return; + } + + CameraMatrix cam_transform = (p_camera_with_transform * CameraMatrix(transform)) * CameraMatrix(storage->gi_probe_get_to_cell_xform(probe).affine_inverse()); + + int level = 0; + Vector3i octree_size = storage->gi_probe_get_octree_size(probe); + + GIProbeDebugPushConstant push_constant; + push_constant.alpha = p_alpha; + push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(probe); + push_constant.cell_offset = mipmaps[level].cell_offset; + push_constant.level = level; + + push_constant.bounds[0] = octree_size.x >> level; + push_constant.bounds[1] = octree_size.y >> level; + push_constant.bounds[2] = octree_size.z >> level; + push_constant.pad = 0; + + for (int i = 0; i < 4; i++) { + for (int j = 0; j < 4; j++) { + push_constant.projection[i * 4 + j] = cam_transform.matrix[i][j]; + } + } + + if (gi->giprobe_debug_uniform_set.is_valid()) { + RD::get_singleton()->free(gi->giprobe_debug_uniform_set); + } + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(storage->gi_probe_get_data_buffer(probe)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 2; + u.ids.push_back(texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 3; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + int cell_count; + if (!p_emission && p_lighting && has_dynamic_object_data) { + cell_count = push_constant.bounds[0] * push_constant.bounds[1] * push_constant.bounds[2]; + } else { + cell_count = mipmaps[level].cell_count; + } + + gi->giprobe_debug_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi->giprobe_debug_shader_version_shaders[0], 0); + + int giprobe_debug_pipeline = GI_PROBE_DEBUG_COLOR; + if (p_emission) { + giprobe_debug_pipeline = GI_PROBE_DEBUG_EMISSION; + } else if (p_lighting) { + giprobe_debug_pipeline = has_dynamic_object_data ? GI_PROBE_DEBUG_LIGHT_FULL : GI_PROBE_DEBUG_LIGHT; + } + RD::get_singleton()->draw_list_bind_render_pipeline( + p_draw_list, + gi->giprobe_debug_shader_version_pipelines[giprobe_debug_pipeline].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, gi->giprobe_debug_uniform_set, 0); + RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(GIProbeDebugPushConstant)); + RD::get_singleton()->draw_list_draw(p_draw_list, false, cell_count, 36); +} + +//////////////////////////////////////////////////////////////////////////////// +// GIRD + +RendererSceneGIRD::RendererSceneGIRD() { + sdfgi_ray_count = RS::EnvironmentSDFGIRayCount(CLAMP(int32_t(GLOBAL_GET("rendering/global_illumination/sdfgi/probe_ray_count")), 0, int32_t(RS::ENV_SDFGI_RAY_COUNT_MAX - 1))); + sdfgi_frames_to_converge = RS::EnvironmentSDFGIFramesToConverge(CLAMP(int32_t(GLOBAL_GET("rendering/global_illumination/sdfgi/frames_to_converge")), 0, int32_t(RS::ENV_SDFGI_CONVERGE_MAX - 1))); + sdfgi_frames_to_update_light = RS::EnvironmentSDFGIFramesToUpdateLight(CLAMP(int32_t(GLOBAL_GET("rendering/global_illumination/sdfgi/frames_to_update_lights")), 0, int32_t(RS::ENV_SDFGI_UPDATE_LIGHT_MAX - 1))); +} + +RendererSceneGIRD::~RendererSceneGIRD() { +} + +void RendererSceneGIRD::init(RendererStorageRD *p_storage, RendererSceneSkyRD *p_sky) { + storage = p_storage; + + /* GI */ + + { + //kinda complicated to compute the amount of slots, we try to use as many as we can + + gi_probe_max_lights = 32; + + gi_probe_lights = memnew_arr(GIProbeLight, gi_probe_max_lights); + gi_probe_lights_uniform = RD::get_singleton()->uniform_buffer_create(gi_probe_max_lights * sizeof(GIProbeLight)); + gi_probe_quality = RS::GIProbeQuality(CLAMP(int(GLOBAL_GET("rendering/global_illumination/gi_probes/quality")), 0, 1)); + + String defines = "\n#define MAX_LIGHTS " + itos(gi_probe_max_lights) + "\n"; + + Vector<String> versions; + versions.push_back("\n#define MODE_COMPUTE_LIGHT\n"); + versions.push_back("\n#define MODE_SECOND_BOUNCE\n"); + versions.push_back("\n#define MODE_UPDATE_MIPMAPS\n"); + versions.push_back("\n#define MODE_WRITE_TEXTURE\n"); + versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_LIGHTING\n"); + versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_WRITE\n"); + versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_PLOT\n"); + versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_PLOT\n#define MODE_DYNAMIC_SHRINK_WRITE\n"); + + giprobe_shader.initialize(versions, defines); + giprobe_lighting_shader_version = giprobe_shader.version_create(); + for (int i = 0; i < GI_PROBE_SHADER_VERSION_MAX; i++) { + giprobe_lighting_shader_version_shaders[i] = giprobe_shader.version_get_shader(giprobe_lighting_shader_version, i); + giprobe_lighting_shader_version_pipelines[i] = RD::get_singleton()->compute_pipeline_create(giprobe_lighting_shader_version_shaders[i]); + } + } + + { + String defines; + Vector<String> versions; + versions.push_back("\n#define MODE_DEBUG_COLOR\n"); + versions.push_back("\n#define MODE_DEBUG_LIGHT\n"); + versions.push_back("\n#define MODE_DEBUG_EMISSION\n"); + versions.push_back("\n#define MODE_DEBUG_LIGHT\n#define MODE_DEBUG_LIGHT_FULL\n"); + + giprobe_debug_shader.initialize(versions, defines); + giprobe_debug_shader_version = giprobe_debug_shader.version_create(); + for (int i = 0; i < GI_PROBE_DEBUG_MAX; i++) { + giprobe_debug_shader_version_shaders[i] = giprobe_debug_shader.version_get_shader(giprobe_debug_shader_version, i); + + RD::PipelineRasterizationState rs; + rs.cull_mode = RD::POLYGON_CULL_FRONT; + RD::PipelineDepthStencilState ds; + ds.enable_depth_test = true; + ds.enable_depth_write = true; + ds.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; + + giprobe_debug_shader_version_pipelines[i].setup(giprobe_debug_shader_version_shaders[i], RD::RENDER_PRIMITIVE_TRIANGLES, rs, RD::PipelineMultisampleState(), ds, RD::PipelineColorBlendState::create_disabled(), 0); + } + } + + /* SDGFI */ + + { + Vector<String> preprocess_modes; + preprocess_modes.push_back("\n#define MODE_SCROLL\n"); + preprocess_modes.push_back("\n#define MODE_SCROLL_OCCLUSION\n"); + preprocess_modes.push_back("\n#define MODE_INITIALIZE_JUMP_FLOOD\n"); + preprocess_modes.push_back("\n#define MODE_INITIALIZE_JUMP_FLOOD_HALF\n"); + preprocess_modes.push_back("\n#define MODE_JUMPFLOOD\n"); + preprocess_modes.push_back("\n#define MODE_JUMPFLOOD_OPTIMIZED\n"); + preprocess_modes.push_back("\n#define MODE_UPSCALE_JUMP_FLOOD\n"); + preprocess_modes.push_back("\n#define MODE_OCCLUSION\n"); + preprocess_modes.push_back("\n#define MODE_STORE\n"); + String defines = "\n#define OCCLUSION_SIZE " + itos(SDFGI::CASCADE_SIZE / SDFGI::PROBE_DIVISOR) + "\n"; + sdfgi_shader.preprocess.initialize(preprocess_modes, defines); + sdfgi_shader.preprocess_shader = sdfgi_shader.preprocess.version_create(); + for (int i = 0; i < SDFGIShader::PRE_PROCESS_MAX; i++) { + sdfgi_shader.preprocess_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, i)); + } + } + + { + //calculate tables + String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; + + Vector<String> direct_light_modes; + direct_light_modes.push_back("\n#define MODE_PROCESS_STATIC\n"); + direct_light_modes.push_back("\n#define MODE_PROCESS_DYNAMIC\n"); + sdfgi_shader.direct_light.initialize(direct_light_modes, defines); + sdfgi_shader.direct_light_shader = sdfgi_shader.direct_light.version_create(); + for (int i = 0; i < SDFGIShader::DIRECT_LIGHT_MODE_MAX; i++) { + sdfgi_shader.direct_light_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.direct_light.version_get_shader(sdfgi_shader.direct_light_shader, i)); + } + } + + { + //calculate tables + String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; + defines += "\n#define SH_SIZE " + itos(SDFGI::SH_SIZE) + "\n"; + if (p_sky->sky_use_cubemap_array) { + defines += "\n#define USE_CUBEMAP_ARRAY\n"; + } + + Vector<String> integrate_modes; + integrate_modes.push_back("\n#define MODE_PROCESS\n"); + integrate_modes.push_back("\n#define MODE_STORE\n"); + integrate_modes.push_back("\n#define MODE_SCROLL\n"); + integrate_modes.push_back("\n#define MODE_SCROLL_STORE\n"); + sdfgi_shader.integrate.initialize(integrate_modes, defines); + sdfgi_shader.integrate_shader = sdfgi_shader.integrate.version_create(); + + for (int i = 0; i < SDFGIShader::INTEGRATE_MODE_MAX; i++) { + sdfgi_shader.integrate_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, i)); + } + + { + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 0; + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_WHITE)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 1; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + sdfgi_shader.integrate_default_sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, 0), 1); + } + } + + //GK + { + //calculate tables + String defines = "\n#define SDFGI_OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; + Vector<String> gi_modes; + gi_modes.push_back("\n#define USE_GIPROBES\n"); + gi_modes.push_back("\n#define USE_SDFGI\n"); + gi_modes.push_back("\n#define USE_SDFGI\n\n#define USE_GIPROBES\n"); + gi_modes.push_back("\n#define MODE_HALF_RES\n#define USE_GIPROBES\n"); + gi_modes.push_back("\n#define MODE_HALF_RES\n#define USE_SDFGI\n"); + gi_modes.push_back("\n#define MODE_HALF_RES\n#define USE_SDFGI\n\n#define USE_GIPROBES\n"); + + shader.initialize(gi_modes, defines); + shader_version = shader.version_create(); + for (int i = 0; i < MODE_MAX; i++) { + pipelines[i] = RD::get_singleton()->compute_pipeline_create(shader.version_get_shader(shader_version, i)); + } + + sdfgi_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(SDFGIData)); + } + { + String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; + Vector<String> debug_modes; + debug_modes.push_back(""); + sdfgi_shader.debug.initialize(debug_modes, defines); + sdfgi_shader.debug_shader = sdfgi_shader.debug.version_create(); + sdfgi_shader.debug_shader_version = sdfgi_shader.debug.version_get_shader(sdfgi_shader.debug_shader, 0); + sdfgi_shader.debug_pipeline = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.debug_shader_version); + } + { + String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; + + Vector<String> versions; + versions.push_back("\n#define MODE_PROBES\n"); + versions.push_back("\n#define MODE_VISIBILITY\n"); + + sdfgi_shader.debug_probes.initialize(versions, defines); + sdfgi_shader.debug_probes_shader = sdfgi_shader.debug_probes.version_create(); + + { + RD::PipelineRasterizationState rs; + rs.cull_mode = RD::POLYGON_CULL_DISABLED; + RD::PipelineDepthStencilState ds; + ds.enable_depth_test = true; + ds.enable_depth_write = true; + ds.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; + for (int i = 0; i < SDFGIShader::PROBE_DEBUG_MAX; i++) { + RID debug_probes_shader_version = sdfgi_shader.debug_probes.version_get_shader(sdfgi_shader.debug_probes_shader, i); + sdfgi_shader.debug_probes_pipeline[i].setup(debug_probes_shader_version, RD::RENDER_PRIMITIVE_TRIANGLE_STRIPS, rs, RD::PipelineMultisampleState(), ds, RD::PipelineColorBlendState::create_disabled(), 0); + } + } + } + default_giprobe_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(GIProbeData) * MAX_GIPROBES); + half_resolution = GLOBAL_GET("rendering/global_illumination/gi/use_half_resolution"); +} + +void RendererSceneGIRD::free() { + RD::get_singleton()->free(default_giprobe_buffer); + RD::get_singleton()->free(gi_probe_lights_uniform); + RD::get_singleton()->free(sdfgi_ubo); + + giprobe_debug_shader.version_free(giprobe_debug_shader_version); + giprobe_shader.version_free(giprobe_lighting_shader_version); + shader.version_free(shader_version); + sdfgi_shader.debug_probes.version_free(sdfgi_shader.debug_probes_shader); + sdfgi_shader.debug.version_free(sdfgi_shader.debug_shader); + sdfgi_shader.direct_light.version_free(sdfgi_shader.direct_light_shader); + sdfgi_shader.integrate.version_free(sdfgi_shader.integrate_shader); + sdfgi_shader.preprocess.version_free(sdfgi_shader.preprocess_shader); + + memdelete_arr(gi_probe_lights); +} + +RendererSceneGIRD::SDFGI *RendererSceneGIRD::create_sdfgi(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size) { + SDFGI *sdfgi = memnew(SDFGI); + + sdfgi->create(p_env, p_world_position, p_requested_history_size, this); + + return sdfgi; +} + +void RendererSceneGIRD::setup_giprobes(RID p_render_buffers, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, uint32_t &r_gi_probes_used, RendererSceneRenderRD *p_scene_render) { + r_gi_probes_used = 0; + + // feels a little dirty to use our container this way but.... + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND(rb == nullptr); + + RID gi_probe_buffer = p_scene_render->render_buffers_get_gi_probe_buffer(p_render_buffers); + + RD::get_singleton()->draw_command_begin_label("GIProbes Setup"); + + GIProbeData gi_probe_data[MAX_GIPROBES]; + + bool giprobes_changed = false; + + Transform to_camera; + to_camera.origin = p_transform.origin; //only translation, make local + + for (int i = 0; i < MAX_GIPROBES; i++) { + RID texture; + if (i < (int)p_gi_probes.size()) { + GIProbeInstance *gipi = get_probe_instance(p_gi_probes[i]); + + if (gipi) { + texture = gipi->texture; + GIProbeData &gipd = gi_probe_data[i]; + + RID base_probe = gipi->probe; + + Transform to_cell = storage->gi_probe_get_to_cell_xform(gipi->probe) * gipi->transform.affine_inverse() * to_camera; + + gipd.xform[0] = to_cell.basis.elements[0][0]; + gipd.xform[1] = to_cell.basis.elements[1][0]; + gipd.xform[2] = to_cell.basis.elements[2][0]; + gipd.xform[3] = 0; + gipd.xform[4] = to_cell.basis.elements[0][1]; + gipd.xform[5] = to_cell.basis.elements[1][1]; + gipd.xform[6] = to_cell.basis.elements[2][1]; + gipd.xform[7] = 0; + gipd.xform[8] = to_cell.basis.elements[0][2]; + gipd.xform[9] = to_cell.basis.elements[1][2]; + gipd.xform[10] = to_cell.basis.elements[2][2]; + gipd.xform[11] = 0; + gipd.xform[12] = to_cell.origin.x; + gipd.xform[13] = to_cell.origin.y; + gipd.xform[14] = to_cell.origin.z; + gipd.xform[15] = 1; + + Vector3 bounds = storage->gi_probe_get_octree_size(base_probe); + + gipd.bounds[0] = bounds.x; + gipd.bounds[1] = bounds.y; + gipd.bounds[2] = bounds.z; + + gipd.dynamic_range = storage->gi_probe_get_dynamic_range(base_probe) * storage->gi_probe_get_energy(base_probe); + gipd.bias = storage->gi_probe_get_bias(base_probe); + gipd.normal_bias = storage->gi_probe_get_normal_bias(base_probe); + gipd.blend_ambient = !storage->gi_probe_is_interior(base_probe); + gipd.anisotropy_strength = 0; + gipd.ao = storage->gi_probe_get_ao(base_probe); + gipd.ao_size = Math::pow(storage->gi_probe_get_ao_size(base_probe), 4.0f); + gipd.mipmaps = gipi->mipmaps.size(); + } + + r_gi_probes_used++; + } + + if (texture == RID()) { + texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); + } + + if (texture != rb->gi.giprobe_textures[i]) { + giprobes_changed = true; + rb->gi.giprobe_textures[i] = texture; + } + } + + if (giprobes_changed) { + if (RD::get_singleton()->uniform_set_is_valid(rb->gi.uniform_set)) { + RD::get_singleton()->free(rb->gi.uniform_set); + } + rb->gi.uniform_set = RID(); + if (rb->volumetric_fog) { + if (RD::get_singleton()->uniform_set_is_valid(rb->volumetric_fog->uniform_set)) { + RD::get_singleton()->free(rb->volumetric_fog->uniform_set); + RD::get_singleton()->free(rb->volumetric_fog->uniform_set2); + } + rb->volumetric_fog->uniform_set = RID(); + rb->volumetric_fog->uniform_set2 = RID(); + } + } + + if (p_gi_probes.size() > 0) { + RD::get_singleton()->buffer_update(gi_probe_buffer, 0, sizeof(GIProbeData) * MIN((uint64_t)MAX_GIPROBES, p_gi_probes.size()), gi_probe_data, RD::BARRIER_MASK_COMPUTE); + } + + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneGIRD::process_gi(RID p_render_buffers, RID p_normal_roughness_buffer, RID p_gi_probe_buffer, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, RendererSceneRenderRD *p_scene_render) { + RD::get_singleton()->draw_command_begin_label("GI Render"); + + RendererSceneRenderRD::RenderBuffers *rb = p_scene_render->render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND(rb == nullptr); + RendererSceneEnvironmentRD *env = p_scene_render->environment_owner.getornull(p_environment); + + if (rb->ambient_buffer.is_null() || rb->gi.using_half_size_gi != half_resolution) { + if (rb->ambient_buffer.is_valid()) { + RD::get_singleton()->free(rb->ambient_buffer); + RD::get_singleton()->free(rb->reflection_buffer); + } + + RD::TextureFormat tf; + tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.width = rb->width; + tf.height = rb->height; + if (half_resolution) { + tf.width >>= 1; + tf.height >>= 1; + } + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + rb->reflection_buffer = RD::get_singleton()->texture_create(tf, RD::TextureView()); + rb->ambient_buffer = RD::get_singleton()->texture_create(tf, RD::TextureView()); + rb->gi.using_half_size_gi = half_resolution; + + p_scene_render->_render_buffers_uniform_set_changed(p_render_buffers); + } + + PushConstant push_constant; + + push_constant.screen_size[0] = rb->width; + push_constant.screen_size[1] = rb->height; + push_constant.z_near = p_projection.get_z_near(); + push_constant.z_far = p_projection.get_z_far(); + push_constant.orthogonal = p_projection.is_orthogonal(); + push_constant.proj_info[0] = -2.0f / (rb->width * p_projection.matrix[0][0]); + push_constant.proj_info[1] = -2.0f / (rb->height * p_projection.matrix[1][1]); + push_constant.proj_info[2] = (1.0f - p_projection.matrix[0][2]) / p_projection.matrix[0][0]; + push_constant.proj_info[3] = (1.0f + p_projection.matrix[1][2]) / p_projection.matrix[1][1]; + push_constant.max_giprobes = MIN((uint64_t)MAX_GIPROBES, p_gi_probes.size()); + push_constant.high_quality_vct = gi_probe_quality == RS::GI_PROBE_QUALITY_HIGH; + + bool use_sdfgi = rb->sdfgi != nullptr; + bool use_giprobes = push_constant.max_giprobes > 0; + + if (env) { + push_constant.ao_color[0] = env->ao_color.r; + push_constant.ao_color[1] = env->ao_color.g; + push_constant.ao_color[2] = env->ao_color.b; + } else { + push_constant.ao_color[0] = 0; + push_constant.ao_color[1] = 0; + push_constant.ao_color[2] = 0; + } + + push_constant.cam_rotation[0] = p_transform.basis[0][0]; + push_constant.cam_rotation[1] = p_transform.basis[1][0]; + push_constant.cam_rotation[2] = p_transform.basis[2][0]; + push_constant.cam_rotation[3] = 0; + push_constant.cam_rotation[4] = p_transform.basis[0][1]; + push_constant.cam_rotation[5] = p_transform.basis[1][1]; + push_constant.cam_rotation[6] = p_transform.basis[2][1]; + push_constant.cam_rotation[7] = 0; + push_constant.cam_rotation[8] = p_transform.basis[0][2]; + push_constant.cam_rotation[9] = p_transform.basis[1][2]; + push_constant.cam_rotation[10] = p_transform.basis[2][2]; + push_constant.cam_rotation[11] = 0; + + if (rb->gi.uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(rb->gi.uniform_set)) { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.binding = 1; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { + u.ids.push_back(rb->sdfgi->cascades[j].sdf_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { + u.ids.push_back(rb->sdfgi->cascades[j].light_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { + u.ids.push_back(rb->sdfgi->cascades[j].light_aniso_0_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.binding = 4; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { + if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { + u.ids.push_back(rb->sdfgi->cascades[j].light_aniso_1_tex); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 5; + if (rb->sdfgi) { + u.ids.push_back(rb->sdfgi->occlusion_texture); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 6; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 7; + u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 9; + u.ids.push_back(rb->ambient_buffer); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_IMAGE; + u.binding = 10; + u.ids.push_back(rb->reflection_buffer); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 11; + if (rb->sdfgi) { + u.ids.push_back(rb->sdfgi->lightprobe_texture); + } else { + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_2D_ARRAY_WHITE)); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 12; + u.ids.push_back(rb->depth_texture); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 13; + u.ids.push_back(p_normal_roughness_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 14; + RID buffer = p_gi_probe_buffer.is_valid() ? p_gi_probe_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); + u.ids.push_back(buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 15; + u.ids.push_back(sdfgi_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 16; + u.ids.push_back(rb->gi.giprobe_buffer); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 17; + for (int i = 0; i < MAX_GIPROBES; i++) { + u.ids.push_back(rb->gi.giprobe_textures[i]); + } + uniforms.push_back(u); + } + + rb->gi.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, shader.version_get_shader(shader_version, 0), 0); + } + + Mode mode; + + if (rb->gi.using_half_size_gi) { + mode = (use_sdfgi && use_giprobes) ? MODE_HALF_RES_COMBINED : (use_sdfgi ? MODE_HALF_RES_SDFGI : MODE_HALF_RES_GIPROBE); + } else { + mode = (use_sdfgi && use_giprobes) ? MODE_COMBINED : (use_sdfgi ? MODE_SDFGI : MODE_GIPROBE); + } + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(true); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, pipelines[mode]); + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->gi.uniform_set, 0); + RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(PushConstant)); + + if (rb->gi.using_half_size_gi) { + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->width >> 1, rb->height >> 1, 1); + } else { + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->width, rb->height, 1); + } + //do barrier later to allow oeverlap + //RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_NO_BARRIER); //no barriers, let other compute, raster and transfer happen at the same time + RD::get_singleton()->draw_command_end_label(); +} + +RID RendererSceneGIRD::gi_probe_instance_create(RID p_base) { + GIProbeInstance gi_probe; + gi_probe.gi = this; + gi_probe.storage = storage; + gi_probe.probe = p_base; + RID rid = gi_probe_instance_owner.make_rid(gi_probe); + return rid; +} + +void RendererSceneGIRD::gi_probe_instance_set_transform_to_data(RID p_probe, const Transform &p_xform) { + GIProbeInstance *gi_probe = get_probe_instance(p_probe); + ERR_FAIL_COND(!gi_probe); + + gi_probe->transform = p_xform; +} + +bool RendererSceneGIRD::gi_probe_needs_update(RID p_probe) const { + GIProbeInstance *gi_probe = get_probe_instance(p_probe); + ERR_FAIL_COND_V(!gi_probe, false); + + return gi_probe->last_probe_version != storage->gi_probe_get_version(gi_probe->probe); +} + +void RendererSceneGIRD::gi_probe_update(RID p_probe, bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<RendererSceneRender::GeometryInstance *> &p_dynamic_objects, RendererSceneRenderRD *p_scene_render) { + GIProbeInstance *gi_probe = get_probe_instance(p_probe); + ERR_FAIL_COND(!gi_probe); + + gi_probe->update(p_update_light_instances, p_light_instances, p_dynamic_objects, p_scene_render); +} + +void RendererSceneGIRD::debug_giprobe(RID p_gi_probe, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha) { + GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_gi_probe); + ERR_FAIL_COND(!gi_probe); + + gi_probe->debug(p_draw_list, p_framebuffer, p_camera_with_transform, p_lighting, p_emission, p_alpha); +} diff --git a/servers/rendering/renderer_rd/renderer_scene_gi_rd.h b/servers/rendering/renderer_rd/renderer_scene_gi_rd.h new file mode 100644 index 0000000000..df20011b23 --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_gi_rd.h @@ -0,0 +1,671 @@ +/*************************************************************************/ +/* renderer_scene_gi_rd.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef RENDERING_SERVER_SCENE_GI_RD_H +#define RENDERING_SERVER_SCENE_GI_RD_H + +#include "core/templates/local_vector.h" +#include "core/templates/rid_owner.h" +#include "servers/rendering/renderer_compositor.h" +#include "servers/rendering/renderer_rd/renderer_scene_environment_rd.h" +#include "servers/rendering/renderer_rd/renderer_scene_sky_rd.h" +#include "servers/rendering/renderer_rd/renderer_storage_rd.h" +#include "servers/rendering/renderer_rd/shaders/gi.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/giprobe.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/giprobe_debug.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl.gen.h" +#include "servers/rendering/renderer_scene_render.h" +#include "servers/rendering/rendering_device.h" + +// Forward declare RendererSceneRenderRD so we can pass it into some of our methods, these classes are pretty tightly bound +class RendererSceneRenderRD; + +class RendererSceneGIRD { +private: + RendererStorageRD *storage; + + /* GIPROBE INSTANCE */ + + struct GIProbeLight { + uint32_t type; + float energy; + float radius; + float attenuation; + + float color[3]; + float cos_spot_angle; + + float position[3]; + float inv_spot_attenuation; + + float direction[3]; + uint32_t has_shadow; + }; + + struct GIProbePushConstant { + int32_t limits[3]; + uint32_t stack_size; + + float emission_scale; + float propagation; + float dynamic_range; + uint32_t light_count; + + uint32_t cell_offset; + uint32_t cell_count; + float aniso_strength; + uint32_t pad; + }; + + struct GIProbeDynamicPushConstant { + int32_t limits[3]; + uint32_t light_count; + int32_t x_dir[3]; + float z_base; + int32_t y_dir[3]; + float z_sign; + int32_t z_dir[3]; + float pos_multiplier; + uint32_t rect_pos[2]; + uint32_t rect_size[2]; + uint32_t prev_rect_ofs[2]; + uint32_t prev_rect_size[2]; + uint32_t flip_x; + uint32_t flip_y; + float dynamic_range; + uint32_t on_mipmap; + float propagation; + float pad[3]; + }; + + GIProbeLight *gi_probe_lights; + uint32_t gi_probe_max_lights; + RID gi_probe_lights_uniform; + + enum { + GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT, + GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE, + GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP, + GI_PROBE_SHADER_VERSION_WRITE_TEXTURE, + GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING, + GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE, + GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT, + GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT, + GI_PROBE_SHADER_VERSION_MAX + }; + + GiprobeShaderRD giprobe_shader; + RID giprobe_lighting_shader_version; + RID giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_MAX]; + RID giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_MAX]; + + enum { + GI_PROBE_DEBUG_COLOR, + GI_PROBE_DEBUG_LIGHT, + GI_PROBE_DEBUG_EMISSION, + GI_PROBE_DEBUG_LIGHT_FULL, + GI_PROBE_DEBUG_MAX + }; + + struct GIProbeDebugPushConstant { + float projection[16]; + uint32_t cell_offset; + float dynamic_range; + float alpha; + uint32_t level; + int32_t bounds[3]; + uint32_t pad; + }; + + GiprobeDebugShaderRD giprobe_debug_shader; + RID giprobe_debug_shader_version; + RID giprobe_debug_shader_version_shaders[GI_PROBE_DEBUG_MAX]; + PipelineCacheRD giprobe_debug_shader_version_pipelines[GI_PROBE_DEBUG_MAX]; + RID giprobe_debug_uniform_set; + + /* SDFGI */ + + struct SDFGIShader { + enum SDFGIPreprocessShaderVersion { + PRE_PROCESS_SCROLL, + PRE_PROCESS_SCROLL_OCCLUSION, + PRE_PROCESS_JUMP_FLOOD_INITIALIZE, + PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF, + PRE_PROCESS_JUMP_FLOOD, + PRE_PROCESS_JUMP_FLOOD_OPTIMIZED, + PRE_PROCESS_JUMP_FLOOD_UPSCALE, + PRE_PROCESS_OCCLUSION, + PRE_PROCESS_STORE, + PRE_PROCESS_MAX + }; + + struct PreprocessPushConstant { + int32_t scroll[3]; + int32_t grid_size; + + int32_t probe_offset[3]; + int32_t step_size; + + int32_t half_size; + uint32_t occlusion_index; + int32_t cascade; + uint32_t pad; + }; + + SdfgiPreprocessShaderRD preprocess; + RID preprocess_shader; + RID preprocess_pipeline[PRE_PROCESS_MAX]; + + struct DebugPushConstant { + float grid_size[3]; + uint32_t max_cascades; + + int32_t screen_size[2]; + uint32_t use_occlusion; + float y_mult; + + float cam_extent[3]; + uint32_t probe_axis_size; + + float cam_transform[16]; + }; + + SdfgiDebugShaderRD debug; + RID debug_shader; + RID debug_shader_version; + RID debug_pipeline; + + enum ProbeDebugMode { + PROBE_DEBUG_PROBES, + PROBE_DEBUG_VISIBILITY, + PROBE_DEBUG_MAX + }; + + struct DebugProbesPushConstant { + float projection[16]; + + uint32_t band_power; + uint32_t sections_in_band; + uint32_t band_mask; + float section_arc; + + float grid_size[3]; + uint32_t cascade; + + uint32_t pad; + float y_mult; + int32_t probe_debug_index; + int32_t probe_axis_size; + }; + + SdfgiDebugProbesShaderRD debug_probes; + RID debug_probes_shader; + RID debug_probes_shader_version; + + PipelineCacheRD debug_probes_pipeline[PROBE_DEBUG_MAX]; + + struct Light { + float color[3]; + float energy; + + float direction[3]; + uint32_t has_shadow; + + float position[3]; + float attenuation; + + uint32_t type; + float cos_spot_angle; + float inv_spot_attenuation; + float radius; + + float shadow_color[4]; + }; + + struct DirectLightPushConstant { + float grid_size[3]; + uint32_t max_cascades; + + uint32_t cascade; + uint32_t light_count; + uint32_t process_offset; + uint32_t process_increment; + + int32_t probe_axis_size; + float bounce_feedback; + float y_mult; + uint32_t use_occlusion; + }; + + enum { + DIRECT_LIGHT_MODE_STATIC, + DIRECT_LIGHT_MODE_DYNAMIC, + DIRECT_LIGHT_MODE_MAX + }; + SdfgiDirectLightShaderRD direct_light; + RID direct_light_shader; + RID direct_light_pipeline[DIRECT_LIGHT_MODE_MAX]; + + enum { + INTEGRATE_MODE_PROCESS, + INTEGRATE_MODE_STORE, + INTEGRATE_MODE_SCROLL, + INTEGRATE_MODE_SCROLL_STORE, + INTEGRATE_MODE_MAX + }; + struct IntegratePushConstant { + enum { + SKY_MODE_DISABLED, + SKY_MODE_COLOR, + SKY_MODE_SKY, + }; + + float grid_size[3]; + uint32_t max_cascades; + + uint32_t probe_axis_size; + uint32_t cascade; + uint32_t history_index; + uint32_t history_size; + + uint32_t ray_count; + float ray_bias; + int32_t image_size[2]; + + int32_t world_offset[3]; + uint32_t sky_mode; + + int32_t scroll[3]; + float sky_energy; + + float sky_color[3]; + float y_mult; + + uint32_t store_ambient_texture; + uint32_t pad[3]; + }; + + SdfgiIntegrateShaderRD integrate; + RID integrate_shader; + RID integrate_pipeline[INTEGRATE_MODE_MAX]; + + RID integrate_default_sky_uniform_set; + + } sdfgi_shader; + +public: + /* GIPROBE INSTANCE */ + + //@TODO GIProbeInstance is still directly used in the render code, we'll address this when we refactor the render code itself. + + struct GIProbeInstance { + // access to our containers + RendererStorageRD *storage; + RendererSceneGIRD *gi; + + RID probe; + RID texture; + RID write_buffer; + + struct Mipmap { + RID texture; + RID uniform_set; + RID second_bounce_uniform_set; + RID write_uniform_set; + uint32_t level; + uint32_t cell_offset; + uint32_t cell_count; + }; + Vector<Mipmap> mipmaps; + + struct DynamicMap { + RID texture; //color normally, or emission on first pass + RID fb_depth; //actual depth buffer for the first pass, float depth for later passes + RID depth; //actual depth buffer for the first pass, float depth for later passes + RID normal; //normal buffer for the first pass + RID albedo; //emission buffer for the first pass + RID orm; //orm buffer for the first pass + RID fb; //used for rendering, only valid on first map + RID uniform_set; + uint32_t size; + int mipmap; // mipmap to write to, -1 if no mipmap assigned + }; + + Vector<DynamicMap> dynamic_maps; + + int slot = -1; + uint32_t last_probe_version = 0; + uint32_t last_probe_data_version = 0; + + //uint64_t last_pass = 0; + uint32_t render_index = 0; + + bool has_dynamic_object_data = false; + + Transform transform; + + void update(bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<RendererSceneRender::GeometryInstance *> &p_dynamic_objects, RendererSceneRenderRD *p_scene_render); + void debug(RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha); + }; + + mutable RID_Owner<GIProbeInstance> gi_probe_instance_owner; + + _FORCE_INLINE_ GIProbeInstance *get_probe_instance(RID p_probe) const { + return gi_probe_instance_owner.getornull(p_probe); + }; + + _FORCE_INLINE_ RID gi_probe_instance_get_texture(RID p_probe) { + GIProbeInstance *gi_probe = get_probe_instance(p_probe); + ERR_FAIL_COND_V(!gi_probe, RID()); + return gi_probe->texture; + }; + + RS::GIProbeQuality gi_probe_quality = RS::GI_PROBE_QUALITY_HIGH; + + /* SDFGI */ + + struct SDFGI { + enum { + MAX_CASCADES = 8, + CASCADE_SIZE = 128, + PROBE_DIVISOR = 16, + ANISOTROPY_SIZE = 6, + MAX_DYNAMIC_LIGHTS = 128, + MAX_STATIC_LIGHTS = 1024, + LIGHTPROBE_OCT_SIZE = 6, + SH_SIZE = 16 + }; + + struct Cascade { + struct UBO { + float offset[3]; + float to_cell; + int32_t probe_offset[3]; + uint32_t pad; + }; + + //cascade blocks are full-size for volume (128^3), half size for albedo/emission + RID sdf_tex; + RID light_tex; + RID light_aniso_0_tex; + RID light_aniso_1_tex; + + RID light_data; + RID light_aniso_0_data; + RID light_aniso_1_data; + + struct SolidCell { // this struct is unused, but remains as reference for size + uint32_t position; + uint32_t albedo; + uint32_t static_light; + uint32_t static_light_aniso; + }; + + RID solid_cell_dispatch_buffer; //buffer for indirect compute dispatch + RID solid_cell_buffer; + + RID lightprobe_history_tex; + RID lightprobe_average_tex; + + float cell_size; + Vector3i position; + + static const Vector3i DIRTY_ALL; + Vector3i dirty_regions; //(0,0,0 is not dirty, negative is refresh from the end, DIRTY_ALL is refresh all. + + RID sdf_store_uniform_set; + RID sdf_direct_light_uniform_set; + RID scroll_uniform_set; + RID scroll_occlusion_uniform_set; + RID integrate_uniform_set; + RID lights_buffer; + + bool all_dynamic_lights_dirty = true; + }; + + // access to our containers + RendererStorageRD *storage; + RendererSceneGIRD *gi; + + // used for rendering (voxelization) + RID render_albedo; + RID render_emission; + RID render_emission_aniso; + RID render_occlusion[8]; + RID render_geom_facing; + + RID render_sdf[2]; + RID render_sdf_half[2]; + + // used for ping pong processing in cascades + RID sdf_initialize_uniform_set; + RID sdf_initialize_half_uniform_set; + RID jump_flood_uniform_set[2]; + RID jump_flood_half_uniform_set[2]; + RID sdf_upscale_uniform_set; + int upscale_jfa_uniform_set_index; + RID occlusion_uniform_set; + + uint32_t cascade_size = 128; + + LocalVector<Cascade> cascades; + + RID lightprobe_texture; + RID lightprobe_data; + RID occlusion_texture; + RID occlusion_data; + RID ambient_texture; //integrates with volumetric fog + + RID lightprobe_history_scroll; //used for scrolling lightprobes + RID lightprobe_average_scroll; //used for scrolling lightprobes + + uint32_t history_size = 0; + float solid_cell_ratio = 0; + uint32_t solid_cell_count = 0; + + RS::EnvironmentSDFGICascades cascade_mode; + float min_cell_size = 0; + uint32_t probe_axis_count = 0; //amount of probes per axis, this is an odd number because it encloses endpoints + + RID debug_uniform_set; + RID debug_probes_uniform_set; + RID cascades_ubo; + + bool uses_occlusion = false; + float bounce_feedback = 0.0; + bool reads_sky = false; + float energy = 1.0; + float normal_bias = 1.1; + float probe_bias = 1.1; + RS::EnvironmentSDFGIYScale y_scale_mode = RS::ENV_SDFGI_Y_SCALE_DISABLED; + + float y_mult = 1.0; + + uint32_t render_pass = 0; + + int32_t cascade_dynamic_light_count[SDFGI::MAX_CASCADES]; //used dynamically + RID integrate_sky_uniform_set; + + void create(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size, RendererSceneGIRD *p_gi); + void erase(); + void update(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position); + void update_light(); + void update_probes(RendererSceneEnvironmentRD *p_env, RendererSceneSkyRD::Sky *p_sky); + void store_probes(); + int get_pending_region_data(int p_region, Vector3i &r_local_offset, Vector3i &r_local_size, AABB &r_bounds) const; + void update_cascades(); + + void debug_draw(const CameraMatrix &p_projection, const Transform &p_transform, int p_width, int p_height, RID p_render_target, RID p_texture); + void debug_probes(RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform); + + void pre_process_gi(const Transform &p_transform, RendererSceneRenderRD *p_scene_render); + void render_region(RID p_render_buffers, int p_region, const PagedArray<RendererSceneRender::GeometryInstance *> &p_instances, RendererSceneRenderRD *p_scene_render); + void render_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_light_cull_result, RendererSceneRenderRD *p_scene_render); + }; + + RS::EnvironmentSDFGIRayCount sdfgi_ray_count = RS::ENV_SDFGI_RAY_COUNT_16; + RS::EnvironmentSDFGIFramesToConverge sdfgi_frames_to_converge = RS::ENV_SDFGI_CONVERGE_IN_10_FRAMES; + RS::EnvironmentSDFGIFramesToUpdateLight sdfgi_frames_to_update_light = RS::ENV_SDFGI_UPDATE_LIGHT_IN_4_FRAMES; + + float sdfgi_solid_cell_ratio = 0.25; + Vector3 sdfgi_debug_probe_pos; + Vector3 sdfgi_debug_probe_dir; + bool sdfgi_debug_probe_enabled = false; + Vector3i sdfgi_debug_probe_index; + + /* SDFGI UPDATE */ + + int sdfgi_get_lightprobe_octahedron_size() const { return SDFGI::LIGHTPROBE_OCT_SIZE; } + + /* GI */ + enum { + MAX_GIPROBES = 8 + }; + + // Struct for use in render buffer + struct RenderBuffersGI { + RID giprobe_textures[MAX_GIPROBES]; + RID giprobe_buffer; + + RID full_buffer; + RID full_dispatch; + RID full_mask; + + RID uniform_set; + bool using_half_size_gi = false; + }; + + struct SDFGIData { + float grid_size[3]; + uint32_t max_cascades; + + uint32_t use_occlusion; + int32_t probe_axis_size; + float probe_to_uvw; + float normal_bias; + + float lightprobe_tex_pixel_size[3]; + float energy; + + float lightprobe_uv_offset[3]; + float y_mult; + + float occlusion_clamp[3]; + uint32_t pad3; + + float occlusion_renormalize[3]; + uint32_t pad4; + + float cascade_probe_size[3]; + uint32_t pad5; + + struct ProbeCascadeData { + float position[3]; //offset of (0,0,0) in world coordinates + float to_probe; // 1/bounds * grid_size + int32_t probe_world_offset[3]; + float to_cell; // 1/bounds * grid_size + }; + + ProbeCascadeData cascades[SDFGI::MAX_CASCADES]; + }; + + struct GIProbeData { + float xform[16]; + float bounds[3]; + float dynamic_range; + + float bias; + float normal_bias; + uint32_t blend_ambient; + uint32_t texture_slot; + + float anisotropy_strength; + float ao; + float ao_size; + uint32_t mipmaps; + }; + + struct PushConstant { + int32_t screen_size[2]; + float z_near; + float z_far; + + float proj_info[4]; + float ao_color[3]; + uint32_t max_giprobes; + + uint32_t high_quality_vct; + uint32_t orthogonal; + uint32_t pad[2]; + + float cam_rotation[12]; + }; + + RID sdfgi_ubo; + enum Mode { + MODE_GIPROBE, + MODE_SDFGI, + MODE_COMBINED, + MODE_HALF_RES_GIPROBE, + MODE_HALF_RES_SDFGI, + MODE_HALF_RES_COMBINED, + MODE_MAX + }; + + RID default_giprobe_buffer; + + bool half_resolution = false; + GiShaderRD shader; + RID shader_version; + RID pipelines[MODE_MAX]; + + RendererSceneGIRD(); + ~RendererSceneGIRD(); + + void init(RendererStorageRD *p_storage, RendererSceneSkyRD *p_sky); + void free(); + + SDFGI *create_sdfgi(RendererSceneEnvironmentRD *p_env, const Vector3 &p_world_position, uint32_t p_requested_history_size); + + void setup_giprobes(RID p_render_buffers, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, uint32_t &r_gi_probes_used, RendererSceneRenderRD *p_scene_render); + void process_gi(RID p_render_buffers, RID p_normal_roughness_buffer, RID p_gi_probe_buffer, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, RendererSceneRenderRD *p_scene_render); + + RID gi_probe_instance_create(RID p_base); + void gi_probe_instance_set_transform_to_data(RID p_probe, const Transform &p_xform); + bool gi_probe_needs_update(RID p_probe) const; + void gi_probe_update(RID p_probe, bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<RendererSceneRender::GeometryInstance *> &p_dynamic_objects, RendererSceneRenderRD *p_scene_render); + void debug_giprobe(RID p_gi_probe, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha); +}; + +#endif /* !RENDERING_SERVER_SCENE_GI_RD_H */ diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp index dad08179e7..4c5bded2ff 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp @@ -35,8 +35,6 @@ #include "renderer_compositor_rd.h" #include "servers/rendering/rendering_server_default.h" -uint64_t RendererSceneRenderRD::auto_exposure_counter = 2; - void get_vogel_disk(float *r_kernel, int p_sample_count) { const float golden_angle = 2.4; @@ -49,975 +47,42 @@ void get_vogel_disk(float *r_kernel, int p_sample_count) { } } -void RendererSceneRenderRD::_clear_reflection_data(ReflectionData &rd) { - rd.layers.clear(); - rd.radiance_base_cubemap = RID(); - if (rd.downsampled_radiance_cubemap.is_valid()) { - RD::get_singleton()->free(rd.downsampled_radiance_cubemap); - } - rd.downsampled_radiance_cubemap = RID(); - rd.downsampled_layer.mipmaps.clear(); - rd.coefficient_buffer = RID(); -} - -void RendererSceneRenderRD::_update_reflection_data(ReflectionData &rd, int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality) { - //recreate radiance and all data - - int mipmaps = p_mipmaps; - uint32_t w = p_size, h = p_size; - - if (p_use_array) { - int layers = p_low_quality ? 8 : roughness_layers; - - for (int i = 0; i < layers; i++) { - ReflectionData::Layer layer; - uint32_t mmw = w; - uint32_t mmh = h; - layer.mipmaps.resize(mipmaps); - layer.views.resize(mipmaps); - for (int j = 0; j < mipmaps; j++) { - ReflectionData::Layer::Mipmap &mm = layer.mipmaps.write[j]; - mm.size.width = mmw; - mm.size.height = mmh; - for (int k = 0; k < 6; k++) { - mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + i * 6 + k, j); - Vector<RID> fbtex; - fbtex.push_back(mm.views[k]); - mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); - } - - layer.views.write[j] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + i * 6, j, RD::TEXTURE_SLICE_CUBEMAP); - - mmw = MAX(1, mmw >> 1); - mmh = MAX(1, mmh >> 1); - } - - rd.layers.push_back(layer); - } - - } else { - mipmaps = p_low_quality ? 8 : mipmaps; - //regular cubemap, lower quality (aliasing, less memory) - ReflectionData::Layer layer; - uint32_t mmw = w; - uint32_t mmh = h; - layer.mipmaps.resize(mipmaps); - layer.views.resize(mipmaps); - for (int j = 0; j < mipmaps; j++) { - ReflectionData::Layer::Mipmap &mm = layer.mipmaps.write[j]; - mm.size.width = mmw; - mm.size.height = mmh; - for (int k = 0; k < 6; k++) { - mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + k, j); - Vector<RID> fbtex; - fbtex.push_back(mm.views[k]); - mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); - } - - layer.views.write[j] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer, j, RD::TEXTURE_SLICE_CUBEMAP); - - mmw = MAX(1, mmw >> 1); - mmh = MAX(1, mmh >> 1); - } - - rd.layers.push_back(layer); - } - - rd.radiance_base_cubemap = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer, 0, RD::TEXTURE_SLICE_CUBEMAP); - - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tf.width = 64; // Always 64x64 - tf.height = 64; - tf.texture_type = RD::TEXTURE_TYPE_CUBE; - tf.array_layers = 6; - tf.mipmaps = 7; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - - rd.downsampled_radiance_cubemap = RD::get_singleton()->texture_create(tf, RD::TextureView()); - { - uint32_t mmw = 64; - uint32_t mmh = 64; - rd.downsampled_layer.mipmaps.resize(7); - for (int j = 0; j < rd.downsampled_layer.mipmaps.size(); j++) { - ReflectionData::DownsampleLayer::Mipmap &mm = rd.downsampled_layer.mipmaps.write[j]; - mm.size.width = mmw; - mm.size.height = mmh; - mm.view = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), rd.downsampled_radiance_cubemap, 0, j, RD::TEXTURE_SLICE_CUBEMAP); - - mmw = MAX(1, mmw >> 1); - mmh = MAX(1, mmh >> 1); - } - } -} - -void RendererSceneRenderRD::_create_reflection_fast_filter(ReflectionData &rd, bool p_use_arrays) { - storage->get_effects()->cubemap_downsample(rd.radiance_base_cubemap, rd.downsampled_layer.mipmaps[0].view, rd.downsampled_layer.mipmaps[0].size); - - for (int i = 1; i < rd.downsampled_layer.mipmaps.size(); i++) { - storage->get_effects()->cubemap_downsample(rd.downsampled_layer.mipmaps[i - 1].view, rd.downsampled_layer.mipmaps[i].view, rd.downsampled_layer.mipmaps[i].size); - } - - Vector<RID> views; - if (p_use_arrays) { - for (int i = 1; i < rd.layers.size(); i++) { - views.push_back(rd.layers[i].views[0]); - } - } else { - for (int i = 1; i < rd.layers[0].views.size(); i++) { - views.push_back(rd.layers[0].views[i]); - } - } - - storage->get_effects()->cubemap_filter(rd.downsampled_radiance_cubemap, views, p_use_arrays); -} - -void RendererSceneRenderRD::_create_reflection_importance_sample(ReflectionData &rd, bool p_use_arrays, int p_cube_side, int p_base_layer) { - if (p_use_arrays) { - //render directly to the layers - storage->get_effects()->cubemap_roughness(rd.radiance_base_cubemap, rd.layers[p_base_layer].views[0], p_cube_side, sky_ggx_samples_quality, float(p_base_layer) / (rd.layers.size() - 1.0), rd.layers[p_base_layer].mipmaps[0].size.x); - } else { - storage->get_effects()->cubemap_roughness(rd.layers[0].views[p_base_layer - 1], rd.layers[0].views[p_base_layer], p_cube_side, sky_ggx_samples_quality, float(p_base_layer) / (rd.layers[0].mipmaps.size() - 1.0), rd.layers[0].mipmaps[p_base_layer].size.x); - } -} - -void RendererSceneRenderRD::_update_reflection_mipmaps(ReflectionData &rd, int p_start, int p_end) { - for (int i = p_start; i < p_end; i++) { - for (int j = 0; j < rd.layers[i].mipmaps.size() - 1; j++) { - for (int k = 0; k < 6; k++) { - RID view = rd.layers[i].mipmaps[j].views[k]; - RID texture = rd.layers[i].mipmaps[j + 1].views[k]; - Size2i size = rd.layers[i].mipmaps[j + 1].size; - storage->get_effects()->make_mipmap(view, texture, size); - } - } - } -} - -void RendererSceneRenderRD::_sdfgi_erase(RenderBuffers *rb) { - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - const SDFGI::Cascade &c = rb->sdfgi->cascades[i]; - RD::get_singleton()->free(c.light_data); - RD::get_singleton()->free(c.light_aniso_0_tex); - RD::get_singleton()->free(c.light_aniso_1_tex); - RD::get_singleton()->free(c.sdf_tex); - RD::get_singleton()->free(c.solid_cell_dispatch_buffer); - RD::get_singleton()->free(c.solid_cell_buffer); - RD::get_singleton()->free(c.lightprobe_history_tex); - RD::get_singleton()->free(c.lightprobe_average_tex); - RD::get_singleton()->free(c.lights_buffer); - } - - RD::get_singleton()->free(rb->sdfgi->render_albedo); - RD::get_singleton()->free(rb->sdfgi->render_emission); - RD::get_singleton()->free(rb->sdfgi->render_emission_aniso); - - RD::get_singleton()->free(rb->sdfgi->render_sdf[0]); - RD::get_singleton()->free(rb->sdfgi->render_sdf[1]); - - RD::get_singleton()->free(rb->sdfgi->render_sdf_half[0]); - RD::get_singleton()->free(rb->sdfgi->render_sdf_half[1]); - - for (int i = 0; i < 8; i++) { - RD::get_singleton()->free(rb->sdfgi->render_occlusion[i]); - } - - RD::get_singleton()->free(rb->sdfgi->render_geom_facing); - - RD::get_singleton()->free(rb->sdfgi->lightprobe_data); - RD::get_singleton()->free(rb->sdfgi->lightprobe_history_scroll); - RD::get_singleton()->free(rb->sdfgi->occlusion_data); - RD::get_singleton()->free(rb->sdfgi->ambient_texture); - - RD::get_singleton()->free(rb->sdfgi->cascades_ubo); - - memdelete(rb->sdfgi); - - rb->sdfgi = nullptr; -} - -const Vector3i RendererSceneRenderRD::SDFGI::Cascade::DIRTY_ALL = Vector3i(0x7FFFFFFF, 0x7FFFFFFF, 0x7FFFFFFF); - void RendererSceneRenderRD::sdfgi_update(RID p_render_buffers, RID p_environment, const Vector3 &p_world_position) { - Environment *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); bool needs_sdfgi = env && env->sdfgi_enabled; if (!needs_sdfgi) { if (rb->sdfgi != nullptr) { //erase it - _sdfgi_erase(rb); + rb->sdfgi->erase(); + memdelete(rb->sdfgi); + rb->sdfgi = nullptr; + _render_buffers_uniform_set_changed(p_render_buffers); } return; } static const uint32_t history_frames_to_converge[RS::ENV_SDFGI_CONVERGE_MAX] = { 5, 10, 15, 20, 25, 30 }; - uint32_t requested_history_size = history_frames_to_converge[sdfgi_frames_to_converge]; + uint32_t requested_history_size = history_frames_to_converge[gi.sdfgi_frames_to_converge]; if (rb->sdfgi && (rb->sdfgi->cascade_mode != env->sdfgi_cascades || rb->sdfgi->min_cell_size != env->sdfgi_min_cell_size || requested_history_size != rb->sdfgi->history_size || rb->sdfgi->uses_occlusion != env->sdfgi_use_occlusion || rb->sdfgi->y_scale_mode != env->sdfgi_y_scale)) { //configuration changed, erase - _sdfgi_erase(rb); + rb->sdfgi->erase(); + memdelete(rb->sdfgi); + rb->sdfgi = nullptr; } - SDFGI *sdfgi = rb->sdfgi; + RendererSceneGIRD::SDFGI *sdfgi = rb->sdfgi; if (sdfgi == nullptr) { - //re-create - rb->sdfgi = memnew(SDFGI); - sdfgi = rb->sdfgi; - sdfgi->cascade_mode = env->sdfgi_cascades; - sdfgi->min_cell_size = env->sdfgi_min_cell_size; - sdfgi->uses_occlusion = env->sdfgi_use_occlusion; - sdfgi->y_scale_mode = env->sdfgi_y_scale; - static const float y_scale[3] = { 1.0, 1.5, 2.0 }; - sdfgi->y_mult = y_scale[sdfgi->y_scale_mode]; - static const int cascasde_size[3] = { 4, 6, 8 }; - sdfgi->cascades.resize(cascasde_size[sdfgi->cascade_mode]); - sdfgi->probe_axis_count = SDFGI::PROBE_DIVISOR + 1; - sdfgi->solid_cell_ratio = sdfgi_solid_cell_ratio; - sdfgi->solid_cell_count = uint32_t(float(sdfgi->cascade_size * sdfgi->cascade_size * sdfgi->cascade_size) * sdfgi->solid_cell_ratio); - - float base_cell_size = sdfgi->min_cell_size; - - RD::TextureFormat tf_sdf; - tf_sdf.format = RD::DATA_FORMAT_R8_UNORM; - tf_sdf.width = sdfgi->cascade_size; // Always 64x64 - tf_sdf.height = sdfgi->cascade_size; - tf_sdf.depth = sdfgi->cascade_size; - tf_sdf.texture_type = RD::TEXTURE_TYPE_3D; - tf_sdf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; - - { - RD::TextureFormat tf_render = tf_sdf; - tf_render.format = RD::DATA_FORMAT_R16_UINT; - sdfgi->render_albedo = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - tf_render.format = RD::DATA_FORMAT_R32_UINT; - sdfgi->render_emission = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - sdfgi->render_emission_aniso = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - - tf_render.format = RD::DATA_FORMAT_R8_UNORM; //at least its easy to visualize - - for (int i = 0; i < 8; i++) { - sdfgi->render_occlusion[i] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - } - - tf_render.format = RD::DATA_FORMAT_R32_UINT; - sdfgi->render_geom_facing = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - - tf_render.format = RD::DATA_FORMAT_R8G8B8A8_UINT; - sdfgi->render_sdf[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - sdfgi->render_sdf[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - - tf_render.width /= 2; - tf_render.height /= 2; - tf_render.depth /= 2; - - sdfgi->render_sdf_half[0] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - sdfgi->render_sdf_half[1] = RD::get_singleton()->texture_create(tf_render, RD::TextureView()); - } - - RD::TextureFormat tf_occlusion = tf_sdf; - tf_occlusion.format = RD::DATA_FORMAT_R16_UINT; - tf_occlusion.shareable_formats.push_back(RD::DATA_FORMAT_R16_UINT); - tf_occlusion.shareable_formats.push_back(RD::DATA_FORMAT_R4G4B4A4_UNORM_PACK16); - tf_occlusion.depth *= sdfgi->cascades.size(); //use depth for occlusion slices - tf_occlusion.width *= 2; //use width for the other half - - RD::TextureFormat tf_light = tf_sdf; - tf_light.format = RD::DATA_FORMAT_R32_UINT; - tf_light.shareable_formats.push_back(RD::DATA_FORMAT_R32_UINT); - tf_light.shareable_formats.push_back(RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32); - - RD::TextureFormat tf_aniso0 = tf_sdf; - tf_aniso0.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; - RD::TextureFormat tf_aniso1 = tf_sdf; - tf_aniso1.format = RD::DATA_FORMAT_R8G8_UNORM; - - int passes = nearest_shift(sdfgi->cascade_size) - 1; - - //store lightprobe SH - RD::TextureFormat tf_probes; - tf_probes.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tf_probes.width = sdfgi->probe_axis_count * sdfgi->probe_axis_count; - tf_probes.height = sdfgi->probe_axis_count * SDFGI::SH_SIZE; - tf_probes.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; - tf_probes.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; - - sdfgi->history_size = requested_history_size; - - RD::TextureFormat tf_probe_history = tf_probes; - tf_probe_history.format = RD::DATA_FORMAT_R16G16B16A16_SINT; //signed integer because SH are signed - tf_probe_history.array_layers = sdfgi->history_size; - - RD::TextureFormat tf_probe_average = tf_probes; - tf_probe_average.format = RD::DATA_FORMAT_R32G32B32A32_SINT; //signed integer because SH are signed - tf_probe_average.texture_type = RD::TEXTURE_TYPE_2D; - - sdfgi->lightprobe_history_scroll = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); - sdfgi->lightprobe_average_scroll = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); - - { - //octahedral lightprobes - RD::TextureFormat tf_octprobes = tf_probes; - tf_octprobes.array_layers = sdfgi->cascades.size() * 2; - tf_octprobes.format = RD::DATA_FORMAT_R32_UINT; //pack well with RGBE - tf_octprobes.width = sdfgi->probe_axis_count * sdfgi->probe_axis_count * (SDFGI::LIGHTPROBE_OCT_SIZE + 2); - tf_octprobes.height = sdfgi->probe_axis_count * (SDFGI::LIGHTPROBE_OCT_SIZE + 2); - tf_octprobes.shareable_formats.push_back(RD::DATA_FORMAT_R32_UINT); - tf_octprobes.shareable_formats.push_back(RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32); - //lightprobe texture is an octahedral texture - - sdfgi->lightprobe_data = RD::get_singleton()->texture_create(tf_octprobes, RD::TextureView()); - RD::TextureView tv; - tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; - sdfgi->lightprobe_texture = RD::get_singleton()->texture_create_shared(tv, sdfgi->lightprobe_data); - - //texture handling ambient data, to integrate with volumetric foc - RD::TextureFormat tf_ambient = tf_probes; - tf_ambient.array_layers = sdfgi->cascades.size(); - tf_ambient.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; //pack well with RGBE - tf_ambient.width = sdfgi->probe_axis_count * sdfgi->probe_axis_count; - tf_ambient.height = sdfgi->probe_axis_count; - tf_ambient.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; - //lightprobe texture is an octahedral texture - sdfgi->ambient_texture = RD::get_singleton()->texture_create(tf_ambient, RD::TextureView()); - } - - sdfgi->cascades_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(SDFGI::Cascade::UBO) * SDFGI::MAX_CASCADES); - - sdfgi->occlusion_data = RD::get_singleton()->texture_create(tf_occlusion, RD::TextureView()); - { - RD::TextureView tv; - tv.format_override = RD::DATA_FORMAT_R4G4B4A4_UNORM_PACK16; - sdfgi->occlusion_texture = RD::get_singleton()->texture_create_shared(tv, sdfgi->occlusion_data); - } - - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - SDFGI::Cascade &cascade = sdfgi->cascades[i]; - - /* 3D Textures */ - - cascade.sdf_tex = RD::get_singleton()->texture_create(tf_sdf, RD::TextureView()); - - cascade.light_data = RD::get_singleton()->texture_create(tf_light, RD::TextureView()); - - cascade.light_aniso_0_tex = RD::get_singleton()->texture_create(tf_aniso0, RD::TextureView()); - cascade.light_aniso_1_tex = RD::get_singleton()->texture_create(tf_aniso1, RD::TextureView()); - - { - RD::TextureView tv; - tv.format_override = RD::DATA_FORMAT_E5B9G9R9_UFLOAT_PACK32; - cascade.light_tex = RD::get_singleton()->texture_create_shared(tv, cascade.light_data); - - RD::get_singleton()->texture_clear(cascade.light_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); - RD::get_singleton()->texture_clear(cascade.light_aniso_0_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); - RD::get_singleton()->texture_clear(cascade.light_aniso_1_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); - } - - cascade.cell_size = base_cell_size; - Vector3 world_position = p_world_position; - world_position.y *= sdfgi->y_mult; - int32_t probe_cells = sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - Vector3 probe_size = Vector3(1, 1, 1) * cascade.cell_size * probe_cells; - Vector3i probe_pos = Vector3i((world_position / probe_size + Vector3(0.5, 0.5, 0.5)).floor()); - cascade.position = probe_pos * probe_cells; - - cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; - - base_cell_size *= 2.0; - - /* Probe History */ - - cascade.lightprobe_history_tex = RD::get_singleton()->texture_create(tf_probe_history, RD::TextureView()); - RD::get_singleton()->texture_clear(cascade.lightprobe_history_tex, Color(0, 0, 0, 0), 0, 1, 0, tf_probe_history.array_layers); //needs to be cleared for average to work - - cascade.lightprobe_average_tex = RD::get_singleton()->texture_create(tf_probe_average, RD::TextureView()); - RD::get_singleton()->texture_clear(cascade.lightprobe_average_tex, Color(0, 0, 0, 0), 0, 1, 0, 1); //needs to be cleared for average to work - - /* Buffers */ - - cascade.solid_cell_buffer = RD::get_singleton()->storage_buffer_create(sizeof(SDFGI::Cascade::SolidCell) * sdfgi->solid_cell_count); - cascade.solid_cell_dispatch_buffer = RD::get_singleton()->storage_buffer_create(sizeof(uint32_t) * 4, Vector<uint8_t>(), RD::STORAGE_BUFFER_USAGE_DISPATCH_INDIRECT); - cascade.lights_buffer = RD::get_singleton()->storage_buffer_create(sizeof(SDGIShader::Light) * MAX(SDFGI::MAX_STATIC_LIGHTS, SDFGI::MAX_DYNAMIC_LIGHTS)); - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_sdf[(passes & 1) ? 1 : 0]); //if passes are even, we read from buffer 0, else we read from buffer 1 - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 3; - for (int j = 0; j < 8; j++) { - u.ids.push_back(sdfgi->render_occlusion[j]); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 4; - u.ids.push_back(sdfgi->render_emission); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 5; - u.ids.push_back(sdfgi->render_emission_aniso); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 6; - u.ids.push_back(sdfgi->render_geom_facing); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 7; - u.ids.push_back(cascade.sdf_tex); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 8; - u.ids.push_back(sdfgi->occlusion_data); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 10; - u.ids.push_back(cascade.solid_cell_dispatch_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 11; - u.ids.push_back(cascade.solid_cell_buffer); - uniforms.push_back(u); - } - - cascade.sdf_store_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_STORE), 0); - } - - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_geom_facing); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 3; - u.ids.push_back(sdfgi->render_emission); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 4; - u.ids.push_back(sdfgi->render_emission_aniso); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 5; - u.ids.push_back(cascade.solid_cell_dispatch_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 6; - u.ids.push_back(cascade.solid_cell_buffer); - uniforms.push_back(u); - } - - cascade.scroll_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_SCROLL), 0); - } - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - for (int j = 0; j < 8; j++) { - u.ids.push_back(sdfgi->render_occlusion[j]); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->occlusion_data); - uniforms.push_back(u); - } - - cascade.scroll_occlusion_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_SCROLL_OCCLUSION), 0); - } - } - - //direct light - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - SDFGI::Cascade &cascade = sdfgi->cascades[i]; - - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.binding = 1; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (j < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[j].sdf_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(cascade.solid_cell_dispatch_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 4; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(cascade.solid_cell_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 5; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.ids.push_back(cascade.light_data); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 6; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.ids.push_back(cascade.light_aniso_0_tex); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 7; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.ids.push_back(cascade.light_aniso_1_tex); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 8; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(rb->sdfgi->cascades_ubo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 9; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.ids.push_back(cascade.lights_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 10; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(rb->sdfgi->lightprobe_texture); - uniforms.push_back(u); - } - - cascade.sdf_direct_light_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.direct_light.version_get_shader(sdfgi_shader.direct_light_shader, 0), 0); - } - - //preprocess initialize uniform set - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_sdf[0]); - uniforms.push_back(u); - } - - sdfgi->sdf_initialize_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE), 0); - } - - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_sdf_half[0]); - uniforms.push_back(u); - } - - sdfgi->sdf_initialize_half_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF), 0); - } - - //jump flood uniform set - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_sdf[0]); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_sdf[1]); - uniforms.push_back(u); - } - - sdfgi->jump_flood_uniform_set[0] = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD), 0); - SWAP(uniforms.write[0].ids.write[0], uniforms.write[1].ids.write[0]); - sdfgi->jump_flood_uniform_set[1] = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD), 0); - } - //jump flood half uniform set - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_sdf_half[0]); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_sdf_half[1]); - uniforms.push_back(u); - } - - sdfgi->jump_flood_half_uniform_set[0] = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD), 0); - SWAP(uniforms.write[0].ids.write[0], uniforms.write[1].ids.write[0]); - sdfgi->jump_flood_half_uniform_set[1] = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD), 0); - } - - //upscale half size sdf - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - u.ids.push_back(sdfgi->render_sdf_half[(passes & 1) ? 0 : 1]); //reverse pass order because half size - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 3; - u.ids.push_back(sdfgi->render_sdf[(passes & 1) ? 0 : 1]); //reverse pass order because it needs an extra JFA pass - uniforms.push_back(u); - } - - sdfgi->upscale_jfa_uniform_set_index = (passes & 1) ? 0 : 1; - sdfgi->sdf_upscale_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_JUMP_FLOOD_UPSCALE), 0); - } - - //occlusion uniform set - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 1; - u.ids.push_back(sdfgi->render_albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 2; - for (int i = 0; i < 8; i++) { - u.ids.push_back(sdfgi->render_occlusion[i]); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 3; - u.ids.push_back(sdfgi->render_geom_facing); - uniforms.push_back(u); - } - - sdfgi->occlusion_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, SDGIShader::PRE_PROCESS_OCCLUSION), 0); - } - - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - //integrate uniform - - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.binding = 1; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (j < sdfgi->cascades.size()) { - u.ids.push_back(sdfgi->cascades[j].sdf_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (j < sdfgi->cascades.size()) { - u.ids.push_back(sdfgi->cascades[j].light_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (j < sdfgi->cascades.size()) { - u.ids.push_back(sdfgi->cascades[j].light_aniso_0_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 4; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (j < sdfgi->cascades.size()) { - u.ids.push_back(sdfgi->cascades[j].light_aniso_1_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 6; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 7; - u.ids.push_back(sdfgi->cascades_ubo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 8; - u.ids.push_back(sdfgi->lightprobe_data); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 9; - u.ids.push_back(sdfgi->cascades[i].lightprobe_history_tex); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 10; - u.ids.push_back(sdfgi->cascades[i].lightprobe_average_tex); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 11; - u.ids.push_back(sdfgi->lightprobe_history_scroll); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 12; - u.ids.push_back(sdfgi->lightprobe_average_scroll); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 13; - RID parent_average; - if (i < sdfgi->cascades.size() - 1) { - parent_average = sdfgi->cascades[i + 1].lightprobe_average_tex; - } else { - parent_average = sdfgi->cascades[i - 1].lightprobe_average_tex; //to use something, but it won't be used - } - u.ids.push_back(parent_average); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 14; - u.ids.push_back(sdfgi->ambient_texture); - uniforms.push_back(u); - } - - sdfgi->cascades[i].integrate_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, 0), 0); - } - - sdfgi->uses_multibounce = env->sdfgi_use_multibounce; - sdfgi->energy = env->sdfgi_energy; - sdfgi->normal_bias = env->sdfgi_normal_bias; - sdfgi->probe_bias = env->sdfgi_probe_bias; - sdfgi->reads_sky = env->sdfgi_read_sky_light; + // re-create + rb->sdfgi = gi.create_sdfgi(env, p_world_position, requested_history_size); _render_buffers_uniform_set_changed(p_render_buffers); - - return; //done. all levels will need to be rendered which its going to take a bit - } - - //check for updates - - sdfgi->uses_multibounce = env->sdfgi_use_multibounce; - sdfgi->energy = env->sdfgi_energy; - sdfgi->normal_bias = env->sdfgi_normal_bias; - sdfgi->probe_bias = env->sdfgi_probe_bias; - sdfgi->reads_sky = env->sdfgi_read_sky_light; - - int32_t drag_margin = (sdfgi->cascade_size / SDFGI::PROBE_DIVISOR) / 2; - - for (uint32_t i = 0; i < sdfgi->cascades.size(); i++) { - SDFGI::Cascade &cascade = sdfgi->cascades[i]; - cascade.dirty_regions = Vector3i(); - - Vector3 probe_half_size = Vector3(1, 1, 1) * cascade.cell_size * float(sdfgi->cascade_size / SDFGI::PROBE_DIVISOR) * 0.5; - probe_half_size = Vector3(0, 0, 0); - - Vector3 world_position = p_world_position; - world_position.y *= sdfgi->y_mult; - Vector3i pos_in_cascade = Vector3i((world_position + probe_half_size) / cascade.cell_size); - - for (int j = 0; j < 3; j++) { - if (pos_in_cascade[j] < cascade.position[j]) { - while (pos_in_cascade[j] < (cascade.position[j] - drag_margin)) { - cascade.position[j] -= drag_margin * 2; - cascade.dirty_regions[j] += drag_margin * 2; - } - } else if (pos_in_cascade[j] > cascade.position[j]) { - while (pos_in_cascade[j] > (cascade.position[j] + drag_margin)) { - cascade.position[j] += drag_margin * 2; - cascade.dirty_regions[j] -= drag_margin * 2; - } - } - - if (cascade.dirty_regions[j] == 0) { - continue; // not dirty - } else if (uint32_t(ABS(cascade.dirty_regions[j])) >= sdfgi->cascade_size) { - //moved too much, just redraw everything (make all dirty) - cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; - break; - } - } - - if (cascade.dirty_regions != Vector3i() && cascade.dirty_regions != SDFGI::Cascade::DIRTY_ALL) { - //see how much the total dirty volume represents from the total volume - uint32_t total_volume = sdfgi->cascade_size * sdfgi->cascade_size * sdfgi->cascade_size; - uint32_t safe_volume = 1; - for (int j = 0; j < 3; j++) { - safe_volume *= sdfgi->cascade_size - ABS(cascade.dirty_regions[j]); - } - uint32_t dirty_volume = total_volume - safe_volume; - if (dirty_volume > (safe_volume / 2)) { - //more than half the volume is dirty, make all dirty so its only rendered once - cascade.dirty_regions = SDFGI::Cascade::DIRTY_ALL; - } - } + } else { + //check for updates + rb->sdfgi->update(env, p_world_position); } } @@ -1032,9 +97,9 @@ int RendererSceneRenderRD::sdfgi_get_pending_region_count(RID p_render_buffers) int dirty_count = 0; for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - const SDFGI::Cascade &c = rb->sdfgi->cascades[i]; + const RendererSceneGIRD::SDFGI::Cascade &c = rb->sdfgi->cascades[i]; - if (c.dirty_regions == SDFGI::Cascade::DIRTY_ALL) { + if (c.dirty_regions == RendererSceneGIRD::SDFGI::Cascade::DIRTY_ALL) { dirty_count++; } else { for (int j = 0; j < 3; j++) { @@ -1048,72 +113,15 @@ int RendererSceneRenderRD::sdfgi_get_pending_region_count(RID p_render_buffers) return dirty_count; } -int RendererSceneRenderRD::_sdfgi_get_pending_region_data(RID p_render_buffers, int p_region, Vector3i &r_local_offset, Vector3i &r_local_size, AABB &r_bounds) const { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND_V(rb == nullptr, -1); - ERR_FAIL_COND_V(rb->sdfgi == nullptr, -1); - - int dirty_count = 0; - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - const SDFGI::Cascade &c = rb->sdfgi->cascades[i]; - - if (c.dirty_regions == SDFGI::Cascade::DIRTY_ALL) { - if (dirty_count == p_region) { - r_local_offset = Vector3i(); - r_local_size = Vector3i(1, 1, 1) * rb->sdfgi->cascade_size; - - r_bounds.position = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + c.position)) * c.cell_size * Vector3(1, 1.0 / rb->sdfgi->y_mult, 1); - r_bounds.size = Vector3(r_local_size) * c.cell_size * Vector3(1, 1.0 / rb->sdfgi->y_mult, 1); - return i; - } - dirty_count++; - } else { - for (int j = 0; j < 3; j++) { - if (c.dirty_regions[j] != 0) { - if (dirty_count == p_region) { - Vector3i from = Vector3i(0, 0, 0); - Vector3i to = Vector3i(1, 1, 1) * rb->sdfgi->cascade_size; - - if (c.dirty_regions[j] > 0) { - //fill from the beginning - to[j] = c.dirty_regions[j]; - } else { - //fill from the end - from[j] = to[j] + c.dirty_regions[j]; - } - - for (int k = 0; k < j; k++) { - // "chip" away previous regions to avoid re-voxelizing the same thing - if (c.dirty_regions[k] > 0) { - from[k] += c.dirty_regions[k]; - } else if (c.dirty_regions[k] < 0) { - to[k] += c.dirty_regions[k]; - } - } - - r_local_offset = from; - r_local_size = to - from; - - r_bounds.position = Vector3(from + Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + c.position) * c.cell_size * Vector3(1, 1.0 / rb->sdfgi->y_mult, 1); - r_bounds.size = Vector3(r_local_size) * c.cell_size * Vector3(1, 1.0 / rb->sdfgi->y_mult, 1); - - return i; - } - - dirty_count++; - } - } - } - } - return -1; -} - AABB RendererSceneRenderRD::sdfgi_get_pending_region_bounds(RID p_render_buffers, int p_region) const { AABB bounds; Vector3i from; Vector3i size; + RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND_V(rb == nullptr, AABB()); + ERR_FAIL_COND_V(rb->sdfgi == nullptr, AABB()); - int c = _sdfgi_get_pending_region_data(p_render_buffers, p_region, from, size, bounds); + int c = rb->sdfgi->get_pending_region_data(p_region, from, size, bounds); ERR_FAIL_COND_V(c == -1, AABB()); return bounds; } @@ -1122,1841 +130,179 @@ uint32_t RendererSceneRenderRD::sdfgi_get_pending_region_cascade(RID p_render_bu AABB bounds; Vector3i from; Vector3i size; - - return _sdfgi_get_pending_region_data(p_render_buffers, p_region, from, size, bounds); -} - -void RendererSceneRenderRD::_sdfgi_update_cascades(RID p_render_buffers) { RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(rb == nullptr); - if (rb->sdfgi == nullptr) { - return; - } - - //update cascades - SDFGI::Cascade::UBO cascade_data[SDFGI::MAX_CASCADES]; - int32_t probe_divisor = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - Vector3 pos = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + rb->sdfgi->cascades[i].position)) * rb->sdfgi->cascades[i].cell_size; - - cascade_data[i].offset[0] = pos.x; - cascade_data[i].offset[1] = pos.y; - cascade_data[i].offset[2] = pos.z; - cascade_data[i].to_cell = 1.0 / rb->sdfgi->cascades[i].cell_size; - cascade_data[i].probe_offset[0] = rb->sdfgi->cascades[i].position.x / probe_divisor; - cascade_data[i].probe_offset[1] = rb->sdfgi->cascades[i].position.y / probe_divisor; - cascade_data[i].probe_offset[2] = rb->sdfgi->cascades[i].position.z / probe_divisor; - cascade_data[i].pad = 0; - } - - RD::get_singleton()->buffer_update(rb->sdfgi->cascades_ubo, 0, sizeof(SDFGI::Cascade::UBO) * SDFGI::MAX_CASCADES, cascade_data, true); -} - -void RendererSceneRenderRD::sdfgi_update_probes(RID p_render_buffers, RID p_environment, const Vector<RID> &p_directional_lights, const RID *p_positional_light_instances, uint32_t p_positional_light_count) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(rb == nullptr); - if (rb->sdfgi == nullptr) { - return; - } - Environment *env = environment_owner.getornull(p_environment); - - RENDER_TIMESTAMP(">SDFGI Update Probes"); - - /* Update Cascades UBO */ - _sdfgi_update_cascades(p_render_buffers); - /* Update Dynamic Lights Buffer */ - - RENDER_TIMESTAMP("Update Lights"); - - /* Update dynamic lights */ - - { - int32_t cascade_light_count[SDFGI::MAX_CASCADES]; - - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - SDFGI::Cascade &cascade = rb->sdfgi->cascades[i]; - - SDGIShader::Light lights[SDFGI::MAX_DYNAMIC_LIGHTS]; - uint32_t idx = 0; - for (uint32_t j = 0; j < (uint32_t)p_directional_lights.size(); j++) { - if (idx == SDFGI::MAX_DYNAMIC_LIGHTS) { - break; - } - - LightInstance *li = light_instance_owner.getornull(p_directional_lights[j]); - ERR_CONTINUE(!li); - - if (storage->light_directional_is_sky_only(li->light)) { - continue; - } - - Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); - dir.y *= rb->sdfgi->y_mult; - dir.normalize(); - lights[idx].direction[0] = dir.x; - lights[idx].direction[1] = dir.y; - lights[idx].direction[2] = dir.z; - Color color = storage->light_get_color(li->light); - color = color.to_linear(); - lights[idx].color[0] = color.r; - lights[idx].color[1] = color.g; - lights[idx].color[2] = color.b; - lights[idx].type = RS::LIGHT_DIRECTIONAL; - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); - lights[idx].has_shadow = storage->light_has_shadow(li->light); - - idx++; - } - - AABB cascade_aabb; - cascade_aabb.position = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + cascade.position)) * cascade.cell_size; - cascade_aabb.size = Vector3(1, 1, 1) * rb->sdfgi->cascade_size * cascade.cell_size; - - for (uint32_t j = 0; j < p_positional_light_count; j++) { - if (idx == SDFGI::MAX_DYNAMIC_LIGHTS) { - break; - } - - LightInstance *li = light_instance_owner.getornull(p_positional_light_instances[j]); - ERR_CONTINUE(!li); - - uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); - if (i > max_sdfgi_cascade) { - continue; - } - - if (!cascade_aabb.intersects(li->aabb)) { - continue; - } - - Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); - //faster to not do this here - //dir.y *= rb->sdfgi->y_mult; - //dir.normalize(); - lights[idx].direction[0] = dir.x; - lights[idx].direction[1] = dir.y; - lights[idx].direction[2] = dir.z; - Vector3 pos = li->transform.origin; - pos.y *= rb->sdfgi->y_mult; - lights[idx].position[0] = pos.x; - lights[idx].position[1] = pos.y; - lights[idx].position[2] = pos.z; - Color color = storage->light_get_color(li->light); - color = color.to_linear(); - lights[idx].color[0] = color.r; - lights[idx].color[1] = color.g; - lights[idx].color[2] = color.b; - lights[idx].type = storage->light_get_type(li->light); - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); - lights[idx].has_shadow = storage->light_has_shadow(li->light); - lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); - lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); - lights[idx].spot_angle = Math::deg2rad(storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ANGLE)); - lights[idx].spot_attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ATTENUATION); - - idx++; - } - - if (idx > 0) { - RD::get_singleton()->buffer_update(cascade.lights_buffer, 0, idx * sizeof(SDGIShader::Light), lights, true); - } - - cascade_light_count[i] = idx; - } - - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.direct_light_pipeline[SDGIShader::DIRECT_LIGHT_MODE_DYNAMIC]); - - SDGIShader::DirectLightPushConstant push_constant; - - push_constant.grid_size[0] = rb->sdfgi->cascade_size; - push_constant.grid_size[1] = rb->sdfgi->cascade_size; - push_constant.grid_size[2] = rb->sdfgi->cascade_size; - push_constant.max_cascades = rb->sdfgi->cascades.size(); - push_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - push_constant.multibounce = rb->sdfgi->uses_multibounce; - push_constant.y_mult = rb->sdfgi->y_mult; - - push_constant.process_offset = 0; - push_constant.process_increment = 1; - - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - SDFGI::Cascade &cascade = rb->sdfgi->cascades[i]; - push_constant.light_count = cascade_light_count[i]; - push_constant.cascade = i; - - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cascade.sdf_direct_light_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::DirectLightPushConstant)); - RD::get_singleton()->compute_list_dispatch_indirect(compute_list, cascade.solid_cell_dispatch_buffer, 0); - } - RD::get_singleton()->compute_list_end(); - } - - RENDER_TIMESTAMP("Raytrace"); - - SDGIShader::IntegratePushConstant push_constant; - push_constant.grid_size[1] = rb->sdfgi->cascade_size; - push_constant.grid_size[2] = rb->sdfgi->cascade_size; - push_constant.grid_size[0] = rb->sdfgi->cascade_size; - push_constant.max_cascades = rb->sdfgi->cascades.size(); - push_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - push_constant.history_index = rb->sdfgi->render_pass % rb->sdfgi->history_size; - push_constant.history_size = rb->sdfgi->history_size; - static const uint32_t ray_count[RS::ENV_SDFGI_RAY_COUNT_MAX] = { 8, 16, 32, 64, 96, 128 }; - push_constant.ray_count = ray_count[sdfgi_ray_count]; - push_constant.ray_bias = rb->sdfgi->probe_bias; - push_constant.image_size[0] = rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count; - push_constant.image_size[1] = rb->sdfgi->probe_axis_count; - push_constant.store_ambient_texture = env->volumetric_fog_enabled; - - RID sky_uniform_set = sdfgi_shader.integrate_default_sky_uniform_set; - push_constant.sky_mode = SDGIShader::IntegratePushConstant::SKY_MODE_DISABLED; - push_constant.y_mult = rb->sdfgi->y_mult; - - if (rb->sdfgi->reads_sky && env) { - push_constant.sky_energy = env->bg_energy; - - if (env->background == RS::ENV_BG_CLEAR_COLOR) { - push_constant.sky_mode = SDGIShader::IntegratePushConstant::SKY_MODE_COLOR; - Color c = storage->get_default_clear_color().to_linear(); - push_constant.sky_color[0] = c.r; - push_constant.sky_color[1] = c.g; - push_constant.sky_color[2] = c.b; - } else if (env->background == RS::ENV_BG_COLOR) { - push_constant.sky_mode = SDGIShader::IntegratePushConstant::SKY_MODE_COLOR; - Color c = env->bg_color; - push_constant.sky_color[0] = c.r; - push_constant.sky_color[1] = c.g; - push_constant.sky_color[2] = c.b; - - } else if (env->background == RS::ENV_BG_SKY) { - Sky *sky = sky_owner.getornull(env->sky); - if (sky && sky->radiance.is_valid()) { - if (sky->sdfgi_integrate_sky_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(sky->sdfgi_integrate_sky_uniform_set)) { - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 0; - u.ids.push_back(sky->radiance); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 1; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - sky->sdfgi_integrate_sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, 0), 1); - } - sky_uniform_set = sky->sdfgi_integrate_sky_uniform_set; - push_constant.sky_mode = SDGIShader::IntegratePushConstant::SKY_MODE_SKY; - } - } - } - - rb->sdfgi->render_pass++; - - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.integrate_pipeline[SDGIShader::INTEGRATE_MODE_PROCESS]); - - int32_t probe_divisor = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - push_constant.cascade = i; - push_constant.world_offset[0] = rb->sdfgi->cascades[i].position.x / probe_divisor; - push_constant.world_offset[1] = rb->sdfgi->cascades[i].position.y / probe_divisor; - push_constant.world_offset[2] = rb->sdfgi->cascades[i].position.z / probe_divisor; - - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[i].integrate_uniform_set, 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sky_uniform_set, 1); - - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::IntegratePushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count, rb->sdfgi->probe_axis_count, 1, 8, 8, 1); - } - - RD::get_singleton()->compute_list_add_barrier(compute_list); //wait until done - - // Then store values into the lightprobe texture. Separating these steps has a small performance hit, but it allows for multiple bounces - RENDER_TIMESTAMP("Average Probes"); - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.integrate_pipeline[SDGIShader::INTEGRATE_MODE_STORE]); - - //convert to octahedral to store - push_constant.image_size[0] *= SDFGI::LIGHTPROBE_OCT_SIZE; - push_constant.image_size[1] *= SDFGI::LIGHTPROBE_OCT_SIZE; - - for (uint32_t i = 0; i < rb->sdfgi->cascades.size(); i++) { - push_constant.cascade = i; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[i].integrate_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::IntegratePushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, rb->sdfgi->probe_axis_count * SDFGI::LIGHTPROBE_OCT_SIZE, 1, 8, 8, 1); - } - - RD::get_singleton()->compute_list_end(); - - RENDER_TIMESTAMP("<SDFGI Update Probes"); -} - -void RendererSceneRenderRD::_setup_giprobes(RID p_render_buffers, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, uint32_t &r_gi_probes_used) { - r_gi_probes_used = 0; - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(rb == nullptr); - - RID gi_probe_buffer = render_buffers_get_gi_probe_buffer(p_render_buffers); - GI::GIProbeData gi_probe_data[RenderBuffers::MAX_GIPROBES]; - - bool giprobes_changed = false; - - Transform to_camera; - to_camera.origin = p_transform.origin; //only translation, make local - - for (int i = 0; i < RenderBuffers::MAX_GIPROBES; i++) { - RID texture; - if (i < (int)p_gi_probes.size()) { - GIProbeInstance *gipi = gi_probe_instance_owner.getornull(p_gi_probes[i]); - - if (gipi) { - texture = gipi->texture; - GI::GIProbeData &gipd = gi_probe_data[i]; - - RID base_probe = gipi->probe; - - Transform to_cell = storage->gi_probe_get_to_cell_xform(gipi->probe) * gipi->transform.affine_inverse() * to_camera; - - gipd.xform[0] = to_cell.basis.elements[0][0]; - gipd.xform[1] = to_cell.basis.elements[1][0]; - gipd.xform[2] = to_cell.basis.elements[2][0]; - gipd.xform[3] = 0; - gipd.xform[4] = to_cell.basis.elements[0][1]; - gipd.xform[5] = to_cell.basis.elements[1][1]; - gipd.xform[6] = to_cell.basis.elements[2][1]; - gipd.xform[7] = 0; - gipd.xform[8] = to_cell.basis.elements[0][2]; - gipd.xform[9] = to_cell.basis.elements[1][2]; - gipd.xform[10] = to_cell.basis.elements[2][2]; - gipd.xform[11] = 0; - gipd.xform[12] = to_cell.origin.x; - gipd.xform[13] = to_cell.origin.y; - gipd.xform[14] = to_cell.origin.z; - gipd.xform[15] = 1; - - Vector3 bounds = storage->gi_probe_get_octree_size(base_probe); - - gipd.bounds[0] = bounds.x; - gipd.bounds[1] = bounds.y; - gipd.bounds[2] = bounds.z; - - gipd.dynamic_range = storage->gi_probe_get_dynamic_range(base_probe) * storage->gi_probe_get_energy(base_probe); - gipd.bias = storage->gi_probe_get_bias(base_probe); - gipd.normal_bias = storage->gi_probe_get_normal_bias(base_probe); - gipd.blend_ambient = !storage->gi_probe_is_interior(base_probe); - gipd.anisotropy_strength = 0; - gipd.ao = storage->gi_probe_get_ao(base_probe); - gipd.ao_size = Math::pow(storage->gi_probe_get_ao_size(base_probe), 4.0f); - gipd.mipmaps = gipi->mipmaps.size(); - } - - r_gi_probes_used++; - } - - if (texture == RID()) { - texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); - } - - if (texture != rb->giprobe_textures[i]) { - giprobes_changed = true; - rb->giprobe_textures[i] = texture; - } - } - - if (giprobes_changed) { - if (RD::get_singleton()->uniform_set_is_valid(rb->gi_uniform_set)) { - RD::get_singleton()->free(rb->gi_uniform_set); - } - rb->gi_uniform_set = RID(); - if (rb->volumetric_fog) { - if (RD::get_singleton()->uniform_set_is_valid(rb->volumetric_fog->uniform_set)) { - RD::get_singleton()->free(rb->volumetric_fog->uniform_set); - RD::get_singleton()->free(rb->volumetric_fog->uniform_set2); - } - rb->volumetric_fog->uniform_set = RID(); - rb->volumetric_fog->uniform_set2 = RID(); - } - } - - if (p_gi_probes.size() > 0) { - RD::get_singleton()->buffer_update(gi_probe_buffer, 0, sizeof(GI::GIProbeData) * MIN((uint64_t)RenderBuffers::MAX_GIPROBES, p_gi_probes.size()), gi_probe_data, true); - } -} - -void RendererSceneRenderRD::_process_gi(RID p_render_buffers, RID p_normal_roughness_buffer, RID p_ambient_buffer, RID p_reflection_buffer, RID p_gi_probe_buffer, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform, const PagedArray<RID> &p_gi_probes) { - RENDER_TIMESTAMP("Render GI"); - - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(rb == nullptr); - Environment *env = environment_owner.getornull(p_environment); - - GI::PushConstant push_constant; - - push_constant.screen_size[0] = rb->width; - push_constant.screen_size[1] = rb->height; - push_constant.z_near = p_projection.get_z_near(); - push_constant.z_far = p_projection.get_z_far(); - push_constant.orthogonal = p_projection.is_orthogonal(); - push_constant.proj_info[0] = -2.0f / (rb->width * p_projection.matrix[0][0]); - push_constant.proj_info[1] = -2.0f / (rb->height * p_projection.matrix[1][1]); - push_constant.proj_info[2] = (1.0f - p_projection.matrix[0][2]) / p_projection.matrix[0][0]; - push_constant.proj_info[3] = (1.0f + p_projection.matrix[1][2]) / p_projection.matrix[1][1]; - push_constant.max_giprobes = MIN((uint64_t)RenderBuffers::MAX_GIPROBES, p_gi_probes.size()); - push_constant.high_quality_vct = gi_probe_quality == RS::GI_PROBE_QUALITY_HIGH; - push_constant.use_sdfgi = rb->sdfgi != nullptr; - - if (env) { - push_constant.ao_color[0] = env->ao_color.r; - push_constant.ao_color[1] = env->ao_color.g; - push_constant.ao_color[2] = env->ao_color.b; - } else { - push_constant.ao_color[0] = 0; - push_constant.ao_color[1] = 0; - push_constant.ao_color[2] = 0; - } - - push_constant.cam_rotation[0] = p_transform.basis[0][0]; - push_constant.cam_rotation[1] = p_transform.basis[1][0]; - push_constant.cam_rotation[2] = p_transform.basis[2][0]; - push_constant.cam_rotation[3] = 0; - push_constant.cam_rotation[4] = p_transform.basis[0][1]; - push_constant.cam_rotation[5] = p_transform.basis[1][1]; - push_constant.cam_rotation[6] = p_transform.basis[2][1]; - push_constant.cam_rotation[7] = 0; - push_constant.cam_rotation[8] = p_transform.basis[0][2]; - push_constant.cam_rotation[9] = p_transform.basis[1][2]; - push_constant.cam_rotation[10] = p_transform.basis[2][2]; - push_constant.cam_rotation[11] = 0; - - if (rb->sdfgi) { - GI::SDFGIData sdfgi_data; - - sdfgi_data.grid_size[0] = rb->sdfgi->cascade_size; - sdfgi_data.grid_size[1] = rb->sdfgi->cascade_size; - sdfgi_data.grid_size[2] = rb->sdfgi->cascade_size; - - sdfgi_data.max_cascades = rb->sdfgi->cascades.size(); - sdfgi_data.probe_axis_size = rb->sdfgi->probe_axis_count; - sdfgi_data.cascade_probe_size[0] = sdfgi_data.probe_axis_size - 1; //float version for performance - sdfgi_data.cascade_probe_size[1] = sdfgi_data.probe_axis_size - 1; - sdfgi_data.cascade_probe_size[2] = sdfgi_data.probe_axis_size - 1; - - float csize = rb->sdfgi->cascade_size; - sdfgi_data.probe_to_uvw = 1.0 / float(sdfgi_data.cascade_probe_size[0]); - sdfgi_data.use_occlusion = rb->sdfgi->uses_occlusion; - //sdfgi_data.energy = rb->sdfgi->energy; - - sdfgi_data.y_mult = rb->sdfgi->y_mult; - - float cascade_voxel_size = (csize / sdfgi_data.cascade_probe_size[0]); - float occlusion_clamp = (cascade_voxel_size - 0.5) / cascade_voxel_size; - sdfgi_data.occlusion_clamp[0] = occlusion_clamp; - sdfgi_data.occlusion_clamp[1] = occlusion_clamp; - sdfgi_data.occlusion_clamp[2] = occlusion_clamp; - sdfgi_data.normal_bias = (rb->sdfgi->normal_bias / csize) * sdfgi_data.cascade_probe_size[0]; - - //vec2 tex_pixel_size = 1.0 / vec2(ivec2( (OCT_SIZE+2) * params.probe_axis_size * params.probe_axis_size, (OCT_SIZE+2) * params.probe_axis_size ) ); - //vec3 probe_uv_offset = (ivec3(OCT_SIZE+2,OCT_SIZE+2,(OCT_SIZE+2) * params.probe_axis_size)) * tex_pixel_size.xyx; - - uint32_t oct_size = SDFGI::LIGHTPROBE_OCT_SIZE; - - sdfgi_data.lightprobe_tex_pixel_size[0] = 1.0 / ((oct_size + 2) * sdfgi_data.probe_axis_size * sdfgi_data.probe_axis_size); - sdfgi_data.lightprobe_tex_pixel_size[1] = 1.0 / ((oct_size + 2) * sdfgi_data.probe_axis_size); - sdfgi_data.lightprobe_tex_pixel_size[2] = 1.0; - - sdfgi_data.energy = rb->sdfgi->energy; - - sdfgi_data.lightprobe_uv_offset[0] = float(oct_size + 2) * sdfgi_data.lightprobe_tex_pixel_size[0]; - sdfgi_data.lightprobe_uv_offset[1] = float(oct_size + 2) * sdfgi_data.lightprobe_tex_pixel_size[1]; - sdfgi_data.lightprobe_uv_offset[2] = float((oct_size + 2) * sdfgi_data.probe_axis_size) * sdfgi_data.lightprobe_tex_pixel_size[0]; - - sdfgi_data.occlusion_renormalize[0] = 0.5; - sdfgi_data.occlusion_renormalize[1] = 1.0; - sdfgi_data.occlusion_renormalize[2] = 1.0 / float(sdfgi_data.max_cascades); - - int32_t probe_divisor = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - - for (uint32_t i = 0; i < sdfgi_data.max_cascades; i++) { - GI::SDFGIData::ProbeCascadeData &c = sdfgi_data.cascades[i]; - Vector3 pos = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + rb->sdfgi->cascades[i].position)) * rb->sdfgi->cascades[i].cell_size; - Vector3 cam_origin = p_transform.origin; - cam_origin.y *= rb->sdfgi->y_mult; - pos -= cam_origin; //make pos local to camera, to reduce numerical error - c.position[0] = pos.x; - c.position[1] = pos.y; - c.position[2] = pos.z; - c.to_probe = 1.0 / (float(rb->sdfgi->cascade_size) * rb->sdfgi->cascades[i].cell_size / float(rb->sdfgi->probe_axis_count - 1)); - - Vector3i probe_ofs = rb->sdfgi->cascades[i].position / probe_divisor; - c.probe_world_offset[0] = probe_ofs.x; - c.probe_world_offset[1] = probe_ofs.y; - c.probe_world_offset[2] = probe_ofs.z; - - c.to_cell = 1.0 / rb->sdfgi->cascades[i].cell_size; - } - - RD::get_singleton()->buffer_update(gi.sdfgi_ubo, 0, sizeof(GI::SDFGIData), &sdfgi_data, true); - } - - if (rb->gi_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(rb->gi_uniform_set)) { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.binding = 1; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[j].sdf_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[j].light_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[j].light_aniso_0_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 4; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t j = 0; j < SDFGI::MAX_CASCADES; j++) { - if (rb->sdfgi && j < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[j].light_aniso_1_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 5; - if (rb->sdfgi) { - u.ids.push_back(rb->sdfgi->occlusion_texture); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 6; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 7; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 9; - u.ids.push_back(p_ambient_buffer); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 10; - u.ids.push_back(p_reflection_buffer); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 11; - if (rb->sdfgi) { - u.ids.push_back(rb->sdfgi->lightprobe_texture); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_2D_ARRAY_WHITE)); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 12; - u.ids.push_back(rb->depth_texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 13; - u.ids.push_back(p_normal_roughness_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 14; - RID buffer = p_gi_probe_buffer.is_valid() ? p_gi_probe_buffer : storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); - u.ids.push_back(buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 15; - u.ids.push_back(gi.sdfgi_ubo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 16; - u.ids.push_back(rb->giprobe_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 17; - for (int i = 0; i < RenderBuffers::MAX_GIPROBES; i++) { - u.ids.push_back(rb->giprobe_textures[i]); - } - uniforms.push_back(u); - } - - rb->gi_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, gi.shader.version_get_shader(gi.shader_version, 0), 0); - } + ERR_FAIL_COND_V(rb == nullptr, -1); + ERR_FAIL_COND_V(rb->sdfgi == nullptr, -1); - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, gi.pipelines[0]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->gi_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GI::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->width, rb->height, 1, 8, 8, 1); - RD::get_singleton()->compute_list_end(); + return rb->sdfgi->get_pending_region_data(p_region, from, size, bounds); } -RID RendererSceneRenderRD::sky_create() { - return sky_owner.make_rid(Sky()); +RID RendererSceneRenderRD::sky_allocate() { + return sky.allocate_sky_rid(); } - -void RendererSceneRenderRD::_sky_invalidate(Sky *p_sky) { - if (!p_sky->dirty) { - p_sky->dirty = true; - p_sky->dirty_list = dirty_sky_list; - dirty_sky_list = p_sky; - } +void RendererSceneRenderRD::sky_initialize(RID p_rid) { + sky.initialize_sky_rid(p_rid); } void RendererSceneRenderRD::sky_set_radiance_size(RID p_sky, int p_radiance_size) { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND(!sky); - ERR_FAIL_COND(p_radiance_size < 32 || p_radiance_size > 2048); - if (sky->radiance_size == p_radiance_size) { - return; - } - sky->radiance_size = p_radiance_size; - - if (sky->mode == RS::SKY_MODE_REALTIME && sky->radiance_size != 256) { - WARN_PRINT("Realtime Skies can only use a radiance size of 256. Radiance size will be set to 256 internally."); - sky->radiance_size = 256; - } - - _sky_invalidate(sky); - if (sky->radiance.is_valid()) { - RD::get_singleton()->free(sky->radiance); - sky->radiance = RID(); - } - _clear_reflection_data(sky->reflection); + sky.sky_set_radiance_size(p_sky, p_radiance_size); } void RendererSceneRenderRD::sky_set_mode(RID p_sky, RS::SkyMode p_mode) { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND(!sky); - - if (sky->mode == p_mode) { - return; - } - - sky->mode = p_mode; - - if (sky->mode == RS::SKY_MODE_REALTIME && sky->radiance_size != 256) { - WARN_PRINT("Realtime Skies can only use a radiance size of 256. Radiance size will be set to 256 internally."); - sky_set_radiance_size(p_sky, 256); - } - - _sky_invalidate(sky); - if (sky->radiance.is_valid()) { - RD::get_singleton()->free(sky->radiance); - sky->radiance = RID(); - } - _clear_reflection_data(sky->reflection); + sky.sky_set_mode(p_sky, p_mode); } void RendererSceneRenderRD::sky_set_material(RID p_sky, RID p_material) { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND(!sky); - sky->material = p_material; - _sky_invalidate(sky); + sky.sky_set_material(p_sky, p_material); } Ref<Image> RendererSceneRenderRD::sky_bake_panorama(RID p_sky, float p_energy, bool p_bake_irradiance, const Size2i &p_size) { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND_V(!sky, Ref<Image>()); - - _update_dirty_skys(); - - if (sky->radiance.is_valid()) { - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R32G32B32A32_SFLOAT; - tf.width = p_size.width; - tf.height = p_size.height; - tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; - - RID rad_tex = RD::get_singleton()->texture_create(tf, RD::TextureView()); - storage->get_effects()->copy_cubemap_to_panorama(sky->radiance, rad_tex, p_size, p_bake_irradiance ? roughness_layers : 0, sky->reflection.layers.size() > 1); - Vector<uint8_t> data = RD::get_singleton()->texture_get_data(rad_tex, 0); - RD::get_singleton()->free(rad_tex); - - Ref<Image> img; - img.instance(); - img->create(p_size.width, p_size.height, false, Image::FORMAT_RGBAF, data); - for (int i = 0; i < p_size.width; i++) { - for (int j = 0; j < p_size.height; j++) { - Color c = img->get_pixel(i, j); - c.r *= p_energy; - c.g *= p_energy; - c.b *= p_energy; - img->set_pixel(i, j, c); - } - } - return img; - } - - return Ref<Image>(); + return sky.sky_bake_panorama(p_sky, p_energy, p_bake_irradiance, p_size); } -void RendererSceneRenderRD::_update_dirty_skys() { - Sky *sky = dirty_sky_list; - - while (sky) { - bool texture_set_dirty = false; - //update sky configuration if texture is missing - - if (sky->radiance.is_null()) { - int mipmaps = Image::get_image_required_mipmaps(sky->radiance_size, sky->radiance_size, Image::FORMAT_RGBAH) + 1; - - uint32_t w = sky->radiance_size, h = sky->radiance_size; - int layers = roughness_layers; - if (sky->mode == RS::SKY_MODE_REALTIME) { - layers = 8; - if (roughness_layers != 8) { - WARN_PRINT("When using REALTIME skies, roughness_layers should be set to 8 in the project settings for best quality reflections"); - } - } - - if (sky_use_cubemap_array) { - //array (higher quality, 6 times more memory) - RD::TextureFormat tf; - tf.array_layers = layers * 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tf.texture_type = RD::TEXTURE_TYPE_CUBE_ARRAY; - tf.mipmaps = mipmaps; - tf.width = w; - tf.height = h; - tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; - - sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); - - _update_reflection_data(sky->reflection, sky->radiance_size, mipmaps, true, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME); - - } else { - //regular cubemap, lower quality (aliasing, less memory) - RD::TextureFormat tf; - tf.array_layers = 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tf.texture_type = RD::TEXTURE_TYPE_CUBE; - tf.mipmaps = MIN(mipmaps, layers); - tf.width = w; - tf.height = h; - tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; - - sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); - - _update_reflection_data(sky->reflection, sky->radiance_size, MIN(mipmaps, layers), false, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME); - } - texture_set_dirty = true; - } - - // Create subpass buffers if they haven't been created already - if (sky->half_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->half_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { - RD::TextureFormat tformat; - tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tformat.width = sky->screen_size.x / 2; - tformat.height = sky->screen_size.y / 2; - tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - tformat.texture_type = RD::TEXTURE_TYPE_2D; - - sky->half_res_pass = RD::get_singleton()->texture_create(tformat, RD::TextureView()); - Vector<RID> texs; - texs.push_back(sky->half_res_pass); - sky->half_res_framebuffer = RD::get_singleton()->framebuffer_create(texs); - texture_set_dirty = true; - } - - if (sky->quarter_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->quarter_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { - RD::TextureFormat tformat; - tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - tformat.width = sky->screen_size.x / 4; - tformat.height = sky->screen_size.y / 4; - tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - tformat.texture_type = RD::TEXTURE_TYPE_2D; - - sky->quarter_res_pass = RD::get_singleton()->texture_create(tformat, RD::TextureView()); - Vector<RID> texs; - texs.push_back(sky->quarter_res_pass); - sky->quarter_res_framebuffer = RD::get_singleton()->framebuffer_create(texs); - texture_set_dirty = true; - } - - if (texture_set_dirty) { - for (int i = 0; i < SKY_TEXTURE_SET_MAX; i++) { - if (sky->texture_uniform_sets[i].is_valid() && RD::get_singleton()->uniform_set_is_valid(sky->texture_uniform_sets[i])) { - RD::get_singleton()->free(sky->texture_uniform_sets[i]); - sky->texture_uniform_sets[i] = RID(); - } - } - } - - sky->reflection.dirty = true; - sky->processing_layer = 0; - - Sky *next = sky->dirty_list; - sky->dirty_list = nullptr; - sky->dirty = false; - sky = next; - } - - dirty_sky_list = nullptr; +RID RendererSceneRenderRD::environment_allocate() { + return environment_owner.allocate_rid(); } - -RID RendererSceneRenderRD::sky_get_radiance_texture_rd(RID p_sky) const { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND_V(!sky, RID()); - - return sky->radiance; -} - -RID RendererSceneRenderRD::sky_get_radiance_uniform_set_rd(RID p_sky, RID p_shader, int p_set) const { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND_V(!sky, RID()); - - if (sky->uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(sky->uniform_set)) { - sky->uniform_set = RID(); - if (sky->radiance.is_valid()) { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 0; - u.ids.push_back(sky->radiance); - uniforms.push_back(u); - } - - sky->uniform_set = RD::get_singleton()->uniform_set_create(uniforms, p_shader, p_set); - } - } - - return sky->uniform_set; -} - -RID RendererSceneRenderRD::_get_sky_textures(Sky *p_sky, SkyTextureSetVersion p_version) { - if (p_sky->texture_uniform_sets[p_version].is_valid() && RD::get_singleton()->uniform_set_is_valid(p_sky->texture_uniform_sets[p_version])) { - return p_sky->texture_uniform_sets[p_version]; - } - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 0; - if (p_sky->radiance.is_valid() && p_version <= SKY_TEXTURE_SET_QUARTER_RES) { - u.ids.push_back(p_sky->radiance); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 1; // half res - if (p_sky->half_res_pass.is_valid() && p_version != SKY_TEXTURE_SET_HALF_RES && p_version != SKY_TEXTURE_SET_CUBEMAP_HALF_RES) { - if (p_version >= SKY_TEXTURE_SET_CUBEMAP) { - u.ids.push_back(p_sky->reflection.layers[0].views[1]); - } else { - u.ids.push_back(p_sky->half_res_pass); - } - } else { - if (p_version < SKY_TEXTURE_SET_CUBEMAP) { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 2; // quarter res - if (p_sky->quarter_res_pass.is_valid() && p_version != SKY_TEXTURE_SET_QUARTER_RES && p_version != SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES) { - if (p_version >= SKY_TEXTURE_SET_CUBEMAP) { - u.ids.push_back(p_sky->reflection.layers[0].views[2]); - } else { - u.ids.push_back(p_sky->quarter_res_pass); - } - } else { - if (p_version < SKY_TEXTURE_SET_CUBEMAP) { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); - } - } - uniforms.push_back(u); - } - - p_sky->texture_uniform_sets[p_version] = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_TEXTURES); - return p_sky->texture_uniform_sets[p_version]; -} - -RID RendererSceneRenderRD::sky_get_material(RID p_sky) const { - Sky *sky = sky_owner.getornull(p_sky); - ERR_FAIL_COND_V(!sky, RID()); - - return sky->material; -} - -void RendererSceneRenderRD::_draw_sky(bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform) { - ERR_FAIL_COND(!is_environment(p_environment)); - - SkyMaterialData *material = nullptr; - - Sky *sky = sky_owner.getornull(environment_get_sky(p_environment)); - - RID sky_material; - - RS::EnvironmentBG background = environment_get_background(p_environment); - - if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { - ERR_FAIL_COND(!sky); - sky_material = sky_get_material(environment_get_sky(p_environment)); - - if (sky_material.is_valid()) { - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - if (!material || !material->shader_data->valid) { - material = nullptr; - } - } - - if (!material) { - sky_material = sky_shader.default_material; - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - } - } - - if (background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) { - sky_material = sky_scene_state.fog_material; - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - } - - ERR_FAIL_COND(!material); - - SkyShaderData *shader_data = material->shader_data; - - ERR_FAIL_COND(!shader_data); - - Basis sky_transform = environment_get_sky_orientation(p_environment); - sky_transform.invert(); - - float multiplier = environment_get_bg_energy(p_environment); - float custom_fov = environment_get_sky_custom_fov(p_environment); - // Camera - CameraMatrix camera; - - if (custom_fov) { - float near_plane = p_projection.get_z_near(); - float far_plane = p_projection.get_z_far(); - float aspect = p_projection.get_aspect(); - - camera.set_perspective(custom_fov, aspect, near_plane, far_plane); - - } else { - camera = p_projection; - } - - sky_transform = p_transform.basis * sky_transform; - - if (shader_data->uses_quarter_res) { - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_QUARTER_RES]; - - RID texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_QUARTER_RES); - - Vector<Color> clear_colors; - clear_colors.push_back(Color(0.0, 0.0, 0.0)); - - RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->quarter_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); - storage->get_effects()->render_sky(draw_list, time, sky->quarter_res_framebuffer, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); - } - - if (shader_data->uses_half_res) { - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_HALF_RES]; - - RID texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_HALF_RES); - - Vector<Color> clear_colors; - clear_colors.push_back(Color(0.0, 0.0, 0.0)); - - RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->half_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); - storage->get_effects()->render_sky(draw_list, time, sky->half_res_framebuffer, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); - } - - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_BACKGROUND]; - - RID texture_uniform_set; - if (sky) { - texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_BACKGROUND); - } else { - texture_uniform_set = sky_scene_state.fog_only_texture_uniform_set; - } - - RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_fb, RD::INITIAL_ACTION_CONTINUE, p_can_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, p_can_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ); - storage->get_effects()->render_sky(draw_list, time, p_fb, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); -} - -void RendererSceneRenderRD::_setup_sky(RID p_environment, RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform, const Size2i p_screen_size) { - ERR_FAIL_COND(!is_environment(p_environment)); - - SkyMaterialData *material = nullptr; - - Sky *sky = sky_owner.getornull(environment_get_sky(p_environment)); - - RID sky_material; - - SkyShaderData *shader_data = nullptr; - - RS::EnvironmentBG background = environment_get_background(p_environment); - - if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { - ERR_FAIL_COND(!sky); - sky_material = sky_get_material(environment_get_sky(p_environment)); - - if (sky_material.is_valid()) { - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - if (!material || !material->shader_data->valid) { - material = nullptr; - } - } - - if (!material) { - sky_material = sky_shader.default_material; - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - } - - ERR_FAIL_COND(!material); - - shader_data = material->shader_data; - - ERR_FAIL_COND(!shader_data); - } - - if (sky) { - // Invalidate supbass buffers if screen size changes - if (sky->screen_size != p_screen_size) { - sky->screen_size = p_screen_size; - sky->screen_size.x = sky->screen_size.x < 4 ? 4 : sky->screen_size.x; - sky->screen_size.y = sky->screen_size.y < 4 ? 4 : sky->screen_size.y; - if (shader_data->uses_half_res) { - if (sky->half_res_pass.is_valid()) { - RD::get_singleton()->free(sky->half_res_pass); - sky->half_res_pass = RID(); - } - _sky_invalidate(sky); - } - if (shader_data->uses_quarter_res) { - if (sky->quarter_res_pass.is_valid()) { - RD::get_singleton()->free(sky->quarter_res_pass); - sky->quarter_res_pass = RID(); - } - _sky_invalidate(sky); - } - } - - // Create new subpass buffers if necessary - if ((shader_data->uses_half_res && sky->half_res_pass.is_null()) || - (shader_data->uses_quarter_res && sky->quarter_res_pass.is_null()) || - sky->radiance.is_null()) { - _sky_invalidate(sky); - _update_dirty_skys(); - } - - if (shader_data->uses_time && time - sky->prev_time > 0.00001) { - sky->prev_time = time; - sky->reflection.dirty = true; - RenderingServerDefault::redraw_request(); - } - - if (material != sky->prev_material) { - sky->prev_material = material; - sky->reflection.dirty = true; - } - - if (material->uniform_set_updated) { - material->uniform_set_updated = false; - sky->reflection.dirty = true; - } - - if (!p_transform.origin.is_equal_approx(sky->prev_position) && shader_data->uses_position) { - sky->prev_position = p_transform.origin; - sky->reflection.dirty = true; - } - - if (shader_data->uses_light) { - // Check whether the directional_light_buffer changes - bool light_data_dirty = false; - - if (sky_scene_state.ubo.directional_light_count != sky_scene_state.last_frame_directional_light_count) { - light_data_dirty = true; - for (uint32_t i = sky_scene_state.ubo.directional_light_count; i < sky_scene_state.max_directional_lights; i++) { - sky_scene_state.directional_lights[i].enabled = false; - } - } - if (!light_data_dirty) { - for (uint32_t i = 0; i < sky_scene_state.ubo.directional_light_count; i++) { - if (sky_scene_state.directional_lights[i].direction[0] != sky_scene_state.last_frame_directional_lights[i].direction[0] || - sky_scene_state.directional_lights[i].direction[1] != sky_scene_state.last_frame_directional_lights[i].direction[1] || - sky_scene_state.directional_lights[i].direction[2] != sky_scene_state.last_frame_directional_lights[i].direction[2] || - sky_scene_state.directional_lights[i].energy != sky_scene_state.last_frame_directional_lights[i].energy || - sky_scene_state.directional_lights[i].color[0] != sky_scene_state.last_frame_directional_lights[i].color[0] || - sky_scene_state.directional_lights[i].color[1] != sky_scene_state.last_frame_directional_lights[i].color[1] || - sky_scene_state.directional_lights[i].color[2] != sky_scene_state.last_frame_directional_lights[i].color[2] || - sky_scene_state.directional_lights[i].enabled != sky_scene_state.last_frame_directional_lights[i].enabled || - sky_scene_state.directional_lights[i].size != sky_scene_state.last_frame_directional_lights[i].size) { - light_data_dirty = true; - break; - } - } - } - - if (light_data_dirty) { - RD::get_singleton()->buffer_update(sky_scene_state.directional_light_buffer, 0, sizeof(SkyDirectionalLightData) * sky_scene_state.max_directional_lights, sky_scene_state.directional_lights, true); - - RendererSceneRenderRD::SkyDirectionalLightData *temp = sky_scene_state.last_frame_directional_lights; - sky_scene_state.last_frame_directional_lights = sky_scene_state.directional_lights; - sky_scene_state.directional_lights = temp; - sky_scene_state.last_frame_directional_light_count = sky_scene_state.ubo.directional_light_count; - sky->reflection.dirty = true; - } - } - } - - //setup fog variables - sky_scene_state.ubo.volumetric_fog_enabled = false; - if (p_render_buffers.is_valid()) { - if (render_buffers_has_volumetric_fog(p_render_buffers)) { - sky_scene_state.ubo.volumetric_fog_enabled = true; - - float fog_end = render_buffers_get_volumetric_fog_end(p_render_buffers); - if (fog_end > 0.0) { - sky_scene_state.ubo.volumetric_fog_inv_length = 1.0 / fog_end; - } else { - sky_scene_state.ubo.volumetric_fog_inv_length = 1.0; - } - - float fog_detail_spread = render_buffers_get_volumetric_fog_detail_spread(p_render_buffers); //reverse lookup - if (fog_detail_spread > 0.0) { - sky_scene_state.ubo.volumetric_fog_detail_spread = 1.0 / fog_detail_spread; - } else { - sky_scene_state.ubo.volumetric_fog_detail_spread = 1.0; - } - } - - RID fog_uniform_set = render_buffers_get_volumetric_fog_sky_uniform_set(p_render_buffers); - - if (fog_uniform_set != RID()) { - sky_scene_state.fog_uniform_set = fog_uniform_set; - } else { - sky_scene_state.fog_uniform_set = sky_scene_state.default_fog_uniform_set; - } - } - - sky_scene_state.ubo.z_far = p_projection.get_z_far(); - sky_scene_state.ubo.fog_enabled = environment_is_fog_enabled(p_environment); - sky_scene_state.ubo.fog_density = environment_get_fog_density(p_environment); - sky_scene_state.ubo.fog_aerial_perspective = environment_get_fog_aerial_perspective(p_environment); - Color fog_color = environment_get_fog_light_color(p_environment).to_linear(); - float fog_energy = environment_get_fog_light_energy(p_environment); - sky_scene_state.ubo.fog_light_color[0] = fog_color.r * fog_energy; - sky_scene_state.ubo.fog_light_color[1] = fog_color.g * fog_energy; - sky_scene_state.ubo.fog_light_color[2] = fog_color.b * fog_energy; - sky_scene_state.ubo.fog_sun_scatter = environment_get_fog_sun_scatter(p_environment); - - RD::get_singleton()->buffer_update(sky_scene_state.uniform_buffer, 0, sizeof(SkySceneState::UBO), &sky_scene_state.ubo, true); -} - -void RendererSceneRenderRD::_update_sky(RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform) { - ERR_FAIL_COND(!is_environment(p_environment)); - - Sky *sky = sky_owner.getornull(environment_get_sky(p_environment)); - ERR_FAIL_COND(!sky); - - RID sky_material = sky_get_material(environment_get_sky(p_environment)); - - SkyMaterialData *material = nullptr; - - if (sky_material.is_valid()) { - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - if (!material || !material->shader_data->valid) { - material = nullptr; - } - } - - if (!material) { - sky_material = sky_shader.default_material; - material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); - } - - ERR_FAIL_COND(!material); - - SkyShaderData *shader_data = material->shader_data; - - ERR_FAIL_COND(!shader_data); - - float multiplier = environment_get_bg_energy(p_environment); - - bool update_single_frame = sky->mode == RS::SKY_MODE_REALTIME || sky->mode == RS::SKY_MODE_QUALITY; - RS::SkyMode sky_mode = sky->mode; - - if (sky_mode == RS::SKY_MODE_AUTOMATIC) { - if (shader_data->uses_time || shader_data->uses_position) { - update_single_frame = true; - sky_mode = RS::SKY_MODE_REALTIME; - } else if (shader_data->uses_light || shader_data->ubo_size > 0) { - update_single_frame = false; - sky_mode = RS::SKY_MODE_INCREMENTAL; - } else { - update_single_frame = true; - sky_mode = RS::SKY_MODE_QUALITY; - } - } - - if (sky->processing_layer == 0 && sky_mode == RS::SKY_MODE_INCREMENTAL) { - // On the first frame after creating sky, rebuild in single frame - update_single_frame = true; - sky_mode = RS::SKY_MODE_QUALITY; - } - - int max_processing_layer = sky_use_cubemap_array ? sky->reflection.layers.size() : sky->reflection.layers[0].mipmaps.size(); - - // Update radiance cubemap - if (sky->reflection.dirty && (sky->processing_layer >= max_processing_layer || update_single_frame)) { - static const Vector3 view_normals[6] = { - Vector3(+1, 0, 0), - Vector3(-1, 0, 0), - Vector3(0, +1, 0), - Vector3(0, -1, 0), - Vector3(0, 0, +1), - Vector3(0, 0, -1) - }; - static const Vector3 view_up[6] = { - Vector3(0, -1, 0), - Vector3(0, -1, 0), - Vector3(0, 0, +1), - Vector3(0, 0, -1), - Vector3(0, -1, 0), - Vector3(0, -1, 0) - }; - - CameraMatrix cm; - cm.set_perspective(90, 1, 0.01, 10.0); - CameraMatrix correction; - correction.set_depth_correction(true); - cm = correction * cm; - - if (shader_data->uses_quarter_res) { - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_QUARTER_RES]; - - Vector<Color> clear_colors; - clear_colors.push_back(Color(0.0, 0.0, 0.0)); - RD::DrawListID cubemap_draw_list; - - for (int i = 0; i < 6; i++) { - Transform local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); - RID texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES); - - cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[2].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - storage->get_effects()->render_sky(cubemap_draw_list, time, sky->reflection.layers[0].mipmaps[2].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); - } - } - - if (shader_data->uses_half_res) { - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_HALF_RES]; - - Vector<Color> clear_colors; - clear_colors.push_back(Color(0.0, 0.0, 0.0)); - RD::DrawListID cubemap_draw_list; - - for (int i = 0; i < 6; i++) { - Transform local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); - RID texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_CUBEMAP_HALF_RES); - - cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[1].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - storage->get_effects()->render_sky(cubemap_draw_list, time, sky->reflection.layers[0].mipmaps[1].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); - } - } - - RD::DrawListID cubemap_draw_list; - PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP]; - - for (int i = 0; i < 6; i++) { - Transform local_view; - local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); - RID texture_uniform_set = _get_sky_textures(sky, SKY_TEXTURE_SET_CUBEMAP); - - cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[0].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); - storage->get_effects()->render_sky(cubemap_draw_list, time, sky->reflection.layers[0].mipmaps[0].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); - RD::get_singleton()->draw_list_end(); - } - - if (sky_mode == RS::SKY_MODE_REALTIME) { - _create_reflection_fast_filter(sky->reflection, sky_use_cubemap_array); - if (sky_use_cubemap_array) { - _update_reflection_mipmaps(sky->reflection, 0, sky->reflection.layers.size()); - } - } else { - if (update_single_frame) { - for (int i = 1; i < max_processing_layer; i++) { - _create_reflection_importance_sample(sky->reflection, sky_use_cubemap_array, 10, i); - } - if (sky_use_cubemap_array) { - _update_reflection_mipmaps(sky->reflection, 0, sky->reflection.layers.size()); - } - } else { - if (sky_use_cubemap_array) { - // Multi-Frame so just update the first array level - _update_reflection_mipmaps(sky->reflection, 0, 1); - } - } - sky->processing_layer = 1; - } - - sky->reflection.dirty = false; - - } else { - if (sky_mode == RS::SKY_MODE_INCREMENTAL && sky->processing_layer < max_processing_layer) { - _create_reflection_importance_sample(sky->reflection, sky_use_cubemap_array, 10, sky->processing_layer); - - if (sky_use_cubemap_array) { - _update_reflection_mipmaps(sky->reflection, sky->processing_layer, sky->processing_layer + 1); - } - - sky->processing_layer++; - } - } -} - -/* SKY SHADER */ - -void RendererSceneRenderRD::SkyShaderData::set_code(const String &p_code) { - //compile - - code = p_code; - valid = false; - ubo_size = 0; - uniforms.clear(); - - if (code == String()) { - return; //just invalid, but no error - } - - ShaderCompilerRD::GeneratedCode gen_code; - ShaderCompilerRD::IdentifierActions actions; - - uses_time = false; - uses_half_res = false; - uses_quarter_res = false; - uses_position = false; - uses_light = false; - - actions.render_mode_flags["use_half_res_pass"] = &uses_half_res; - actions.render_mode_flags["use_quarter_res_pass"] = &uses_quarter_res; - - actions.usage_flag_pointers["TIME"] = &uses_time; - actions.usage_flag_pointers["POSITION"] = &uses_position; - actions.usage_flag_pointers["LIGHT0_ENABLED"] = &uses_light; - actions.usage_flag_pointers["LIGHT0_ENERGY"] = &uses_light; - actions.usage_flag_pointers["LIGHT0_DIRECTION"] = &uses_light; - actions.usage_flag_pointers["LIGHT0_COLOR"] = &uses_light; - actions.usage_flag_pointers["LIGHT0_SIZE"] = &uses_light; - actions.usage_flag_pointers["LIGHT1_ENABLED"] = &uses_light; - actions.usage_flag_pointers["LIGHT1_ENERGY"] = &uses_light; - actions.usage_flag_pointers["LIGHT1_DIRECTION"] = &uses_light; - actions.usage_flag_pointers["LIGHT1_COLOR"] = &uses_light; - actions.usage_flag_pointers["LIGHT1_SIZE"] = &uses_light; - actions.usage_flag_pointers["LIGHT2_ENABLED"] = &uses_light; - actions.usage_flag_pointers["LIGHT2_ENERGY"] = &uses_light; - actions.usage_flag_pointers["LIGHT2_DIRECTION"] = &uses_light; - actions.usage_flag_pointers["LIGHT2_COLOR"] = &uses_light; - actions.usage_flag_pointers["LIGHT2_SIZE"] = &uses_light; - actions.usage_flag_pointers["LIGHT3_ENABLED"] = &uses_light; - actions.usage_flag_pointers["LIGHT3_ENERGY"] = &uses_light; - actions.usage_flag_pointers["LIGHT3_DIRECTION"] = &uses_light; - actions.usage_flag_pointers["LIGHT3_COLOR"] = &uses_light; - actions.usage_flag_pointers["LIGHT3_SIZE"] = &uses_light; - - actions.uniforms = &uniforms; - - RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; - - Error err = scene_singleton->sky_shader.compiler.compile(RS::SHADER_SKY, code, &actions, path, gen_code); - - ERR_FAIL_COND(err != OK); - - if (version.is_null()) { - version = scene_singleton->sky_shader.shader.version_create(); - } - -#if 0 - print_line("**compiling shader:"); - print_line("**defines:\n"); - for (int i = 0; i < gen_code.defines.size(); i++) { - print_line(gen_code.defines[i]); - } - print_line("\n**uniforms:\n" + gen_code.uniforms); - // print_line("\n**vertex_globals:\n" + gen_code.vertex_global); - // print_line("\n**vertex_code:\n" + gen_code.vertex); - print_line("\n**fragment_globals:\n" + gen_code.fragment_global); - print_line("\n**fragment_code:\n" + gen_code.fragment); - print_line("\n**light_code:\n" + gen_code.light); -#endif - - scene_singleton->sky_shader.shader.version_set_code(version, gen_code.uniforms, gen_code.vertex_global, gen_code.vertex, gen_code.fragment_global, gen_code.light, gen_code.fragment, gen_code.defines); - ERR_FAIL_COND(!scene_singleton->sky_shader.shader.version_is_valid(version)); - - ubo_size = gen_code.uniform_total_size; - ubo_offsets = gen_code.uniform_offsets; - texture_uniforms = gen_code.texture_uniforms; - - //update pipelines - - for (int i = 0; i < SKY_VERSION_MAX; i++) { - RD::PipelineDepthStencilState depth_stencil_state; - depth_stencil_state.enable_depth_test = true; - depth_stencil_state.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; - - RID shader_variant = scene_singleton->sky_shader.shader.version_get_shader(version, i); - pipelines[i].setup(shader_variant, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), depth_stencil_state, RD::PipelineColorBlendState::create_disabled(), 0); - } - - valid = true; -} - -void RendererSceneRenderRD::SkyShaderData::set_default_texture_param(const StringName &p_name, RID p_texture) { - if (!p_texture.is_valid()) { - default_texture_params.erase(p_name); - } else { - default_texture_params[p_name] = p_texture; - } -} - -void RendererSceneRenderRD::SkyShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { - Map<int, StringName> order; - - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { - continue; - } - - if (E->get().texture_order >= 0) { - order[E->get().texture_order + 100000] = E->key(); - } else { - order[E->get().order] = E->key(); - } - } - - for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { - PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); - pi.name = E->get(); - p_param_list->push_back(pi); - } -} - -void RendererSceneRenderRD::SkyShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { - for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { - if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { - continue; - } - - RendererStorage::InstanceShaderParam p; - p.info = ShaderLanguage::uniform_to_property_info(E->get()); - p.info.name = E->key(); //supply name - p.index = E->get().instance_index; - p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); - p_param_list->push_back(p); - } -} - -bool RendererSceneRenderRD::SkyShaderData::is_param_texture(const StringName &p_param) const { - if (!uniforms.has(p_param)) { - return false; - } - - return uniforms[p_param].texture_order >= 0; -} - -bool RendererSceneRenderRD::SkyShaderData::is_animated() const { - return false; -} - -bool RendererSceneRenderRD::SkyShaderData::casts_shadows() const { - return false; -} - -Variant RendererSceneRenderRD::SkyShaderData::get_default_parameter(const StringName &p_parameter) const { - if (uniforms.has(p_parameter)) { - ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; - Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; - return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); - } - return Variant(); -} - -RendererSceneRenderRD::SkyShaderData::SkyShaderData() { - valid = false; -} - -RendererSceneRenderRD::SkyShaderData::~SkyShaderData() { - RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; - ERR_FAIL_COND(!scene_singleton); - //pipeline variants will clear themselves if shader is gone - if (version.is_valid()) { - scene_singleton->sky_shader.shader.version_free(version); - } -} - -RendererStorageRD::ShaderData *RendererSceneRenderRD::_create_sky_shader_func() { - SkyShaderData *shader_data = memnew(SkyShaderData); - return shader_data; -} - -void RendererSceneRenderRD::SkyMaterialData::update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty) { - RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; - - uniform_set_updated = true; - - if ((uint32_t)ubo_data.size() != shader_data->ubo_size) { - p_uniform_dirty = true; - if (uniform_buffer.is_valid()) { - RD::get_singleton()->free(uniform_buffer); - uniform_buffer = RID(); - } - - ubo_data.resize(shader_data->ubo_size); - if (ubo_data.size()) { - uniform_buffer = RD::get_singleton()->uniform_buffer_create(ubo_data.size()); - memset(ubo_data.ptrw(), 0, ubo_data.size()); //clear - } - - //clear previous uniform set - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - uniform_set = RID(); - } - } - - //check whether buffer changed - if (p_uniform_dirty && ubo_data.size()) { - update_uniform_buffer(shader_data->uniforms, shader_data->ubo_offsets.ptr(), p_parameters, ubo_data.ptrw(), ubo_data.size(), false); - RD::get_singleton()->buffer_update(uniform_buffer, 0, ubo_data.size(), ubo_data.ptrw()); - } - - uint32_t tex_uniform_count = shader_data->texture_uniforms.size(); - - if ((uint32_t)texture_cache.size() != tex_uniform_count) { - texture_cache.resize(tex_uniform_count); - p_textures_dirty = true; - - //clear previous uniform set - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - uniform_set = RID(); - } - } - - if (p_textures_dirty && tex_uniform_count) { - update_textures(p_parameters, shader_data->default_texture_params, shader_data->texture_uniforms, texture_cache.ptrw(), true); - } - - if (shader_data->ubo_size == 0 && shader_data->texture_uniforms.size() == 0) { - // This material does not require an uniform set, so don't create it. - return; - } - - if (!p_textures_dirty && uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - //no reason to update uniform set, only UBO (or nothing) was needed to update - return; - } - - Vector<RD::Uniform> uniforms; - - { - if (shader_data->ubo_size) { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 0; - u.ids.push_back(uniform_buffer); - uniforms.push_back(u); - } - - const RID *textures = texture_cache.ptrw(); - for (uint32_t i = 0; i < tex_uniform_count; i++) { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 1 + i; - u.ids.push_back(textures[i]); - uniforms.push_back(u); - } - } - - uniform_set = RD::get_singleton()->uniform_set_create(uniforms, scene_singleton->sky_shader.shader.version_get_shader(shader_data->version, 0), SKY_SET_MATERIAL); -} - -RendererSceneRenderRD::SkyMaterialData::~SkyMaterialData() { - if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { - RD::get_singleton()->free(uniform_set); - } - - if (uniform_buffer.is_valid()) { - RD::get_singleton()->free(uniform_buffer); - } -} - -RendererStorageRD::MaterialData *RendererSceneRenderRD::_create_sky_material_func(SkyShaderData *p_shader) { - SkyMaterialData *material_data = memnew(SkyMaterialData); - material_data->shader_data = p_shader; - material_data->last_frame = false; - //update will happen later anyway so do nothing. - return material_data; -} - -RID RendererSceneRenderRD::environment_create() { - return environment_owner.make_rid(Environment()); +void RendererSceneRenderRD::environment_initialize(RID p_rid) { + environment_owner.initialize_rid(p_rid, RendererSceneEnvironmentRD()); } void RendererSceneRenderRD::environment_set_background(RID p_env, RS::EnvironmentBG p_bg) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->background = p_bg; } void RendererSceneRenderRD::environment_set_sky(RID p_env, RID p_sky) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->sky = p_sky; } void RendererSceneRenderRD::environment_set_sky_custom_fov(RID p_env, float p_scale) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->sky_custom_fov = p_scale; } void RendererSceneRenderRD::environment_set_sky_orientation(RID p_env, const Basis &p_orientation) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->sky_orientation = p_orientation; } void RendererSceneRenderRD::environment_set_bg_color(RID p_env, const Color &p_color) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->bg_color = p_color; } void RendererSceneRenderRD::environment_set_bg_energy(RID p_env, float p_energy) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->bg_energy = p_energy; } void RendererSceneRenderRD::environment_set_canvas_max_layer(RID p_env, int p_max_layer) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->canvas_max_layer = p_max_layer; } void RendererSceneRenderRD::environment_set_ambient_light(RID p_env, const Color &p_color, RS::EnvironmentAmbientSource p_ambient, float p_energy, float p_sky_contribution, RS::EnvironmentReflectionSource p_reflection_source, const Color &p_ao_color) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); - env->ambient_light = p_color; - env->ambient_source = p_ambient; - env->ambient_light_energy = p_energy; - env->ambient_sky_contribution = p_sky_contribution; - env->reflection_source = p_reflection_source; - env->ao_color = p_ao_color; + env->set_ambient_light(p_color, p_ambient, p_energy, p_sky_contribution, p_reflection_source, p_ao_color); } RS::EnvironmentBG RendererSceneRenderRD::environment_get_background(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, RS::ENV_BG_MAX); return env->background; } RID RendererSceneRenderRD::environment_get_sky(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, RID()); return env->sky; } float RendererSceneRenderRD::environment_get_sky_custom_fov(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->sky_custom_fov; } Basis RendererSceneRenderRD::environment_get_sky_orientation(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Basis()); return env->sky_orientation; } Color RendererSceneRenderRD::environment_get_bg_color(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Color()); return env->bg_color; } float RendererSceneRenderRD::environment_get_bg_energy(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->bg_energy; } int RendererSceneRenderRD::environment_get_canvas_max_layer(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->canvas_max_layer; } Color RendererSceneRenderRD::environment_get_ambient_light_color(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Color()); return env->ambient_light; } RS::EnvironmentAmbientSource RendererSceneRenderRD::environment_get_ambient_source(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, RS::ENV_AMBIENT_SOURCE_BG); return env->ambient_source; } float RendererSceneRenderRD::environment_get_ambient_light_energy(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->ambient_light_energy; } float RendererSceneRenderRD::environment_get_ambient_sky_contribution(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->ambient_sky_contribution; } RS::EnvironmentReflectionSource RendererSceneRenderRD::environment_get_reflection_source(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, RS::ENV_REFLECTION_SOURCE_DISABLED); return env->reflection_source; } Color RendererSceneRenderRD::environment_get_ao_color(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Color()); return env->ao_color; } void RendererSceneRenderRD::environment_set_tonemap(RID p_env, RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); - env->exposure = p_exposure; - env->tone_mapper = p_tone_mapper; - if (!env->auto_exposure && p_auto_exposure) { - env->auto_exposure_version = ++auto_exposure_counter; - } - env->auto_exposure = p_auto_exposure; - env->white = p_white; - env->min_luminance = p_min_luminance; - env->max_luminance = p_max_luminance; - env->auto_exp_speed = p_auto_exp_speed; - env->auto_exp_scale = p_auto_exp_scale; + env->set_tonemap(p_tone_mapper, p_exposure, p_white, p_auto_exposure, p_min_luminance, p_max_luminance, p_auto_exp_speed, p_auto_exp_scale); } void RendererSceneRenderRD::environment_set_glow(RID p_env, bool p_enable, Vector<float> p_levels, float p_intensity, float p_strength, float p_mix, float p_bloom_threshold, RS::EnvironmentGlowBlendMode p_blend_mode, float p_hdr_bleed_threshold, float p_hdr_bleed_scale, float p_hdr_luminance_cap) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); - ERR_FAIL_COND_MSG(p_levels.size() != 7, "Size of array of glow levels must be 7"); - env->glow_enabled = p_enable; - env->glow_levels = p_levels; - env->glow_intensity = p_intensity; - env->glow_strength = p_strength; - env->glow_mix = p_mix; - env->glow_bloom = p_bloom_threshold; - env->glow_blend_mode = p_blend_mode; - env->glow_hdr_bleed_threshold = p_hdr_bleed_threshold; - env->glow_hdr_bleed_scale = p_hdr_bleed_scale; - env->glow_hdr_luminance_cap = p_hdr_luminance_cap; + env->set_glow(p_enable, p_levels, p_intensity, p_strength, p_mix, p_bloom_threshold, p_blend_mode, p_hdr_bleed_threshold, p_hdr_bleed_scale, p_hdr_luminance_cap); } void RendererSceneRenderRD::environment_glow_set_use_bicubic_upscale(bool p_enable) { @@ -2967,100 +313,77 @@ void RendererSceneRenderRD::environment_glow_set_use_high_quality(bool p_enable) glow_high_quality = p_enable; } -void RendererSceneRenderRD::environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, bool p_use_multibounce, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) { - Environment *env = environment_owner.getornull(p_env); +void RendererSceneRenderRD::environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) { + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); if (low_end) { return; } - env->sdfgi_enabled = p_enable; - env->sdfgi_cascades = p_cascades; - env->sdfgi_min_cell_size = p_min_cell_size; - env->sdfgi_use_occlusion = p_use_occlusion; - env->sdfgi_use_multibounce = p_use_multibounce; - env->sdfgi_read_sky_light = p_read_sky; - env->sdfgi_energy = p_energy; - env->sdfgi_normal_bias = p_normal_bias; - env->sdfgi_probe_bias = p_probe_bias; - env->sdfgi_y_scale = p_y_scale; + env->set_sdfgi(p_enable, p_cascades, p_min_cell_size, p_y_scale, p_use_occlusion, p_bounce_feedback, p_read_sky, p_energy, p_normal_bias, p_probe_bias); } void RendererSceneRenderRD::environment_set_fog(RID p_env, bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_fog_aerial_perspective) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); - env->fog_enabled = p_enable; - env->fog_light_color = p_light_color; - env->fog_light_energy = p_light_energy; - env->fog_sun_scatter = p_sun_scatter; - env->fog_density = p_density; - env->fog_height = p_height; - env->fog_height_density = p_height_density; - env->fog_aerial_perspective = p_fog_aerial_perspective; + env->set_fog(p_enable, p_light_color, p_light_energy, p_sun_scatter, p_density, p_height, p_height_density, p_fog_aerial_perspective); } bool RendererSceneRenderRD::environment_is_fog_enabled(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, false); return env->fog_enabled; } Color RendererSceneRenderRD::environment_get_fog_light_color(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Color()); return env->fog_light_color; } float RendererSceneRenderRD::environment_get_fog_light_energy(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_light_energy; } float RendererSceneRenderRD::environment_get_fog_sun_scatter(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_sun_scatter; } float RendererSceneRenderRD::environment_get_fog_density(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_density; } float RendererSceneRenderRD::environment_get_fog_height(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_height; } float RendererSceneRenderRD::environment_get_fog_height_density(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_height_density; } float RendererSceneRenderRD::environment_get_fog_aerial_perspective(RID p_env) const { - const Environment *env = environment_owner.getornull(p_env); + const RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0); return env->fog_aerial_perspective; } -void RendererSceneRenderRD::environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, RenderingServer::EnvVolumetricFogShadowFilter p_shadow_filter) { - Environment *env = environment_owner.getornull(p_env); +void RendererSceneRenderRD::environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) { + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); if (low_end) { return; } - env->volumetric_fog_enabled = p_enable; - env->volumetric_fog_density = p_density; - env->volumetric_fog_light = p_light; - env->volumetric_fog_light_energy = p_light_energy; - env->volumetric_fog_length = p_length; - env->volumetric_fog_detail_spread = p_detail_spread; - env->volumetric_fog_shadow_filter = p_shadow_filter; - env->volumetric_fog_gi_inject = p_gi_inject; + env->set_volumetric_fog(p_enable, p_density, p_light, p_light_energy, p_length, p_detail_spread, p_gi_inject, p_temporal_reprojection, p_temporal_reprojection_amount); } void RendererSceneRenderRD::environment_set_volumetric_fog_volume_size(int p_size, int p_depth) { @@ -3071,47 +394,27 @@ void RendererSceneRenderRD::environment_set_volumetric_fog_volume_size(int p_siz void RendererSceneRenderRD::environment_set_volumetric_fog_filter_active(bool p_enable) { volumetric_fog_filter_active = p_enable; } -void RendererSceneRenderRD::environment_set_volumetric_fog_directional_shadow_shrink_size(int p_shrink_size) { - p_shrink_size = nearest_power_of_2_templated(p_shrink_size); - if (volumetric_fog_directional_shadow_shrink == (uint32_t)p_shrink_size) { - return; - } - - _clear_shadow_shrink_stages(directional_shadow.shrink_stages); -} -void RendererSceneRenderRD::environment_set_volumetric_fog_positional_shadow_shrink_size(int p_shrink_size) { - p_shrink_size = nearest_power_of_2_templated(p_shrink_size); - if (volumetric_fog_positional_shadow_shrink == (uint32_t)p_shrink_size) { - return; - } - - for (uint32_t i = 0; i < shadow_atlas_owner.get_rid_count(); i++) { - ShadowAtlas *sa = shadow_atlas_owner.get_ptr_by_index(i); - _clear_shadow_shrink_stages(sa->shrink_stages); - } -} void RendererSceneRenderRD::environment_set_sdfgi_ray_count(RS::EnvironmentSDFGIRayCount p_ray_count) { - sdfgi_ray_count = p_ray_count; + gi.sdfgi_ray_count = p_ray_count; } void RendererSceneRenderRD::environment_set_sdfgi_frames_to_converge(RS::EnvironmentSDFGIFramesToConverge p_frames) { - sdfgi_frames_to_converge = p_frames; + gi.sdfgi_frames_to_converge = p_frames; +} +void RendererSceneRenderRD::environment_set_sdfgi_frames_to_update_light(RS::EnvironmentSDFGIFramesToUpdateLight p_update) { + gi.sdfgi_frames_to_update_light = p_update; } void RendererSceneRenderRD::environment_set_ssr(RID p_env, bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); if (low_end) { return; } - env->ssr_enabled = p_enable; - env->ssr_max_steps = p_max_steps; - env->ssr_fade_in = p_fade_int; - env->ssr_fade_out = p_fade_out; - env->ssr_depth_tolerance = p_depth_tolerance; + env->set_ssr(p_enable, p_max_steps, p_fade_int, p_fade_out, p_depth_tolerance); } void RendererSceneRenderRD::environment_set_ssr_roughness_quality(RS::EnvironmentSSRRoughnessQuality p_quality) { @@ -3123,22 +426,14 @@ RS::EnvironmentSSRRoughnessQuality RendererSceneRenderRD::environment_get_ssr_ro } void RendererSceneRenderRD::environment_set_ssao(RID p_env, bool p_enable, float p_radius, float p_intensity, float p_power, float p_detail, float p_horizon, float p_sharpness, float p_light_affect, float p_ao_channel_affect) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); if (low_end) { return; } - env->ssao_enabled = p_enable; - env->ssao_radius = p_radius; - env->ssao_intensity = p_intensity; - env->ssao_power = p_power; - env->ssao_detail = p_detail; - env->ssao_horizon = p_horizon; - env->ssao_sharpness = p_sharpness; - env->ssao_direct_light_affect = p_light_affect; - env->ssao_ao_channel_affect = p_ao_channel_affect; + env->set_ssao(p_enable, p_radius, p_intensity, p_power, p_detail, p_horizon, p_sharpness, p_light_affect, p_ao_channel_affect); } void RendererSceneRenderRD::environment_set_ssao_quality(RS::EnvironmentSSAOQuality p_quality, bool p_half_size, float p_adaptive_target, int p_blur_passes, float p_fadeout_from, float p_fadeout_to) { @@ -3151,30 +446,30 @@ void RendererSceneRenderRD::environment_set_ssao_quality(RS::EnvironmentSSAOQual } bool RendererSceneRenderRD::environment_is_ssao_enabled(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, false); return env->ssao_enabled; } float RendererSceneRenderRD::environment_get_ssao_ao_affect(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0.0); return env->ssao_ao_channel_affect; } float RendererSceneRenderRD::environment_get_ssao_light_affect(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, 0.0); return env->ssao_direct_light_affect; } bool RendererSceneRenderRD::environment_is_ssr_enabled(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, false); return env->ssr_enabled; } bool RendererSceneRenderRD::environment_is_sdfgi_enabled(RID p_env) const { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, false); return env->sdfgi_enabled; } @@ -3184,7 +479,7 @@ bool RendererSceneRenderRD::is_environment(RID p_env) const { } Ref<Image> RendererSceneRenderRD::environment_bake_panorama(RID p_env, bool p_bake_irradiance, const Size2i &p_size) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND_V(!env, Ref<Image>()); if (env->background == RS::ENV_BG_CAMERA_FEED || env->background == RS::ENV_BG_CANVAS || env->background == RS::ENV_BG_KEEP) { @@ -3224,8 +519,16 @@ Ref<Image> RendererSceneRenderRD::environment_bake_panorama(RID p_env, bool p_ba RID RendererSceneRenderRD::reflection_atlas_create() { ReflectionAtlas ra; - ra.count = GLOBAL_GET("rendering/quality/reflection_atlas/reflection_count"); - ra.size = GLOBAL_GET("rendering/quality/reflection_atlas/reflection_size"); + ra.count = GLOBAL_GET("rendering/reflections/reflection_atlas/reflection_count"); + ra.size = GLOBAL_GET("rendering/reflections/reflection_atlas/reflection_size"); + + if (is_clustered_enabled()) { + ra.cluster_builder = memnew(ClusterBuilderRD); + ra.cluster_builder->set_shared(&cluster_builder_shared); + ra.cluster_builder->setup(Size2i(ra.size, ra.size), max_cluster_elements, RID(), RID(), RID()); + } else { + ra.cluster_builder = nullptr; + } return reflection_atlas_owner.make_rid(ra); } @@ -3238,6 +541,11 @@ void RendererSceneRenderRD::reflection_atlas_set_size(RID p_ref_atlas, int p_ref return; //no changes } + if (ra->cluster_builder) { + // only if we're using our cluster + ra->cluster_builder->setup(Size2i(ra->size, ra->size), max_cluster_elements, RID(), RID(), RID()); + } + ra->size = p_reflection_size; ra->count = p_reflection_count; @@ -3247,9 +555,8 @@ void RendererSceneRenderRD::reflection_atlas_set_size(RID p_ref_atlas, int p_ref ra->reflection = RID(); RD::get_singleton()->free(ra->depth_buffer); ra->depth_buffer = RID(); - for (int i = 0; i < ra->reflections.size(); i++) { - _clear_reflection_data(ra->reflections.write[i].data); + ra->reflections.write[i].data.clear_reflection_data(); if (ra->reflections[i].owner.is_null()) { continue; } @@ -3353,7 +660,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc } if (atlas->reflection.is_null()) { - int mipmaps = MIN(roughness_layers, Image::get_image_required_mipmaps(atlas->size, atlas->size, Image::FORMAT_RGBAH) + 1); + int mipmaps = MIN(sky.roughness_layers, Image::get_image_required_mipmaps(atlas->size, atlas->size, Image::FORMAT_RGBAH) + 1); mipmaps = storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS ? 8 : mipmaps; // always use 8 mipmaps with real time filtering { //reflection atlas was unused, create: @@ -3378,7 +685,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc } atlas->reflections.resize(atlas->count); for (int i = 0; i < atlas->count; i++) { - _update_reflection_data(atlas->reflections.write[i].data, atlas->size, mipmaps, false, atlas->reflection, i * 6, storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS); + atlas->reflections.write[i].data.update_reflection_data(atlas->size, mipmaps, false, atlas->reflection, i * 6, storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS, sky.roughness_layers); for (int j = 0; j < 6; j++) { Vector<RID> fb; fb.push_back(atlas->reflections.write[i].data.layers[0].mipmaps[0].views[j]); @@ -3438,7 +745,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_ins if (storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS) { // Using real time reflections, all roughness is done in one step - _create_reflection_fast_filter(atlas->reflections.write[rpi->atlas_index].data, false); + atlas->reflections.write[rpi->atlas_index].data.create_reflection_fast_filter(storage, false); rpi->rendering = false; rpi->processing_side = 0; rpi->processing_layer = 1; @@ -3446,7 +753,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_ins } if (rpi->processing_layer > 1) { - _create_reflection_importance_sample(atlas->reflections.write[rpi->atlas_index].data, false, 10, rpi->processing_layer); + atlas->reflections.write[rpi->atlas_index].data.create_reflection_importance_sample(storage, false, 10, rpi->processing_layer, sky.sky_ggx_samples_quality); rpi->processing_layer++; if (rpi->processing_layer == atlas->reflections[rpi->atlas_index].data.layers[0].mipmaps.size()) { rpi->rendering = false; @@ -3457,7 +764,7 @@ bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_ins return false; } else { - _create_reflection_importance_sample(atlas->reflections.write[rpi->atlas_index].data, false, rpi->processing_side, rpi->processing_layer); + atlas->reflections.write[rpi->atlas_index].data.create_reflection_importance_sample(storage, false, rpi->processing_side, rpi->processing_layer, sky.sky_ggx_samples_quality); } rpi->processing_side++; @@ -3504,13 +811,28 @@ RID RendererSceneRenderRD::shadow_atlas_create() { return shadow_atlas_owner.make_rid(ShadowAtlas()); } -void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size) { +void RendererSceneRenderRD::_update_shadow_atlas(ShadowAtlas *shadow_atlas) { + if (shadow_atlas->size > 0 && shadow_atlas->depth.is_null()) { + RD::TextureFormat tf; + tf.format = shadow_atlas->use_16_bits ? RD::DATA_FORMAT_D16_UNORM : RD::DATA_FORMAT_D32_SFLOAT; + tf.width = shadow_atlas->size; + tf.height = shadow_atlas->size; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT; + + shadow_atlas->depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); + Vector<RID> fb_tex; + fb_tex.push_back(shadow_atlas->depth); + shadow_atlas->fb = RD::get_singleton()->framebuffer_create(fb_tex); + } +} + +void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size, bool p_16_bits) { ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_atlas); ERR_FAIL_COND(!shadow_atlas); ERR_FAIL_COND(p_size < 0); p_size = next_power_of_2(p_size); - if (p_size == shadow_atlas->size) { + if (p_size == shadow_atlas->size && p_16_bits == shadow_atlas->use_16_bits) { return; } @@ -3518,7 +840,6 @@ void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size) { if (shadow_atlas->depth.is_valid()) { RD::get_singleton()->free(shadow_atlas->depth); shadow_atlas->depth = RID(); - _clear_shadow_shrink_stages(shadow_atlas->shrink_stages); } for (int i = 0; i < 4; i++) { //clear subdivisions @@ -3537,16 +858,7 @@ void RendererSceneRenderRD::shadow_atlas_set_size(RID p_atlas, int p_size) { shadow_atlas->shadow_owners.clear(); shadow_atlas->size = p_size; - - if (shadow_atlas->size) { - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R32_SFLOAT; - tf.width = shadow_atlas->size; - tf.height = shadow_atlas->size; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; - - shadow_atlas->depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); - } + shadow_atlas->use_16_bits = p_size; } void RendererSceneRenderRD::shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision) { @@ -3801,10 +1113,24 @@ bool RendererSceneRenderRD::shadow_atlas_update_light(RID p_atlas, RID p_light_i return false; } -void RendererSceneRenderRD::directional_shadow_atlas_set_size(int p_size) { +void RendererSceneRenderRD::_update_directional_shadow_atlas() { + if (directional_shadow.depth.is_null() && directional_shadow.size > 0) { + RD::TextureFormat tf; + tf.format = directional_shadow.use_16_bits ? RD::DATA_FORMAT_D16_UNORM : RD::DATA_FORMAT_D32_SFLOAT; + tf.width = directional_shadow.size; + tf.height = directional_shadow.size; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT; + + directional_shadow.depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); + Vector<RID> fb_tex; + fb_tex.push_back(directional_shadow.depth); + directional_shadow.fb = RD::get_singleton()->framebuffer_create(fb_tex); + } +} +void RendererSceneRenderRD::directional_shadow_atlas_set_size(int p_size, bool p_16_bits) { p_size = nearest_power_of_2_templated(p_size); - if (directional_shadow.size == p_size) { + if (directional_shadow.size == p_size && directional_shadow.use_16_bits == p_16_bits) { return; } @@ -3812,21 +1138,9 @@ void RendererSceneRenderRD::directional_shadow_atlas_set_size(int p_size) { if (directional_shadow.depth.is_valid()) { RD::get_singleton()->free(directional_shadow.depth); - _clear_shadow_shrink_stages(directional_shadow.shrink_stages); directional_shadow.depth = RID(); + _base_uniforms_changed(); } - - if (p_size > 0) { - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R32_SFLOAT; - tf.width = p_size; - tf.height = p_size; - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; - - directional_shadow.depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); - } - - _base_uniforms_changed(); } void RendererSceneRenderRD::set_directional_shadow_count(int p_count) { @@ -3880,8 +1194,11 @@ int RendererSceneRenderRD::get_directional_light_shadow_size(RID p_light_intance ////////////////////////////////////////////////// -RID RendererSceneRenderRD::camera_effects_create() { - return camera_effects_owner.make_rid(CameraEffects()); +RID RendererSceneRenderRD::camera_effects_allocate() { + return camera_effects_owner.allocate_rid(); +} +void RendererSceneRenderRD::camera_effects_initialize(RID p_rid) { + camera_effects_owner.initialize_rid(p_rid, CameraEffects()); } void RendererSceneRenderRD::camera_effects_set_dof_blur_quality(RS::DOFBlurQuality p_quality, bool p_use_jitter) { @@ -3946,11 +1263,7 @@ void RendererSceneRenderRD::light_instance_set_shadow_transform(RID p_light_inst LightInstance *light_instance = light_instance_owner.getornull(p_light_instance); ERR_FAIL_COND(!light_instance); - if (storage->light_get_type(light_instance->light) != RS::LIGHT_DIRECTIONAL) { - p_pass = 0; - } - - ERR_FAIL_INDEX(p_pass, 4); + ERR_FAIL_INDEX(p_pass, 6); light_instance->shadow_transform[p_pass].camera = p_projection; light_instance->shadow_transform[p_pass].transform = p_transform; @@ -3996,29 +1309,6 @@ RendererSceneRenderRD::ShadowCubemap *RendererSceneRenderRD::_get_shadow_cubemap return &shadow_cubemaps[p_size]; } -RendererSceneRenderRD::ShadowMap *RendererSceneRenderRD::_get_shadow_map(const Size2i &p_size) { - if (!shadow_maps.has(p_size)) { - ShadowMap sm; - { - RD::TextureFormat tf; - tf.format = RD::get_singleton()->texture_is_format_supported_for_usage(RD::DATA_FORMAT_D32_SFLOAT, RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT) ? RD::DATA_FORMAT_D32_SFLOAT : RD::DATA_FORMAT_X8_D24_UNORM_PACK32; - tf.width = p_size.width; - tf.height = p_size.height; - tf.usage_bits = RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT; - - sm.depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); - } - - Vector<RID> fbtex; - fbtex.push_back(sm.depth); - sm.fb = RD::get_singleton()->framebuffer_create(fbtex); - - shadow_maps[p_size] = sm; - } - - return &shadow_maps[p_size]; -} - ////////////////////////// RID RendererSceneRenderRD::decal_instance_create(RID p_decal) { @@ -4049,805 +1339,27 @@ void RendererSceneRenderRD::lightmap_instance_set_transform(RID p_lightmap, cons ///////////////////////////////// RID RendererSceneRenderRD::gi_probe_instance_create(RID p_base) { - GIProbeInstance gi_probe; - gi_probe.probe = p_base; - RID rid = gi_probe_instance_owner.make_rid(gi_probe); - return rid; + return gi.gi_probe_instance_create(p_base); } void RendererSceneRenderRD::gi_probe_instance_set_transform_to_data(RID p_probe, const Transform &p_xform) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - ERR_FAIL_COND(!gi_probe); - - gi_probe->transform = p_xform; + gi.gi_probe_instance_set_transform_to_data(p_probe, p_xform); } bool RendererSceneRenderRD::gi_probe_needs_update(RID p_probe) const { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - ERR_FAIL_COND_V(!gi_probe, false); - if (low_end) { return false; } - //return true; - return gi_probe->last_probe_version != storage->gi_probe_get_version(gi_probe->probe); + return gi.gi_probe_needs_update(p_probe); } void RendererSceneRenderRD::gi_probe_update(RID p_probe, bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<GeometryInstance *> &p_dynamic_objects) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - ERR_FAIL_COND(!gi_probe); - if (low_end) { return; } - uint32_t data_version = storage->gi_probe_get_data_version(gi_probe->probe); - - // (RE)CREATE IF NEEDED - - if (gi_probe->last_probe_data_version != data_version) { - //need to re-create everything - if (gi_probe->texture.is_valid()) { - RD::get_singleton()->free(gi_probe->texture); - RD::get_singleton()->free(gi_probe->write_buffer); - gi_probe->mipmaps.clear(); - } - - for (int i = 0; i < gi_probe->dynamic_maps.size(); i++) { - RD::get_singleton()->free(gi_probe->dynamic_maps[i].texture); - RD::get_singleton()->free(gi_probe->dynamic_maps[i].depth); - } - - gi_probe->dynamic_maps.clear(); - - Vector3i octree_size = storage->gi_probe_get_octree_size(gi_probe->probe); - - if (octree_size != Vector3i()) { - //can create a 3D texture - Vector<int> levels = storage->gi_probe_get_level_counts(gi_probe->probe); - - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; - tf.width = octree_size.x; - tf.height = octree_size.y; - tf.depth = octree_size.z; - tf.texture_type = RD::TEXTURE_TYPE_3D; - tf.mipmaps = levels.size(); - - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; - - gi_probe->texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); - - RD::get_singleton()->texture_clear(gi_probe->texture, Color(0, 0, 0, 0), 0, levels.size(), 0, 1, false); - - { - int total_elements = 0; - for (int i = 0; i < levels.size(); i++) { - total_elements += levels[i]; - } - - gi_probe->write_buffer = RD::get_singleton()->storage_buffer_create(total_elements * 16); - } - - for (int i = 0; i < levels.size(); i++) { - GIProbeInstance::Mipmap mipmap; - mipmap.texture = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), gi_probe->texture, 0, i, RD::TEXTURE_SLICE_3D); - mipmap.level = levels.size() - i - 1; - mipmap.cell_offset = 0; - for (uint32_t j = 0; j < mipmap.level; j++) { - mipmap.cell_offset += levels[j]; - } - mipmap.cell_count = levels[mipmap.level]; - - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 1; - u.ids.push_back(storage->gi_probe_get_octree_buffer(gi_probe->probe)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 2; - u.ids.push_back(storage->gi_probe_get_data_buffer(gi_probe->probe)); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 4; - u.ids.push_back(gi_probe->write_buffer); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 9; - u.ids.push_back(storage->gi_probe_get_sdf_texture(gi_probe->probe)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 10; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - { - Vector<RD::Uniform> copy_uniforms = uniforms; - if (i == 0) { - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 3; - u.ids.push_back(gi_probe_lights_uniform); - copy_uniforms.push_back(u); - } - - mipmap.uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT], 0); - - copy_uniforms = uniforms; //restore - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 5; - u.ids.push_back(gi_probe->texture); - copy_uniforms.push_back(u); - } - mipmap.second_bounce_uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE], 0); - } else { - mipmap.uniform_set = RD::get_singleton()->uniform_set_create(copy_uniforms, giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP], 0); - } - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 5; - u.ids.push_back(mipmap.texture); - uniforms.push_back(u); - } - - mipmap.write_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_WRITE_TEXTURE], 0); - - gi_probe->mipmaps.push_back(mipmap); - } - - { - uint32_t dynamic_map_size = MAX(MAX(octree_size.x, octree_size.y), octree_size.z); - uint32_t oversample = nearest_power_of_2_templated(4); - int mipmap_index = 0; - - while (mipmap_index < gi_probe->mipmaps.size()) { - GIProbeInstance::DynamicMap dmap; - - if (oversample > 0) { - dmap.size = dynamic_map_size * (1 << oversample); - dmap.mipmap = -1; - oversample--; - } else { - dmap.size = dynamic_map_size >> mipmap_index; - dmap.mipmap = mipmap_index; - mipmap_index++; - } - - RD::TextureFormat dtf; - dtf.width = dmap.size; - dtf.height = dmap.size; - dtf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; - dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; - - if (gi_probe->dynamic_maps.size() == 0) { - dtf.usage_bits |= RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - } - dmap.texture = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - - if (gi_probe->dynamic_maps.size() == 0) { - //render depth for first one - dtf.format = RD::get_singleton()->texture_is_format_supported_for_usage(RD::DATA_FORMAT_D32_SFLOAT, RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT) ? RD::DATA_FORMAT_D32_SFLOAT : RD::DATA_FORMAT_X8_D24_UNORM_PACK32; - dtf.usage_bits = RD::TEXTURE_USAGE_DEPTH_STENCIL_ATTACHMENT_BIT; - dmap.fb_depth = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - } - - //just use depth as-is - dtf.format = RD::DATA_FORMAT_R32_SFLOAT; - dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - - dmap.depth = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - - if (gi_probe->dynamic_maps.size() == 0) { - dtf.format = RD::DATA_FORMAT_R8G8B8A8_UNORM; - dtf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; - dmap.albedo = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - dmap.normal = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - dmap.orm = RD::get_singleton()->texture_create(dtf, RD::TextureView()); - - Vector<RID> fb; - fb.push_back(dmap.albedo); - fb.push_back(dmap.normal); - fb.push_back(dmap.orm); - fb.push_back(dmap.texture); //emission - fb.push_back(dmap.depth); - fb.push_back(dmap.fb_depth); - - dmap.fb = RD::get_singleton()->framebuffer_create(fb); - - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.binding = 3; - u.ids.push_back(gi_probe_lights_uniform); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 5; - u.ids.push_back(dmap.albedo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 6; - u.ids.push_back(dmap.normal); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 7; - u.ids.push_back(dmap.orm); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 8; - u.ids.push_back(dmap.fb_depth); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 9; - u.ids.push_back(storage->gi_probe_get_sdf_texture(gi_probe->probe)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 10; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 11; - u.ids.push_back(dmap.texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 12; - u.ids.push_back(dmap.depth); - uniforms.push_back(u); - } - - dmap.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING], 0); - } - } else { - bool plot = dmap.mipmap >= 0; - bool write = dmap.mipmap < (gi_probe->mipmaps.size() - 1); - - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 5; - u.ids.push_back(gi_probe->dynamic_maps[gi_probe->dynamic_maps.size() - 1].texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 6; - u.ids.push_back(gi_probe->dynamic_maps[gi_probe->dynamic_maps.size() - 1].depth); - uniforms.push_back(u); - } - - if (write) { - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 7; - u.ids.push_back(dmap.texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 8; - u.ids.push_back(dmap.depth); - uniforms.push_back(u); - } - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 9; - u.ids.push_back(storage->gi_probe_get_sdf_texture(gi_probe->probe)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 10; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - if (plot) { - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.binding = 11; - u.ids.push_back(gi_probe->mipmaps[dmap.mipmap].texture); - uniforms.push_back(u); - } - } - - dmap.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, giprobe_lighting_shader_version_shaders[(write && plot) ? GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT : write ? GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE : GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT], 0); - } - - gi_probe->dynamic_maps.push_back(dmap); - } - } - } - - gi_probe->last_probe_data_version = data_version; - p_update_light_instances = true; //just in case - - _base_uniforms_changed(); - } - - // UDPDATE TIME - - if (gi_probe->has_dynamic_object_data) { - //if it has dynamic object data, it needs to be cleared - RD::get_singleton()->texture_clear(gi_probe->texture, Color(0, 0, 0, 0), 0, gi_probe->mipmaps.size(), 0, 1, true); - } - - uint32_t light_count = 0; - - if (p_update_light_instances || p_dynamic_objects.size() > 0) { - light_count = MIN(gi_probe_max_lights, (uint32_t)p_light_instances.size()); - - { - Transform to_cell = storage->gi_probe_get_to_cell_xform(gi_probe->probe); - Transform to_probe_xform = (gi_probe->transform * to_cell.affine_inverse()).affine_inverse(); - //update lights - - for (uint32_t i = 0; i < light_count; i++) { - GIProbeLight &l = gi_probe_lights[i]; - RID light_instance = p_light_instances[i]; - RID light = light_instance_get_base_light(light_instance); - - l.type = storage->light_get_type(light); - if (l.type == RS::LIGHT_DIRECTIONAL && storage->light_directional_is_sky_only(light)) { - light_count--; - continue; - } - - l.attenuation = storage->light_get_param(light, RS::LIGHT_PARAM_ATTENUATION); - l.energy = storage->light_get_param(light, RS::LIGHT_PARAM_ENERGY) * storage->light_get_param(light, RS::LIGHT_PARAM_INDIRECT_ENERGY); - l.radius = to_cell.basis.xform(Vector3(storage->light_get_param(light, RS::LIGHT_PARAM_RANGE), 0, 0)).length(); - Color color = storage->light_get_color(light).to_linear(); - l.color[0] = color.r; - l.color[1] = color.g; - l.color[2] = color.b; - - l.spot_angle_radians = Math::deg2rad(storage->light_get_param(light, RS::LIGHT_PARAM_SPOT_ANGLE)); - l.spot_attenuation = storage->light_get_param(light, RS::LIGHT_PARAM_SPOT_ATTENUATION); - - Transform xform = light_instance_get_base_transform(light_instance); - - Vector3 pos = to_probe_xform.xform(xform.origin); - Vector3 dir = to_probe_xform.basis.xform(-xform.basis.get_axis(2)).normalized(); - - l.position[0] = pos.x; - l.position[1] = pos.y; - l.position[2] = pos.z; - - l.direction[0] = dir.x; - l.direction[1] = dir.y; - l.direction[2] = dir.z; - - l.has_shadow = storage->light_has_shadow(light); - } - - RD::get_singleton()->buffer_update(gi_probe_lights_uniform, 0, sizeof(GIProbeLight) * light_count, gi_probe_lights, true); - } - } - - if (gi_probe->has_dynamic_object_data || p_update_light_instances || p_dynamic_objects.size()) { - // PROCESS MIPMAPS - if (gi_probe->mipmaps.size()) { - //can update mipmaps - - Vector3i probe_size = storage->gi_probe_get_octree_size(gi_probe->probe); - - GIProbePushConstant push_constant; - - push_constant.limits[0] = probe_size.x; - push_constant.limits[1] = probe_size.y; - push_constant.limits[2] = probe_size.z; - push_constant.stack_size = gi_probe->mipmaps.size(); - push_constant.emission_scale = 1.0; - push_constant.propagation = storage->gi_probe_get_propagation(gi_probe->probe); - push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(gi_probe->probe); - push_constant.light_count = light_count; - push_constant.aniso_strength = 0; - - /* print_line("probe update to version " + itos(gi_probe->last_probe_version)); - print_line("propagation " + rtos(push_constant.propagation)); - print_line("dynrange " + rtos(push_constant.dynamic_range)); - */ - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - - int passes; - if (p_update_light_instances) { - passes = storage->gi_probe_is_using_two_bounces(gi_probe->probe) ? 2 : 1; - } else { - passes = 1; //only re-blitting is necessary - } - int wg_size = 64; - int wg_limit_x = RD::get_singleton()->limit_get(RD::LIMIT_MAX_COMPUTE_WORKGROUP_COUNT_X); - - for (int pass = 0; pass < passes; pass++) { - if (p_update_light_instances) { - for (int i = 0; i < gi_probe->mipmaps.size(); i++) { - if (i == 0) { - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[pass == 0 ? GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT : GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE]); - } else if (i == 1) { - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP]); - } - - if (pass == 1 || i > 0) { - RD::get_singleton()->compute_list_add_barrier(compute_list); //wait til previous step is done - } - if (pass == 0 || i > 0) { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi_probe->mipmaps[i].uniform_set, 0); - } else { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi_probe->mipmaps[i].second_bounce_uniform_set, 0); - } - - push_constant.cell_offset = gi_probe->mipmaps[i].cell_offset; - push_constant.cell_count = gi_probe->mipmaps[i].cell_count; - - int wg_todo = (gi_probe->mipmaps[i].cell_count - 1) / wg_size + 1; - while (wg_todo) { - int wg_count = MIN(wg_todo, wg_limit_x); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbePushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, wg_count, 1, 1); - wg_todo -= wg_count; - push_constant.cell_offset += wg_count * wg_size; - } - } - - RD::get_singleton()->compute_list_add_barrier(compute_list); //wait til previous step is done - } - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_WRITE_TEXTURE]); - - for (int i = 0; i < gi_probe->mipmaps.size(); i++) { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi_probe->mipmaps[i].write_uniform_set, 0); - - push_constant.cell_offset = gi_probe->mipmaps[i].cell_offset; - push_constant.cell_count = gi_probe->mipmaps[i].cell_count; - - int wg_todo = (gi_probe->mipmaps[i].cell_count - 1) / wg_size + 1; - while (wg_todo) { - int wg_count = MIN(wg_todo, wg_limit_x); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbePushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, wg_count, 1, 1); - wg_todo -= wg_count; - push_constant.cell_offset += wg_count * wg_size; - } - } - } - - RD::get_singleton()->compute_list_end(); - } - } - - gi_probe->has_dynamic_object_data = false; //clear until dynamic object data is used again - - if (p_dynamic_objects.size() && gi_probe->dynamic_maps.size()) { - Vector3i octree_size = storage->gi_probe_get_octree_size(gi_probe->probe); - int multiplier = gi_probe->dynamic_maps[0].size / MAX(MAX(octree_size.x, octree_size.y), octree_size.z); - - Transform oversample_scale; - oversample_scale.basis.scale(Vector3(multiplier, multiplier, multiplier)); - - Transform to_cell = oversample_scale * storage->gi_probe_get_to_cell_xform(gi_probe->probe); - Transform to_world_xform = gi_probe->transform * to_cell.affine_inverse(); - Transform to_probe_xform = to_world_xform.affine_inverse(); - - AABB probe_aabb(Vector3(), octree_size); - - //this could probably be better parallelized in compute.. - for (int i = 0; i < (int)p_dynamic_objects.size(); i++) { - GeometryInstance *instance = p_dynamic_objects[i]; - - //transform aabb to giprobe - AABB aabb = (to_probe_xform * geometry_instance_get_transform(instance)).xform(geometry_instance_get_aabb(instance)); - - //this needs to wrap to grid resolution to avoid jitter - //also extend margin a bit just in case - Vector3i begin = aabb.position - Vector3i(1, 1, 1); - Vector3i end = aabb.position + aabb.size + Vector3i(1, 1, 1); - - for (int j = 0; j < 3; j++) { - if ((end[j] - begin[j]) & 1) { - end[j]++; //for half extents split, it needs to be even - } - begin[j] = MAX(begin[j], 0); - end[j] = MIN(end[j], octree_size[j] * multiplier); - } - - //aabb = aabb.intersection(probe_aabb); //intersect - aabb.position = begin; - aabb.size = end - begin; - - //print_line("aabb: " + aabb); - - for (int j = 0; j < 6; j++) { - //if (j != 0 && j != 3) { - // continue; - //} - static const Vector3 render_z[6] = { - Vector3(1, 0, 0), - Vector3(0, 1, 0), - Vector3(0, 0, 1), - Vector3(-1, 0, 0), - Vector3(0, -1, 0), - Vector3(0, 0, -1), - }; - static const Vector3 render_up[6] = { - Vector3(0, 1, 0), - Vector3(0, 0, 1), - Vector3(0, 1, 0), - Vector3(0, 1, 0), - Vector3(0, 0, 1), - Vector3(0, 1, 0), - }; - - Vector3 render_dir = render_z[j]; - Vector3 up_dir = render_up[j]; - - Vector3 center = aabb.position + aabb.size * 0.5; - Transform xform; - xform.set_look_at(center - aabb.size * 0.5 * render_dir, center, up_dir); - - Vector3 x_dir = xform.basis.get_axis(0).abs(); - int x_axis = int(Vector3(0, 1, 2).dot(x_dir)); - Vector3 y_dir = xform.basis.get_axis(1).abs(); - int y_axis = int(Vector3(0, 1, 2).dot(y_dir)); - Vector3 z_dir = -xform.basis.get_axis(2); - int z_axis = int(Vector3(0, 1, 2).dot(z_dir.abs())); - - Rect2i rect(aabb.position[x_axis], aabb.position[y_axis], aabb.size[x_axis], aabb.size[y_axis]); - bool x_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(0)) < 0); - bool y_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(1)) < 0); - bool z_flip = bool(Vector3(1, 1, 1).dot(xform.basis.get_axis(2)) > 0); - - CameraMatrix cm; - cm.set_orthogonal(-rect.size.width / 2, rect.size.width / 2, -rect.size.height / 2, rect.size.height / 2, 0.0001, aabb.size[z_axis]); - - if (cull_argument.size() == 0) { - cull_argument.push_back(nullptr); - } - cull_argument[0] = instance; - - _render_material(to_world_xform * xform, cm, true, cull_argument, gi_probe->dynamic_maps[0].fb, Rect2i(Vector2i(), rect.size)); - - GIProbeDynamicPushConstant push_constant; - zeromem(&push_constant, sizeof(GIProbeDynamicPushConstant)); - push_constant.limits[0] = octree_size.x; - push_constant.limits[1] = octree_size.y; - push_constant.limits[2] = octree_size.z; - push_constant.light_count = p_light_instances.size(); - push_constant.x_dir[0] = x_dir[0]; - push_constant.x_dir[1] = x_dir[1]; - push_constant.x_dir[2] = x_dir[2]; - push_constant.y_dir[0] = y_dir[0]; - push_constant.y_dir[1] = y_dir[1]; - push_constant.y_dir[2] = y_dir[2]; - push_constant.z_dir[0] = z_dir[0]; - push_constant.z_dir[1] = z_dir[1]; - push_constant.z_dir[2] = z_dir[2]; - push_constant.z_base = xform.origin[z_axis]; - push_constant.z_sign = (z_flip ? -1.0 : 1.0); - push_constant.pos_multiplier = float(1.0) / multiplier; - push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(gi_probe->probe); - push_constant.flip_x = x_flip; - push_constant.flip_y = y_flip; - push_constant.rect_pos[0] = rect.position[0]; - push_constant.rect_pos[1] = rect.position[1]; - push_constant.rect_size[0] = rect.size[0]; - push_constant.rect_size[1] = rect.size[1]; - push_constant.prev_rect_ofs[0] = 0; - push_constant.prev_rect_ofs[1] = 0; - push_constant.prev_rect_size[0] = 0; - push_constant.prev_rect_size[1] = 0; - push_constant.on_mipmap = false; - push_constant.propagation = storage->gi_probe_get_propagation(gi_probe->probe); - push_constant.pad[0] = 0; - push_constant.pad[1] = 0; - push_constant.pad[2] = 0; - - //process lighting - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi_probe->dynamic_maps[0].uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbeDynamicPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, (rect.size.x - 1) / 8 + 1, (rect.size.y - 1) / 8 + 1, 1); - //print_line("rect: " + itos(i) + ": " + rect); - - for (int k = 1; k < gi_probe->dynamic_maps.size(); k++) { - // enlarge the rect if needed so all pixels fit when downscaled, - // this ensures downsampling is smooth and optimal because no pixels are left behind - - //x - if (rect.position.x & 1) { - rect.size.x++; - push_constant.prev_rect_ofs[0] = 1; //this is used to ensure reading is also optimal - } else { - push_constant.prev_rect_ofs[0] = 0; - } - if (rect.size.x & 1) { - rect.size.x++; - } - - rect.position.x >>= 1; - rect.size.x = MAX(1, rect.size.x >> 1); - - //y - if (rect.position.y & 1) { - rect.size.y++; - push_constant.prev_rect_ofs[1] = 1; - } else { - push_constant.prev_rect_ofs[1] = 0; - } - if (rect.size.y & 1) { - rect.size.y++; - } - - rect.position.y >>= 1; - rect.size.y = MAX(1, rect.size.y >> 1); - - //shrink limits to ensure plot does not go outside map - if (gi_probe->dynamic_maps[k].mipmap > 0) { - for (int l = 0; l < 3; l++) { - push_constant.limits[l] = MAX(1, push_constant.limits[l] >> 1); - } - } - - //print_line("rect: " + itos(i) + ": " + rect); - push_constant.rect_pos[0] = rect.position[0]; - push_constant.rect_pos[1] = rect.position[1]; - push_constant.prev_rect_size[0] = push_constant.rect_size[0]; - push_constant.prev_rect_size[1] = push_constant.rect_size[1]; - push_constant.rect_size[0] = rect.size[0]; - push_constant.rect_size[1] = rect.size[1]; - push_constant.on_mipmap = gi_probe->dynamic_maps[k].mipmap > 0; - - RD::get_singleton()->compute_list_add_barrier(compute_list); - - if (gi_probe->dynamic_maps[k].mipmap < 0) { - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE]); - } else if (k < gi_probe->dynamic_maps.size() - 1) { - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT]); - } else { - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT]); - } - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, gi_probe->dynamic_maps[k].uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(GIProbeDynamicPushConstant)); - RD::get_singleton()->compute_list_dispatch(compute_list, (rect.size.x - 1) / 8 + 1, (rect.size.y - 1) / 8 + 1, 1); - } - - RD::get_singleton()->compute_list_end(); - } - } - - gi_probe->has_dynamic_object_data = true; //clear until dynamic object data is used again - } - - gi_probe->last_probe_version = storage->gi_probe_get_version(gi_probe->probe); -} - -void RendererSceneRenderRD::_debug_giprobe(RID p_gi_probe, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_gi_probe); - ERR_FAIL_COND(!gi_probe); - - if (gi_probe->mipmaps.size() == 0) { - return; - } - - CameraMatrix transform = (p_camera_with_transform * CameraMatrix(gi_probe->transform)) * CameraMatrix(storage->gi_probe_get_to_cell_xform(gi_probe->probe).affine_inverse()); - - int level = 0; - Vector3i octree_size = storage->gi_probe_get_octree_size(gi_probe->probe); - - GIProbeDebugPushConstant push_constant; - push_constant.alpha = p_alpha; - push_constant.dynamic_range = storage->gi_probe_get_dynamic_range(gi_probe->probe); - push_constant.cell_offset = gi_probe->mipmaps[level].cell_offset; - push_constant.level = level; - - push_constant.bounds[0] = octree_size.x >> level; - push_constant.bounds[1] = octree_size.y >> level; - push_constant.bounds[2] = octree_size.z >> level; - push_constant.pad = 0; - - for (int i = 0; i < 4; i++) { - for (int j = 0; j < 4; j++) { - push_constant.projection[i * 4 + j] = transform.matrix[i][j]; - } - } - - if (giprobe_debug_uniform_set.is_valid()) { - RD::get_singleton()->free(giprobe_debug_uniform_set); - } - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 1; - u.ids.push_back(storage->gi_probe_get_data_buffer(gi_probe->probe)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 2; - u.ids.push_back(gi_probe->texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 3; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - int cell_count; - if (!p_emission && p_lighting && gi_probe->has_dynamic_object_data) { - cell_count = push_constant.bounds[0] * push_constant.bounds[1] * push_constant.bounds[2]; - } else { - cell_count = gi_probe->mipmaps[level].cell_count; - } - - giprobe_debug_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, giprobe_debug_shader_version_shaders[0], 0); - RD::get_singleton()->draw_list_bind_render_pipeline(p_draw_list, giprobe_debug_shader_version_pipelines[p_emission ? GI_PROBE_DEBUG_EMISSION : p_lighting ? (gi_probe->has_dynamic_object_data ? GI_PROBE_DEBUG_LIGHT_FULL : GI_PROBE_DEBUG_LIGHT) : GI_PROBE_DEBUG_COLOR].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); - RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, giprobe_debug_uniform_set, 0); - RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(GIProbeDebugPushConstant)); - RD::get_singleton()->draw_list_draw(p_draw_list, false, cell_count, 36); + gi.gi_probe_update(p_probe, p_update_light_instances, p_light_instances, p_dynamic_objects, this); } void RendererSceneRenderRD::_debug_sdfgi_probes(RID p_render_buffers, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform) { @@ -4858,132 +1370,7 @@ void RendererSceneRenderRD::_debug_sdfgi_probes(RID p_render_buffers, RD::DrawLi return; //nothing to debug } - SDGIShader::DebugProbesPushConstant push_constant; - - for (int i = 0; i < 4; i++) { - for (int j = 0; j < 4; j++) { - push_constant.projection[i * 4 + j] = p_camera_with_transform.matrix[i][j]; - } - } - - //gen spheres from strips - uint32_t band_points = 16; - push_constant.band_power = 4; - push_constant.sections_in_band = ((band_points / 2) - 1); - push_constant.band_mask = band_points - 2; - push_constant.section_arc = (Math_PI * 2.0) / float(push_constant.sections_in_band); - push_constant.y_mult = rb->sdfgi->y_mult; - - uint32_t total_points = push_constant.sections_in_band * band_points; - uint32_t total_probes = rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count; - - push_constant.grid_size[0] = rb->sdfgi->cascade_size; - push_constant.grid_size[1] = rb->sdfgi->cascade_size; - push_constant.grid_size[2] = rb->sdfgi->cascade_size; - push_constant.cascade = 0; - - push_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - - if (!rb->sdfgi->debug_probes_uniform_set.is_valid() || !RD::get_singleton()->uniform_set_is_valid(rb->sdfgi->debug_probes_uniform_set)) { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.binding = 1; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(rb->sdfgi->cascades_ubo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(rb->sdfgi->lightprobe_texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 4; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(rb->sdfgi->occlusion_texture); - uniforms.push_back(u); - } - - rb->sdfgi->debug_probes_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.debug_probes.version_get_shader(sdfgi_shader.debug_probes_shader, 0), 0); - } - - RD::get_singleton()->draw_list_bind_render_pipeline(p_draw_list, sdfgi_shader.debug_probes_pipeline[SDGIShader::PROBE_DEBUG_PROBES].get_render_pipeline(RD::INVALID_FORMAT_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); - RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, rb->sdfgi->debug_probes_uniform_set, 0); - RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(SDGIShader::DebugProbesPushConstant)); - RD::get_singleton()->draw_list_draw(p_draw_list, false, total_probes, total_points); - - if (sdfgi_debug_probe_dir != Vector3()) { - print_line("CLICK DEBUG ME?"); - uint32_t cascade = 0; - Vector3 offset = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + rb->sdfgi->cascades[cascade].position)) * rb->sdfgi->cascades[cascade].cell_size * Vector3(1.0, 1.0 / rb->sdfgi->y_mult, 1.0); - Vector3 probe_size = rb->sdfgi->cascades[cascade].cell_size * (rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR) * Vector3(1.0, 1.0 / rb->sdfgi->y_mult, 1.0); - Vector3 ray_from = sdfgi_debug_probe_pos; - Vector3 ray_to = sdfgi_debug_probe_pos + sdfgi_debug_probe_dir * rb->sdfgi->cascades[cascade].cell_size * Math::sqrt(3.0) * rb->sdfgi->cascade_size; - float sphere_radius = 0.2; - float closest_dist = 1e20; - sdfgi_debug_probe_enabled = false; - - Vector3i probe_from = rb->sdfgi->cascades[cascade].position / (rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR); - for (int i = 0; i < (SDFGI::PROBE_DIVISOR + 1); i++) { - for (int j = 0; j < (SDFGI::PROBE_DIVISOR + 1); j++) { - for (int k = 0; k < (SDFGI::PROBE_DIVISOR + 1); k++) { - Vector3 pos = offset + probe_size * Vector3(i, j, k); - Vector3 res; - if (Geometry3D::segment_intersects_sphere(ray_from, ray_to, pos, sphere_radius, &res)) { - float d = ray_from.distance_to(res); - if (d < closest_dist) { - closest_dist = d; - sdfgi_debug_probe_enabled = true; - sdfgi_debug_probe_index = probe_from + Vector3i(i, j, k); - } - } - } - } - } - - if (sdfgi_debug_probe_enabled) { - print_line("found: " + sdfgi_debug_probe_index); - } else { - print_line("no found"); - } - sdfgi_debug_probe_dir = Vector3(); - } - - if (sdfgi_debug_probe_enabled) { - uint32_t cascade = 0; - uint32_t probe_cells = (rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR); - Vector3i probe_from = rb->sdfgi->cascades[cascade].position / probe_cells; - Vector3i ofs = sdfgi_debug_probe_index - probe_from; - if (ofs.x < 0 || ofs.y < 0 || ofs.z < 0) { - return; - } - if (ofs.x > SDFGI::PROBE_DIVISOR || ofs.y > SDFGI::PROBE_DIVISOR || ofs.z > SDFGI::PROBE_DIVISOR) { - return; - } - - uint32_t mult = (SDFGI::PROBE_DIVISOR + 1); - uint32_t index = ofs.z * mult * mult + ofs.y * mult + ofs.x; - - push_constant.probe_debug_index = index; - - uint32_t cell_count = probe_cells * 2 * probe_cells * 2 * probe_cells * 2; - - RD::get_singleton()->draw_list_bind_render_pipeline(p_draw_list, sdfgi_shader.debug_probes_pipeline[SDGIShader::PROBE_DEBUG_VISIBILITY].get_render_pipeline(RD::INVALID_FORMAT_ID, RD::get_singleton()->framebuffer_get_format(p_framebuffer))); - RD::get_singleton()->draw_list_bind_uniform_set(p_draw_list, rb->sdfgi->debug_probes_uniform_set, 0); - RD::get_singleton()->draw_list_set_push_constant(p_draw_list, &push_constant, sizeof(SDGIShader::DebugProbesPushConstant)); - RD::get_singleton()->draw_list_draw(p_draw_list, false, cell_count, total_points); - } + rb->sdfgi->debug_probes(p_draw_list, p_framebuffer, p_camera_with_transform); } //////////////////////////////// @@ -5092,9 +1479,6 @@ void RendererSceneRenderRD::_free_render_buffer_data(RenderBuffers *rb) { RD::get_singleton()->free(rb->luminance.reduce[i]); } - for (int i = 0; i < rb->luminance.reduce.size(); i++) { - RD::get_singleton()->free(rb->luminance.reduce[i]); - } rb->luminance.reduce.clear(); if (rb->luminance.current.is_valid()) { @@ -5135,6 +1519,13 @@ void RendererSceneRenderRD::_free_render_buffer_data(RenderBuffers *rb) { RD::get_singleton()->free(rb->ssr.normal_scaled); rb->ssr.normal_scaled = RID(); } + + if (rb->ambient_buffer.is_valid()) { + RD::get_singleton()->free(rb->ambient_buffer); + RD::get_singleton()->free(rb->reflection_buffer); + rb->ambient_buffer = RID(); + rb->reflection_buffer = RID(); + } } void RendererSceneRenderRD::_process_sss(RID p_render_buffers, const CameraMatrix &p_camera) { @@ -5168,7 +1559,7 @@ void RendererSceneRenderRD::_process_ssr(RID p_render_buffers, RID p_dest_frameb return; } - Environment *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); ERR_FAIL_COND(!env); ERR_FAIL_COND(!env->ssr_enabled); @@ -5213,7 +1604,7 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); ERR_FAIL_COND(!rb); - Environment *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); ERR_FAIL_COND(!env); RENDER_TIMESTAMP("Process SSAO"); @@ -5267,9 +1658,11 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen tf.array_layers = 4; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; rb->ssao.depth = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.depth, "SSAO Depth"); for (uint32_t i = 0; i < tf.mipmaps; i++) { RID slice = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), rb->ssao.depth, 0, i, RD::TEXTURE_SLICE_2D_ARRAY); rb->ssao.depth_slices.push_back(slice); + RD::get_singleton()->set_resource_name(rb->ssao.depth_slices[i], "SSAO Depth Mip " + itos(i) + " "); } } @@ -5282,9 +1675,11 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen tf.array_layers = 4; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; rb->ssao.ao_deinterleaved = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.ao_deinterleaved, "SSAO De-interleaved Array"); for (uint32_t i = 0; i < 4; i++) { RID slice = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), rb->ssao.ao_deinterleaved, i, 0); rb->ssao.ao_deinterleaved_slices.push_back(slice); + RD::get_singleton()->set_resource_name(rb->ssao.ao_deinterleaved_slices[i], "SSAO De-interleaved Array Layer " + itos(i) + " "); } } @@ -5297,9 +1692,11 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen tf.array_layers = 4; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; rb->ssao.ao_pong = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.ao_pong, "SSAO De-interleaved Array Pong"); for (uint32_t i = 0; i < 4; i++) { RID slice = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), rb->ssao.ao_pong, i, 0); rb->ssao.ao_pong_slices.push_back(slice); + RD::get_singleton()->set_resource_name(rb->ssao.ao_deinterleaved_slices[i], "SSAO De-interleaved Array Layer " + itos(i) + " Pong"); } } @@ -5310,7 +1707,9 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen tf.height = half_height; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; rb->ssao.importance_map[0] = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.importance_map[0], "SSAO Importance Map"); rb->ssao.importance_map[1] = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.importance_map[1], "SSAO Importance Map Pong"); } { RD::TextureFormat tf; @@ -5319,6 +1718,7 @@ void RendererSceneRenderRD::_process_ssao(RID p_render_buffers, RID p_environmen tf.height = rb->height; tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; rb->ssao.ao_final = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->set_resource_name(rb->ssao.ao_final, "SSAO Final"); _render_buffers_uniform_set_changed(p_render_buffers); } ssao_using_half_size = ssao_half_size; @@ -5350,7 +1750,7 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(RID p_rende RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); ERR_FAIL_COND(!rb); - Environment *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); //glow (if enabled) CameraEffects *camfx = camera_effects_owner.getornull(p_camera_effects); @@ -5534,16 +1934,16 @@ void RendererSceneRenderRD::_render_buffers_debug_draw(RID p_render_buffers, RID effects->copy_to_fb_rect(_render_buffers_get_normal_texture(p_render_buffers), storage->render_target_get_rd_framebuffer(rb->render_target), Rect2(Vector2(), rtsize), false, false); } - if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_GI_BUFFER && _render_buffers_get_ambient_texture(p_render_buffers).is_valid()) { + if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_GI_BUFFER && rb->ambient_buffer.is_valid()) { Size2 rtsize = storage->render_target_get_size(rb->render_target); - RID ambient_texture = _render_buffers_get_ambient_texture(p_render_buffers); - RID reflection_texture = _render_buffers_get_reflection_texture(p_render_buffers); + RID ambient_texture = rb->ambient_buffer; + RID reflection_texture = rb->reflection_buffer; effects->copy_to_fb_rect(ambient_texture, storage->render_target_get_rd_framebuffer(rb->render_target), Rect2(Vector2(), rtsize), false, false, false, true, reflection_texture); } } void RendererSceneRenderRD::environment_set_adjustment(RID p_env, bool p_enable, float p_brightness, float p_contrast, float p_saturation, bool p_use_1d_color_correction, RID p_color_correction) { - Environment *env = environment_owner.getornull(p_env); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_env); ERR_FAIL_COND(!env); env->adjustments_enabled = p_enable; @@ -5554,152 +1954,6 @@ void RendererSceneRenderRD::environment_set_adjustment(RID p_env, bool p_enable, env->color_correction = p_color_correction; } -void RendererSceneRenderRD::_sdfgi_debug_draw(RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(!rb); - - if (!rb->sdfgi) { - return; //eh - } - - if (!rb->sdfgi->debug_uniform_set.is_valid() || !RD::get_singleton()->uniform_set_is_valid(rb->sdfgi->debug_uniform_set)) { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.binding = 1; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { - if (i < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[i].sdf_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { - if (i < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[i].light_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { - if (i < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[i].light_aniso_0_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 4; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - for (uint32_t i = 0; i < SDFGI::MAX_CASCADES; i++) { - if (i < rb->sdfgi->cascades.size()) { - u.ids.push_back(rb->sdfgi->cascades[i].light_aniso_1_tex); - } else { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE)); - } - } - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 5; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(rb->sdfgi->occlusion_texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 8; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 9; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(rb->sdfgi->cascades_ubo); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 10; - u.uniform_type = RD::UNIFORM_TYPE_IMAGE; - u.ids.push_back(rb->texture); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.binding = 11; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.ids.push_back(rb->sdfgi->lightprobe_texture); - uniforms.push_back(u); - } - rb->sdfgi->debug_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.debug_shader_version, 0); - } - - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.debug_pipeline); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->debug_uniform_set, 0); - - SDGIShader::DebugPushConstant push_constant; - push_constant.grid_size[0] = rb->sdfgi->cascade_size; - push_constant.grid_size[1] = rb->sdfgi->cascade_size; - push_constant.grid_size[2] = rb->sdfgi->cascade_size; - push_constant.max_cascades = rb->sdfgi->cascades.size(); - push_constant.screen_size[0] = rb->width; - push_constant.screen_size[1] = rb->height; - push_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - push_constant.use_occlusion = rb->sdfgi->uses_occlusion; - push_constant.y_mult = rb->sdfgi->y_mult; - - Vector2 vp_half = p_projection.get_viewport_half_extents(); - push_constant.cam_extent[0] = vp_half.x; - push_constant.cam_extent[1] = vp_half.y; - push_constant.cam_extent[2] = -p_projection.get_z_near(); - - push_constant.cam_transform[0] = p_transform.basis.elements[0][0]; - push_constant.cam_transform[1] = p_transform.basis.elements[1][0]; - push_constant.cam_transform[2] = p_transform.basis.elements[2][0]; - push_constant.cam_transform[3] = 0; - push_constant.cam_transform[4] = p_transform.basis.elements[0][1]; - push_constant.cam_transform[5] = p_transform.basis.elements[1][1]; - push_constant.cam_transform[6] = p_transform.basis.elements[2][1]; - push_constant.cam_transform[7] = 0; - push_constant.cam_transform[8] = p_transform.basis.elements[0][2]; - push_constant.cam_transform[9] = p_transform.basis.elements[1][2]; - push_constant.cam_transform[10] = p_transform.basis.elements[2][2]; - push_constant.cam_transform[11] = 0; - push_constant.cam_transform[12] = p_transform.origin.x; - push_constant.cam_transform[13] = p_transform.origin.y; - push_constant.cam_transform[14] = p_transform.origin.z; - push_constant.cam_transform[15] = 1; - - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::DebugPushConstant)); - - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->width, rb->height, 1, 8, 8, 1); - RD::get_singleton()->compute_list_end(); - - Size2 rtsize = storage->render_target_get_size(rb->render_target); - storage->get_effects()->copy_to_fb_rect(rb->texture, storage->render_target_get_rd_framebuffer(rb->render_target), Rect2(Vector2(), rtsize), true); -} - RID RendererSceneRenderRD::render_buffers_get_back_buffer_texture(RID p_render_buffers) { RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); @@ -5719,14 +1973,25 @@ RID RendererSceneRenderRD::render_buffers_get_ao_texture(RID p_render_buffers) { RID RendererSceneRenderRD::render_buffers_get_gi_probe_buffer(RID p_render_buffers) { RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); ERR_FAIL_COND_V(!rb, RID()); - if (rb->giprobe_buffer.is_null()) { - rb->giprobe_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(GI::GIProbeData) * RenderBuffers::MAX_GIPROBES); + if (rb->gi.giprobe_buffer.is_null()) { + rb->gi.giprobe_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(RendererSceneGIRD::GIProbeData) * RendererSceneGIRD::MAX_GIPROBES); } - return rb->giprobe_buffer; + return rb->gi.giprobe_buffer; } RID RendererSceneRenderRD::render_buffers_get_default_gi_probe_buffer() { - return default_giprobe_buffer; + return gi.default_giprobe_buffer; +} + +RID RendererSceneRenderRD::render_buffers_get_gi_ambient_texture(RID p_render_buffers) { + RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND_V(!rb, RID()); + return rb->ambient_buffer; +} +RID RendererSceneRenderRD::render_buffers_get_gi_reflection_texture(RID p_render_buffers) { + RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND_V(!rb, RID()); + return rb->reflection_buffer; } uint32_t RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_count(RID p_render_buffers) const { @@ -5764,7 +2029,7 @@ Vector3i RendererSceneRenderRD::render_buffers_get_sdfgi_cascade_probe_offset(RI ERR_FAIL_COND_V(!rb, Vector3i()); ERR_FAIL_COND_V(!rb->sdfgi, Vector3i()); ERR_FAIL_UNSIGNED_INDEX_V(p_cascade, rb->sdfgi->cascades.size(), Vector3i()); - int32_t probe_divisor = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; + int32_t probe_divisor = rb->sdfgi->cascade_size / RendererSceneGIRD::SDFGI::PROBE_DIVISOR; return rb->sdfgi->cascades[p_cascade].position / probe_divisor; } @@ -5866,6 +2131,14 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p rb->msaa = p_msaa; rb->screen_space_aa = p_screen_space_aa; rb->use_debanding = p_use_debanding; + + if (is_clustered_enabled()) { + if (rb->cluster_builder == nullptr) { + rb->cluster_builder = memnew(ClusterBuilderRD); + } + rb->cluster_builder->set_shared(&cluster_builder_shared); + } + _free_render_buffer_data(rb); { @@ -5906,6 +2179,14 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p rb->data->configure(rb->texture, rb->depth_texture, p_width, p_height, p_msaa); _render_buffers_uniform_set_changed(p_render_buffers); + + if (is_clustered_enabled()) { + rb->cluster_builder->setup(Size2i(p_width, p_height), max_cluster_elements, rb->depth_texture, storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED), rb->texture); + } +} + +void RendererSceneRenderRD::gi_set_use_half_resolution(bool p_enable) { + gi.half_resolution = p_enable; } void RendererSceneRenderRD::sub_surface_scattering_set_quality(RS::SubSurfaceScatteringQuality p_quality) { @@ -6002,11 +2283,11 @@ void RendererSceneRenderRD::directional_shadow_quality_set(RS::ShadowQuality p_q } int RendererSceneRenderRD::get_roughness_layers() const { - return roughness_layers; + return sky.roughness_layers; } bool RendererSceneRenderRD::is_using_radiance_cubemap_array() const { - return sky_use_cubemap_array; + return sky.sky_use_cubemap_array; } RendererSceneRenderRD::RenderBufferData *RendererSceneRenderRD::render_buffers_get_data(RID p_render_buffers) { @@ -6016,17 +2297,34 @@ RendererSceneRenderRD::RenderBufferData *RendererSceneRenderRD::render_buffers_g } void RendererSceneRenderRD::_setup_reflections(const PagedArray<RID> &p_reflections, const Transform &p_camera_inverse_transform, RID p_environment) { + cluster.reflection_count = 0; + for (uint32_t i = 0; i < (uint32_t)p_reflections.size(); i++) { - RID rpi = p_reflections[i]; + if (cluster.reflection_count == cluster.max_reflections) { + break; + } - if (i >= cluster.max_reflections) { - reflection_probe_instance_set_render_index(rpi, 0); //invalid, but something needs to be set + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_reflections[i]); + if (!rpi) { continue; } - reflection_probe_instance_set_render_index(rpi, i); + cluster.reflection_sort[cluster.reflection_count].instance = rpi; + cluster.reflection_sort[cluster.reflection_count].depth = -p_camera_inverse_transform.xform(rpi->transform.origin).z; + cluster.reflection_count++; + } - RID base_probe = reflection_probe_instance_get_probe(rpi); + if (cluster.reflection_count > 0) { + SortArray<Cluster::InstanceSort<ReflectionProbeInstance>> sort_array; + sort_array.sort(cluster.reflection_sort, cluster.reflection_count); + } + + for (uint32_t i = 0; i < cluster.reflection_count; i++) { + ReflectionProbeInstance *rpi = cluster.reflection_sort[i].instance; + + rpi->render_index = i; + + RID base_probe = rpi->probe; Cluster::ReflectionData &reflection_ubo = cluster.reflections[i]; @@ -6035,7 +2333,7 @@ void RendererSceneRenderRD::_setup_reflections(const PagedArray<RID> &p_reflecti reflection_ubo.box_extents[0] = extents.x; reflection_ubo.box_extents[1] = extents.y; reflection_ubo.box_extents[2] = extents.z; - reflection_ubo.index = reflection_probe_instance_get_atlas_index(rpi); + reflection_ubo.index = rpi->atlas_index; Vector3 origin_offset = storage->reflection_probe_get_origin_offset(base_probe); @@ -6044,46 +2342,50 @@ void RendererSceneRenderRD::_setup_reflections(const PagedArray<RID> &p_reflecti reflection_ubo.box_offset[2] = origin_offset.z; reflection_ubo.mask = storage->reflection_probe_get_cull_mask(base_probe); - float intensity = storage->reflection_probe_get_intensity(base_probe); - bool interior = storage->reflection_probe_is_interior(base_probe); - bool box_projection = storage->reflection_probe_is_box_projection(base_probe); + reflection_ubo.intensity = storage->reflection_probe_get_intensity(base_probe); + reflection_ubo.ambient_mode = storage->reflection_probe_get_ambient_mode(base_probe); - reflection_ubo.params[0] = intensity; - reflection_ubo.params[1] = 0; - reflection_ubo.params[2] = interior ? 1.0 : 0.0; - reflection_ubo.params[3] = box_projection ? 1.0 : 0.0; + reflection_ubo.exterior = !storage->reflection_probe_is_interior(base_probe); + reflection_ubo.box_project = storage->reflection_probe_is_box_projection(base_probe); Color ambient_linear = storage->reflection_probe_get_ambient_color(base_probe).to_linear(); float interior_ambient_energy = storage->reflection_probe_get_ambient_color_energy(base_probe); - uint32_t ambient_mode = storage->reflection_probe_get_ambient_mode(base_probe); reflection_ubo.ambient[0] = ambient_linear.r * interior_ambient_energy; reflection_ubo.ambient[1] = ambient_linear.g * interior_ambient_energy; reflection_ubo.ambient[2] = ambient_linear.b * interior_ambient_energy; - reflection_ubo.ambient_mode = ambient_mode; - Transform transform = reflection_probe_instance_get_transform(rpi); + Transform transform = rpi->transform; Transform proj = (p_camera_inverse_transform * transform).inverse(); RendererStorageRD::store_transform(proj, reflection_ubo.local_matrix); - cluster.builder.add_reflection_probe(transform, extents); + current_cluster_builder->add_box(ClusterBuilderRD::BOX_TYPE_REFLECTION_PROBE, transform, extents); - reflection_probe_instance_set_render_pass(rpi, RSG::rasterizer->get_frame_number()); + rpi->last_pass = RSG::rasterizer->get_frame_number(); } - if (p_reflections.size()) { - RD::get_singleton()->buffer_update(cluster.reflection_buffer, 0, MIN(cluster.max_reflections, (unsigned int)p_reflections.size()) * sizeof(ReflectionData), cluster.reflections, true); + if (cluster.reflection_count) { + RD::get_singleton()->buffer_update(cluster.reflection_buffer, 0, cluster.reflection_count * sizeof(RendererSceneSkyRD::ReflectionData), cluster.reflections, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } } -void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const Transform &p_camera_inverse_transform, RID p_shadow_atlas, bool p_using_shadows, uint32_t &r_directional_light_count, uint32_t &r_positional_light_count) { - uint32_t light_count = 0; +void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const Transform &p_camera_transform, RID p_shadow_atlas, bool p_using_shadows, uint32_t &r_directional_light_count, uint32_t &r_positional_light_count) { + Transform inverse_transform = p_camera_transform.affine_inverse(); + r_directional_light_count = 0; r_positional_light_count = 0; - sky_scene_state.ubo.directional_light_count = 0; + sky.sky_scene_state.ubo.directional_light_count = 0; + + Plane camera_plane(p_camera_transform.origin, -p_camera_transform.basis.get_axis(Vector3::AXIS_Z).normalized()); + + cluster.omni_light_count = 0; + cluster.spot_light_count = 0; for (int i = 0; i < (int)p_lights.size(); i++) { - RID li = p_lights[i]; - RID base = light_instance_get_base_light(li); + LightInstance *li = light_instance_owner.getornull(p_lights[i]); + if (!li) { + continue; + } + RID base = li->light; ERR_CONTINUE(base.is_null()); @@ -6091,9 +2393,9 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const switch (type) { case RS::LIGHT_DIRECTIONAL: { // Copy to SkyDirectionalLightData - if (r_directional_light_count < sky_scene_state.max_directional_lights) { - SkyDirectionalLightData &sky_light_data = sky_scene_state.directional_lights[r_directional_light_count]; - Transform light_transform = light_instance_get_base_transform(li); + if (r_directional_light_count < sky.sky_scene_state.max_directional_lights) { + RendererSceneSkyRD::SkyDirectionalLightData &sky_light_data = sky.sky_scene_state.directional_lights[r_directional_light_count]; + Transform light_transform = li->transform; Vector3 world_direction = light_transform.basis.xform(Vector3(0, 0, 1)).normalized(); sky_light_data.direction[0] = world_direction.x; @@ -6120,7 +2422,7 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const angular_diameter = 0.0; } sky_light_data.size = angular_diameter; - sky_scene_state.ubo.directional_light_count++; + sky.sky_scene_state.ubo.directional_light_count++; } if (r_directional_light_count >= cluster.max_directional_lights || storage->light_directional_is_sky_only(base)) { @@ -6129,9 +2431,9 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const Cluster::DirectionalLightData &light_data = cluster.directional_lights[r_directional_light_count]; - Transform light_transform = light_instance_get_base_transform(li); + Transform light_transform = li->transform; - Vector3 direction = p_camera_inverse_transform.basis.xform(light_transform.basis.xform(Vector3(0, 0, 1))).normalized(); + Vector3 direction = inverse_transform.basis.xform(light_transform.basis.xform(Vector3(0, 0, 1))).normalized(); light_data.direction[0] = direction.x; light_data.direction[1] = direction.y; @@ -6210,28 +2512,28 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const int limit = smode == RS::LIGHT_DIRECTIONAL_SHADOW_ORTHOGONAL ? 0 : (smode == RS::LIGHT_DIRECTIONAL_SHADOW_PARALLEL_2_SPLITS ? 1 : 3); light_data.blend_splits = storage->light_directional_get_blend_splits(base); for (int j = 0; j < 4; j++) { - Rect2 atlas_rect = light_instance_get_directional_shadow_atlas_rect(li, j); - CameraMatrix matrix = light_instance_get_shadow_camera(li, j); - float split = light_instance_get_directional_shadow_split(li, MIN(limit, j)); + Rect2 atlas_rect = li->shadow_transform[j].atlas_rect; + CameraMatrix matrix = li->shadow_transform[j].camera; + float split = li->shadow_transform[MIN(limit, j)].split; CameraMatrix bias; bias.set_light_bias(); CameraMatrix rectm; rectm.set_light_atlas_rect(atlas_rect); - Transform modelview = (p_camera_inverse_transform * light_instance_get_shadow_transform(li, j)).inverse(); + Transform modelview = (inverse_transform * li->shadow_transform[j].transform).inverse(); CameraMatrix shadow_mtx = rectm * bias * matrix * modelview; light_data.shadow_split_offsets[j] = split; - float bias_scale = light_instance_get_shadow_bias_scale(li, j); + float bias_scale = li->shadow_transform[j].bias_scale; light_data.shadow_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * bias_scale; - light_data.shadow_normal_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * light_instance_get_directional_shadow_texel_size(li, j); + light_data.shadow_normal_bias[j] = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * li->shadow_transform[j].shadow_texel_size; light_data.shadow_transmittance_bias[j] = storage->light_get_transmittance_bias(base) * bias_scale; - light_data.shadow_z_range[j] = light_instance_get_shadow_range(li, j); - light_data.shadow_range_begin[j] = light_instance_get_shadow_range_begin(li, j); + light_data.shadow_z_range[j] = li->shadow_transform[j].farplane; + light_data.shadow_range_begin[j] = li->shadow_transform[j].range_begin; RendererStorageRD::store_camera(shadow_mtx, light_data.shadow_matrices[j]); - Vector2 uv_scale = light_instance_get_shadow_uv_scale(li, j); + Vector2 uv_scale = li->shadow_transform[j].uv_scale; uv_scale *= atlas_rect.size; //adapt to atlas size switch (j) { case 0: { @@ -6268,170 +2570,201 @@ void RendererSceneRenderRD::_setup_lights(const PagedArray<RID> &p_lights, const r_directional_light_count++; } break; - case RS::LIGHT_SPOT: case RS::LIGHT_OMNI: { - if (light_count >= cluster.max_lights) { + if (cluster.omni_light_count >= cluster.max_lights) { continue; } - Transform light_transform = light_instance_get_base_transform(li); + cluster.omni_light_sort[cluster.omni_light_count].instance = li; + cluster.omni_light_sort[cluster.omni_light_count].depth = camera_plane.distance_to(li->transform.origin); + cluster.omni_light_count++; + } break; + case RS::LIGHT_SPOT: { + if (cluster.spot_light_count >= cluster.max_lights) { + continue; + } - Cluster::LightData &light_data = cluster.lights[light_count]; - cluster.lights_instances[light_count] = li; + cluster.spot_light_sort[cluster.spot_light_count].instance = li; + cluster.spot_light_sort[cluster.spot_light_count].depth = camera_plane.distance_to(li->transform.origin); + cluster.spot_light_count++; + } break; + } - float sign = storage->light_is_negative(base) ? -1 : 1; - Color linear_col = storage->light_get_color(base).to_linear(); + li->last_pass = RSG::rasterizer->get_frame_number(); + } - light_data.attenuation_energy[0] = Math::make_half_float(storage->light_get_param(base, RS::LIGHT_PARAM_ATTENUATION)); - light_data.attenuation_energy[1] = Math::make_half_float(sign * storage->light_get_param(base, RS::LIGHT_PARAM_ENERGY) * Math_PI); + if (cluster.omni_light_count) { + SortArray<Cluster::InstanceSort<LightInstance>> sorter; + sorter.sort(cluster.omni_light_sort, cluster.omni_light_count); + } - light_data.color_specular[0] = MIN(uint32_t(linear_col.r * 255), 255); - light_data.color_specular[1] = MIN(uint32_t(linear_col.g * 255), 255); - light_data.color_specular[2] = MIN(uint32_t(linear_col.b * 255), 255); - light_data.color_specular[3] = MIN(uint32_t(storage->light_get_param(base, RS::LIGHT_PARAM_SPECULAR) * 255), 255); + if (cluster.spot_light_count) { + SortArray<Cluster::InstanceSort<LightInstance>> sorter; + sorter.sort(cluster.spot_light_sort, cluster.spot_light_count); + } - float radius = MAX(0.001, storage->light_get_param(base, RS::LIGHT_PARAM_RANGE)); - light_data.inv_radius = 1.0 / radius; + ShadowAtlas *shadow_atlas = nullptr; - Vector3 pos = p_camera_inverse_transform.xform(light_transform.origin); + if (p_shadow_atlas.is_valid() && p_using_shadows) { + shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + } - light_data.position[0] = pos.x; - light_data.position[1] = pos.y; - light_data.position[2] = pos.z; + for (uint32_t i = 0; i < (cluster.omni_light_count + cluster.spot_light_count); i++) { + uint32_t index = (i < cluster.omni_light_count) ? i : i - (cluster.omni_light_count); + Cluster::LightData &light_data = (i < cluster.omni_light_count) ? cluster.omni_lights[index] : cluster.spot_lights[index]; + RS::LightType type = (i < cluster.omni_light_count) ? RS::LIGHT_OMNI : RS::LIGHT_SPOT; + LightInstance *li = (i < cluster.omni_light_count) ? cluster.omni_light_sort[index].instance : cluster.spot_light_sort[index].instance; + RID base = li->light; - Vector3 direction = p_camera_inverse_transform.basis.xform(light_transform.basis.xform(Vector3(0, 0, -1))).normalized(); + Transform light_transform = li->transform; - light_data.direction[0] = direction.x; - light_data.direction[1] = direction.y; - light_data.direction[2] = direction.z; + float sign = storage->light_is_negative(base) ? -1 : 1; + Color linear_col = storage->light_get_color(base).to_linear(); - float size = storage->light_get_param(base, RS::LIGHT_PARAM_SIZE); + light_data.attenuation = storage->light_get_param(base, RS::LIGHT_PARAM_ATTENUATION); - light_data.size = size; + float energy = sign * storage->light_get_param(base, RS::LIGHT_PARAM_ENERGY) * Math_PI; - light_data.cone_attenuation_angle[0] = Math::make_half_float(storage->light_get_param(base, RS::LIGHT_PARAM_SPOT_ATTENUATION)); - float spot_angle = storage->light_get_param(base, RS::LIGHT_PARAM_SPOT_ANGLE); - light_data.cone_attenuation_angle[1] = Math::make_half_float(Math::cos(Math::deg2rad(spot_angle))); + light_data.color[0] = linear_col.r * energy; + light_data.color[1] = linear_col.g * energy; + light_data.color[2] = linear_col.b * energy; + light_data.specular_amount = storage->light_get_param(base, RS::LIGHT_PARAM_SPECULAR) * 2.0; - light_data.mask = storage->light_get_cull_mask(base); + float radius = MAX(0.001, storage->light_get_param(base, RS::LIGHT_PARAM_RANGE)); + light_data.inv_radius = 1.0 / radius; - light_data.atlas_rect[0] = 0; - light_data.atlas_rect[1] = 0; - light_data.atlas_rect[2] = 0; - light_data.atlas_rect[3] = 0; + Vector3 pos = inverse_transform.xform(light_transform.origin); - RID projector = storage->light_get_projector(base); + light_data.position[0] = pos.x; + light_data.position[1] = pos.y; + light_data.position[2] = pos.z; - if (projector.is_valid()) { - Rect2 rect = storage->decal_atlas_get_texture_rect(projector); + Vector3 direction = inverse_transform.basis.xform(light_transform.basis.xform(Vector3(0, 0, -1))).normalized(); - if (type == RS::LIGHT_SPOT) { - light_data.projector_rect[0] = rect.position.x; - light_data.projector_rect[1] = rect.position.y + rect.size.height; //flip because shadow is flipped - light_data.projector_rect[2] = rect.size.width; - light_data.projector_rect[3] = -rect.size.height; - } else { - light_data.projector_rect[0] = rect.position.x; - light_data.projector_rect[1] = rect.position.y; - light_data.projector_rect[2] = rect.size.width; - light_data.projector_rect[3] = rect.size.height * 0.5; //used by dp, so needs to be half - } - } else { - light_data.projector_rect[0] = 0; - light_data.projector_rect[1] = 0; - light_data.projector_rect[2] = 0; - light_data.projector_rect[3] = 0; - } + light_data.direction[0] = direction.x; + light_data.direction[1] = direction.y; + light_data.direction[2] = direction.z; - if (p_using_shadows && p_shadow_atlas.is_valid() && shadow_atlas_owns_light_instance(p_shadow_atlas, li)) { - // fill in the shadow information + float size = storage->light_get_param(base, RS::LIGHT_PARAM_SIZE); - Color shadow_color = storage->light_get_shadow_color(base); + light_data.size = size; - light_data.shadow_color_enabled[0] = MIN(uint32_t(shadow_color.r * 255), 255); - light_data.shadow_color_enabled[1] = MIN(uint32_t(shadow_color.g * 255), 255); - light_data.shadow_color_enabled[2] = MIN(uint32_t(shadow_color.b * 255), 255); - light_data.shadow_color_enabled[3] = 255; + light_data.inv_spot_attenuation = 1.0f / storage->light_get_param(base, RS::LIGHT_PARAM_SPOT_ATTENUATION); + float spot_angle = storage->light_get_param(base, RS::LIGHT_PARAM_SPOT_ANGLE); + light_data.cos_spot_angle = Math::cos(Math::deg2rad(spot_angle)); - if (type == RS::LIGHT_SPOT) { - light_data.shadow_bias = (storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0); - float shadow_texel_size = Math::tan(Math::deg2rad(spot_angle)) * radius * 2.0; - shadow_texel_size *= light_instance_get_shadow_texel_size(li, p_shadow_atlas); + light_data.mask = storage->light_get_cull_mask(base); - light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size; + light_data.atlas_rect[0] = 0; + light_data.atlas_rect[1] = 0; + light_data.atlas_rect[2] = 0; + light_data.atlas_rect[3] = 0; - } else { //omni - light_data.shadow_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0; - float shadow_texel_size = light_instance_get_shadow_texel_size(li, p_shadow_atlas); - light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size * 2.0; // applied in -1 .. 1 space - } + RID projector = storage->light_get_projector(base); + + if (projector.is_valid()) { + Rect2 rect = storage->decal_atlas_get_texture_rect(projector); + + if (type == RS::LIGHT_SPOT) { + light_data.projector_rect[0] = rect.position.x; + light_data.projector_rect[1] = rect.position.y + rect.size.height; //flip because shadow is flipped + light_data.projector_rect[2] = rect.size.width; + light_data.projector_rect[3] = -rect.size.height; + } else { + light_data.projector_rect[0] = rect.position.x; + light_data.projector_rect[1] = rect.position.y; + light_data.projector_rect[2] = rect.size.width; + light_data.projector_rect[3] = rect.size.height * 0.5; //used by dp, so needs to be half + } + } else { + light_data.projector_rect[0] = 0; + light_data.projector_rect[1] = 0; + light_data.projector_rect[2] = 0; + light_data.projector_rect[3] = 0; + } - light_data.transmittance_bias = storage->light_get_transmittance_bias(base); + if (shadow_atlas && shadow_atlas->shadow_owners.has(li->self)) { + // fill in the shadow information - Rect2 rect = light_instance_get_shadow_atlas_rect(li, p_shadow_atlas); + light_data.shadow_enabled = true; - light_data.atlas_rect[0] = rect.position.x; - light_data.atlas_rect[1] = rect.position.y; - light_data.atlas_rect[2] = rect.size.width; - light_data.atlas_rect[3] = rect.size.height; + if (type == RS::LIGHT_SPOT) { + light_data.shadow_bias = (storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0); + float shadow_texel_size = Math::tan(Math::deg2rad(spot_angle)) * radius * 2.0; + shadow_texel_size *= light_instance_get_shadow_texel_size(li->self, p_shadow_atlas); - light_data.soft_shadow_scale = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BLUR); - light_data.shadow_volumetric_fog_fade = 1.0 / storage->light_get_shadow_volumetric_fog_fade(base); + light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size; - if (type == RS::LIGHT_OMNI) { - light_data.atlas_rect[3] *= 0.5; //one paraboloid on top of another - Transform proj = (p_camera_inverse_transform * light_transform).inverse(); + } else { //omni + light_data.shadow_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BIAS) * radius / 10.0; + float shadow_texel_size = light_instance_get_shadow_texel_size(li->self, p_shadow_atlas); + light_data.shadow_normal_bias = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * shadow_texel_size * 2.0; // applied in -1 .. 1 space + } - RendererStorageRD::store_transform(proj, light_data.shadow_matrix); + light_data.transmittance_bias = storage->light_get_transmittance_bias(base); - if (size > 0.0) { - light_data.soft_shadow_size = size; - } else { - light_data.soft_shadow_size = 0.0; - light_data.soft_shadow_scale *= shadows_quality_radius_get(); // Only use quality radius for PCF - } + Rect2 rect = light_instance_get_shadow_atlas_rect(li->self, p_shadow_atlas); - } else if (type == RS::LIGHT_SPOT) { - Transform modelview = (p_camera_inverse_transform * light_transform).inverse(); - CameraMatrix bias; - bias.set_light_bias(); + light_data.atlas_rect[0] = rect.position.x; + light_data.atlas_rect[1] = rect.position.y; + light_data.atlas_rect[2] = rect.size.width; + light_data.atlas_rect[3] = rect.size.height; - CameraMatrix shadow_mtx = bias * light_instance_get_shadow_camera(li, 0) * modelview; - RendererStorageRD::store_camera(shadow_mtx, light_data.shadow_matrix); + light_data.soft_shadow_scale = storage->light_get_param(base, RS::LIGHT_PARAM_SHADOW_BLUR); + light_data.shadow_volumetric_fog_fade = 1.0 / storage->light_get_shadow_volumetric_fog_fade(base); - if (size > 0.0) { - CameraMatrix cm = light_instance_get_shadow_camera(li, 0); - float half_np = cm.get_z_near() * Math::tan(Math::deg2rad(spot_angle)); - light_data.soft_shadow_size = (size * 0.5 / radius) / (half_np / cm.get_z_near()) * rect.size.width; - } else { - light_data.soft_shadow_size = 0.0; - light_data.soft_shadow_scale *= shadows_quality_radius_get(); // Only use quality radius for PCF - } - } + if (type == RS::LIGHT_OMNI) { + light_data.atlas_rect[3] *= 0.5; //one paraboloid on top of another + Transform proj = (inverse_transform * light_transform).inverse(); + + RendererStorageRD::store_transform(proj, light_data.shadow_matrix); + + if (size > 0.0) { + light_data.soft_shadow_size = size; } else { - light_data.shadow_color_enabled[3] = 0; + light_data.soft_shadow_size = 0.0; + light_data.soft_shadow_scale *= shadows_quality_radius_get(); // Only use quality radius for PCF } - light_instance_set_index(li, light_count); + } else if (type == RS::LIGHT_SPOT) { + Transform modelview = (inverse_transform * light_transform).inverse(); + CameraMatrix bias; + bias.set_light_bias(); - cluster.builder.add_light(type == RS::LIGHT_SPOT ? LightClusterBuilder::LIGHT_TYPE_SPOT : LightClusterBuilder::LIGHT_TYPE_OMNI, light_transform, radius, spot_angle); + CameraMatrix shadow_mtx = bias * li->shadow_transform[0].camera * modelview; + RendererStorageRD::store_camera(shadow_mtx, light_data.shadow_matrix); - light_count++; - r_positional_light_count++; - } break; + if (size > 0.0) { + CameraMatrix cm = li->shadow_transform[0].camera; + float half_np = cm.get_z_near() * Math::tan(Math::deg2rad(spot_angle)); + light_data.soft_shadow_size = (size * 0.5 / radius) / (half_np / cm.get_z_near()) * rect.size.width; + } else { + light_data.soft_shadow_size = 0.0; + light_data.soft_shadow_scale *= shadows_quality_radius_get(); // Only use quality radius for PCF + } + } + } else { + light_data.shadow_enabled = false; } - light_instance_set_render_pass(li, RSG::rasterizer->get_frame_number()); + li->light_index = index; + + current_cluster_builder->add_light(type == RS::LIGHT_SPOT ? ClusterBuilderRD::LIGHT_TYPE_SPOT : ClusterBuilderRD::LIGHT_TYPE_OMNI, light_transform, radius, spot_angle); + + r_positional_light_count++; + } - //update UBO for forward rendering, blit to texture for clustered + //update without barriers + if (cluster.omni_light_count) { + RD::get_singleton()->buffer_update(cluster.omni_light_buffer, 0, sizeof(Cluster::LightData) * cluster.omni_light_count, cluster.omni_lights, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } - if (light_count) { - RD::get_singleton()->buffer_update(cluster.light_buffer, 0, sizeof(Cluster::LightData) * light_count, cluster.lights, true); + if (cluster.spot_light_count) { + RD::get_singleton()->buffer_update(cluster.spot_light_buffer, 0, sizeof(Cluster::LightData) * cluster.spot_light_count, cluster.spot_lights, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } if (r_directional_light_count) { - RD::get_singleton()->buffer_update(cluster.directional_light_buffer, 0, sizeof(Cluster::DirectionalLightData) * r_directional_light_count, cluster.directional_lights, true); + RD::get_singleton()->buffer_update(cluster.directional_light_buffer, 0, sizeof(Cluster::DirectionalLightData) * r_directional_light_count, cluster.directional_lights, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } } @@ -6440,18 +2773,26 @@ void RendererSceneRenderRD::_setup_decals(const PagedArray<RID> &p_decals, const uv_xform.basis.scale(Vector3(2.0, 1.0, 2.0)); uv_xform.origin = Vector3(-1.0, 0.0, -1.0); - uint32_t decal_count = MIN((uint32_t)p_decals.size(), cluster.max_decals); - int idx = 0; + uint32_t decal_count = p_decals.size(); + + cluster.decal_count = 0; + for (uint32_t i = 0; i < decal_count; i++) { - RID di = p_decals[i]; - RID decal = decal_instance_get_base(di); + if (cluster.decal_count == cluster.max_decals) { + break; + } - Transform xform = decal_instance_get_transform(di); + DecalInstance *di = decal_instance_owner.getornull(p_decals[i]); + if (!di) { + continue; + } + RID decal = di->decal; - float fade = 1.0; + Transform xform = di->transform; + + real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; if (storage->decal_is_distance_fade_enabled(decal)) { - real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; float fade_begin = storage->decal_get_distance_fade_begin(decal); float fade_length = storage->decal_get_distance_fade_length(decal); @@ -6459,18 +2800,43 @@ void RendererSceneRenderRD::_setup_decals(const PagedArray<RID> &p_decals, const if (distance > fade_begin + fade_length) { continue; // do not use this decal, its invisible } + } + } + cluster.decal_sort[cluster.decal_count].instance = di; + cluster.decal_sort[cluster.decal_count].depth = distance; + cluster.decal_count++; + } + + if (cluster.decal_count > 0) { + SortArray<Cluster::InstanceSort<DecalInstance>> sort_array; + sort_array.sort(cluster.decal_sort, cluster.decal_count); + } + + for (uint32_t i = 0; i < cluster.decal_count; i++) { + DecalInstance *di = cluster.decal_sort[i].instance; + RID decal = di->decal; + + Transform xform = di->transform; + float fade = 1.0; + + if (storage->decal_is_distance_fade_enabled(decal)) { + real_t distance = -p_camera_inverse_xform.xform(xform.origin).z; + float fade_begin = storage->decal_get_distance_fade_begin(decal); + float fade_length = storage->decal_get_distance_fade_length(decal); + + if (distance > fade_begin) { fade = 1.0 - (distance - fade_begin) / fade_length; } } - Cluster::DecalData &dd = cluster.decals[idx]; + Cluster::DecalData &dd = cluster.decals[i]; Vector3 decal_extents = storage->decal_get_extents(decal); Transform scale_xform; scale_xform.basis.scale(Vector3(decal_extents.x, decal_extents.y, decal_extents.z)); - Transform to_decal_xform = (p_camera_inverse_xform * decal_instance_get_transform(di) * scale_xform * uv_xform).affine_inverse(); + Transform to_decal_xform = (p_camera_inverse_xform * di->transform * scale_xform * uv_xform).affine_inverse(); RendererStorageRD::store_transform(to_decal_xform, dd.xform); Vector3 normal = xform.basis.get_axis(Vector3::AXIS_Y).normalized(); @@ -6555,19 +2921,18 @@ void RendererSceneRenderRD::_setup_decals(const PagedArray<RID> &p_decals, const dd.upper_fade = storage->decal_get_upper_fade(decal); dd.lower_fade = storage->decal_get_lower_fade(decal); - cluster.builder.add_decal(xform, decal_extents); - - idx++; + current_cluster_builder->add_box(ClusterBuilderRD::BOX_TYPE_DECAL, xform, decal_extents); } - if (idx > 0) { - RD::get_singleton()->buffer_update(cluster.decal_buffer, 0, sizeof(Cluster::DecalData) * idx, cluster.decals, true); + if (cluster.decal_count > 0) { + RD::get_singleton()->buffer_update(cluster.decal_buffer, 0, sizeof(Cluster::DecalData) * cluster.decal_count, cluster.decals, RD::BARRIER_MASK_RASTER | RD::BARRIER_MASK_COMPUTE); } } void RendererSceneRenderRD::_volumetric_fog_erase(RenderBuffers *rb) { ERR_FAIL_COND(!rb->volumetric_fog); + RD::get_singleton()->free(rb->volumetric_fog->prev_light_density_map); RD::get_singleton()->free(rb->volumetric_fog->light_density_map); RD::get_singleton()->free(rb->volumetric_fog->fog_map); @@ -6589,53 +2954,11 @@ void RendererSceneRenderRD::_volumetric_fog_erase(RenderBuffers *rb) { rb->volumetric_fog = nullptr; } -void RendererSceneRenderRD::_allocate_shadow_shrink_stages(RID p_base, int p_base_size, Vector<ShadowShrinkStage> &shrink_stages, uint32_t p_target_size) { - //create fog mipmaps - uint32_t fog_texture_size = p_target_size; - uint32_t base_texture_size = p_base_size; - - ShadowShrinkStage first; - first.size = base_texture_size; - first.texture = p_base; - shrink_stages.push_back(first); //put depth first in case we dont find smaller ones - - while (fog_texture_size < base_texture_size) { - base_texture_size = MAX(base_texture_size / 8, fog_texture_size); - - ShadowShrinkStage s; - s.size = base_texture_size; - - RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R32_SFLOAT; - tf.width = base_texture_size; - tf.height = base_texture_size; - tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; - - if (base_texture_size == fog_texture_size) { - s.filter_texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); - tf.usage_bits |= RD::TEXTURE_USAGE_SAMPLING_BIT; - } - - s.texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); - - shrink_stages.push_back(s); - } -} - -void RendererSceneRenderRD::_clear_shadow_shrink_stages(Vector<ShadowShrinkStage> &shrink_stages) { - for (int i = 1; i < shrink_stages.size(); i++) { - RD::get_singleton()->free(shrink_stages[i].texture); - if (shrink_stages[i].filter_texture.is_valid()) { - RD::get_singleton()->free(shrink_stages[i].filter_texture); - } - } - shrink_stages.clear(); -} - void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_environment, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_shadow_atlas, int p_directional_light_count, bool p_use_directional_shadows, int p_positional_light_count, int p_gi_probe_count) { + ERR_FAIL_COND(!is_clustered_enabled()); // can't use volumetric fog without clustered RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); ERR_FAIL_COND(!rb); - Environment *env = environment_owner.getornull(p_environment); + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_environment); float ratio = float(rb->width) / float((rb->width + rb->height) / 2); uint32_t target_width = uint32_t(float(volumetric_fog_size) * ratio); @@ -6654,6 +2977,8 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e return; } + RENDER_TIMESTAMP(">Volumetric Fog"); + if (env && env->volumetric_fog_enabled && !rb->volumetric_fog) { //required volumetric fog but not existing, create rb->volumetric_fog = memnew(VolumetricFog); @@ -6667,11 +2992,16 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e tf.height = target_height; tf.depth = volumetric_fog_depth; tf.texture_type = RD::TEXTURE_TYPE_3D; - tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; rb->volumetric_fog->light_density_map = RD::get_singleton()->texture_create(tf, RD::TextureView()); - tf.usage_bits |= RD::TEXTURE_USAGE_SAMPLING_BIT; + tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; + + rb->volumetric_fog->prev_light_density_map = RD::get_singleton()->texture_create(tf, RD::TextureView()); + RD::get_singleton()->texture_clear(rb->volumetric_fog->prev_light_density_map, Color(0, 0, 0, 0), 0, 1, 0, 1); + + tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT; rb->volumetric_fog->fog_map = RD::get_singleton()->texture_create(tf, RD::TextureView()); _render_buffers_uniform_set_changed(p_render_buffers); @@ -6685,163 +3015,7 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e uniforms.push_back(u); } - rb->volumetric_fog->sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_FOG); - } - - //update directional shadow - - if (p_use_directional_shadows) { - if (directional_shadow.shrink_stages.is_empty()) { - if (rb->volumetric_fog->uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(rb->volumetric_fog->uniform_set)) { - //invalidate uniform set, we will need a new one - RD::get_singleton()->free(rb->volumetric_fog->uniform_set); - rb->volumetric_fog->uniform_set = RID(); - } - _allocate_shadow_shrink_stages(directional_shadow.depth, directional_shadow.size, directional_shadow.shrink_stages, volumetric_fog_directional_shadow_shrink); - } - - if (directional_shadow.shrink_stages.size() > 1) { - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - for (int i = 1; i < directional_shadow.shrink_stages.size(); i++) { - int32_t src_size = directional_shadow.shrink_stages[i - 1].size; - int32_t dst_size = directional_shadow.shrink_stages[i].size; - Rect2i r(0, 0, src_size, src_size); - int32_t shrink_limit = 8 / (src_size / dst_size); - - storage->get_effects()->reduce_shadow(directional_shadow.shrink_stages[i - 1].texture, directional_shadow.shrink_stages[i].texture, Size2i(src_size, src_size), r, shrink_limit, compute_list); - RD::get_singleton()->compute_list_add_barrier(compute_list); - if (env->volumetric_fog_shadow_filter != RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_DISABLED && directional_shadow.shrink_stages[i].filter_texture.is_valid()) { - Rect2i rf(0, 0, dst_size, dst_size); - storage->get_effects()->filter_shadow(directional_shadow.shrink_stages[i].texture, directional_shadow.shrink_stages[i].filter_texture, Size2i(dst_size, dst_size), rf, env->volumetric_fog_shadow_filter, compute_list); - } - } - RD::get_singleton()->compute_list_end(); - } - } - - ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); - - if (shadow_atlas) { - //shrink shadows that need to be shrunk - - bool force_shrink_shadows = false; - - if (shadow_atlas->shrink_stages.is_empty()) { - if (rb->volumetric_fog->uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(rb->volumetric_fog->uniform_set)) { - //invalidate uniform set, we will need a new one - RD::get_singleton()->free(rb->volumetric_fog->uniform_set); - rb->volumetric_fog->uniform_set = RID(); - } - _allocate_shadow_shrink_stages(shadow_atlas->depth, shadow_atlas->size, shadow_atlas->shrink_stages, volumetric_fog_positional_shadow_shrink); - force_shrink_shadows = true; - } - - if (rb->volumetric_fog->last_shadow_filter != env->volumetric_fog_shadow_filter) { - //if shadow filter changed, invalidate caches - rb->volumetric_fog->last_shadow_filter = env->volumetric_fog_shadow_filter; - force_shrink_shadows = true; - } - - cluster.lights_shadow_rect_cache_count = 0; - - for (int i = 0; i < p_positional_light_count; i++) { - if (cluster.lights[i].shadow_color_enabled[3] > 127) { - RID li = cluster.lights_instances[i]; - - ERR_CONTINUE(!shadow_atlas->shadow_owners.has(li)); - - uint32_t key = shadow_atlas->shadow_owners[li]; - - uint32_t quadrant = (key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; - uint32_t shadow = key & ShadowAtlas::SHADOW_INDEX_MASK; - - ERR_CONTINUE((int)shadow >= shadow_atlas->quadrants[quadrant].shadows.size()); - - ShadowAtlas::Quadrant::Shadow &s = shadow_atlas->quadrants[quadrant].shadows.write[shadow]; - - if (!force_shrink_shadows && s.fog_version == s.version) { - continue; //do not update, no need - } - - s.fog_version = s.version; - - uint32_t quadrant_size = shadow_atlas->size >> 1; - - Rect2i atlas_rect; - - atlas_rect.position.x = (quadrant & 1) * quadrant_size; - atlas_rect.position.y = (quadrant >> 1) * quadrant_size; - - uint32_t shadow_size = (quadrant_size / shadow_atlas->quadrants[quadrant].subdivision); - atlas_rect.position.x += (shadow % shadow_atlas->quadrants[quadrant].subdivision) * shadow_size; - atlas_rect.position.y += (shadow / shadow_atlas->quadrants[quadrant].subdivision) * shadow_size; - - atlas_rect.size.x = shadow_size; - atlas_rect.size.y = shadow_size; - - cluster.lights_shadow_rect_cache[cluster.lights_shadow_rect_cache_count] = atlas_rect; - - cluster.lights_shadow_rect_cache_count++; - - if (cluster.lights_shadow_rect_cache_count == cluster.max_lights) { - break; //light limit reached - } - } - } - - if (cluster.lights_shadow_rect_cache_count > 0) { - //there are shadows to be shrunk, try to do them in parallel - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - - for (int i = 1; i < shadow_atlas->shrink_stages.size(); i++) { - int32_t base_size = shadow_atlas->shrink_stages[0].size; - int32_t src_size = shadow_atlas->shrink_stages[i - 1].size; - int32_t dst_size = shadow_atlas->shrink_stages[i].size; - - uint32_t rect_divisor = base_size / src_size; - - int32_t shrink_limit = 8 / (src_size / dst_size); - - //shrink in parallel for more performance - for (uint32_t j = 0; j < cluster.lights_shadow_rect_cache_count; j++) { - Rect2i src_rect = cluster.lights_shadow_rect_cache[j]; - - src_rect.position /= rect_divisor; - src_rect.size /= rect_divisor; - - storage->get_effects()->reduce_shadow(shadow_atlas->shrink_stages[i - 1].texture, shadow_atlas->shrink_stages[i].texture, Size2i(src_size, src_size), src_rect, shrink_limit, compute_list); - } - - RD::get_singleton()->compute_list_add_barrier(compute_list); - - if (env->volumetric_fog_shadow_filter != RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_DISABLED && shadow_atlas->shrink_stages[i].filter_texture.is_valid()) { - uint32_t filter_divisor = base_size / dst_size; - - //filter in parallel for more performance - for (uint32_t j = 0; j < cluster.lights_shadow_rect_cache_count; j++) { - Rect2i dst_rect = cluster.lights_shadow_rect_cache[j]; - - dst_rect.position /= filter_divisor; - dst_rect.size /= filter_divisor; - - storage->get_effects()->filter_shadow(shadow_atlas->shrink_stages[i].texture, shadow_atlas->shrink_stages[i].filter_texture, Size2i(dst_size, dst_size), dst_rect, env->volumetric_fog_shadow_filter, compute_list, true, false); - } - - RD::get_singleton()->compute_list_add_barrier(compute_list); - - for (uint32_t j = 0; j < cluster.lights_shadow_rect_cache_count; j++) { - Rect2i dst_rect = cluster.lights_shadow_rect_cache[j]; - - dst_rect.position /= filter_divisor; - dst_rect.size /= filter_divisor; - - storage->get_effects()->filter_shadow(shadow_atlas->shrink_stages[i].texture, shadow_atlas->shrink_stages[i].filter_texture, Size2i(dst_size, dst_size), dst_rect, env->volumetric_fog_shadow_filter, compute_list, false, true); - } - } - } - - RD::get_singleton()->compute_list_end(); - } + rb->volumetric_fog->sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky.sky_shader.default_shader_rd, RendererSceneSkyRD::SKY_SET_FOG); } //update volumetric fog @@ -6855,10 +3029,11 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 1; - if (shadow_atlas == nullptr || shadow_atlas->shrink_stages.size() == 0) { + ShadowAtlas *shadow_atlas = shadow_atlas_owner.getornull(p_shadow_atlas); + if (shadow_atlas == nullptr || shadow_atlas->depth.is_null()) { u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK)); } else { - u.ids.push_back(shadow_atlas->shrink_stages[shadow_atlas->shrink_stages.size() - 1].texture); + u.ids.push_back(shadow_atlas->depth); } uniforms.push_back(u); @@ -6868,10 +3043,10 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 2; - if (directional_shadow.shrink_stages.size() == 0) { - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK)); + if (directional_shadow.depth.is_valid()) { + u.ids.push_back(directional_shadow.depth); } else { - u.ids.push_back(directional_shadow.shrink_stages[directional_shadow.shrink_stages.size() - 1].texture); + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK)); } uniforms.push_back(u); } @@ -6880,23 +3055,22 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.binding = 3; - u.ids.push_back(get_positional_light_buffer()); + u.ids.push_back(get_omni_light_buffer()); uniforms.push_back(u); } - { RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.binding = 4; - u.ids.push_back(get_directional_light_buffer()); + u.ids.push_back(get_spot_light_buffer()); uniforms.push_back(u); } { RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; u.binding = 5; - u.ids.push_back(get_cluster_builder_texture()); + u.ids.push_back(get_directional_light_buffer()); uniforms.push_back(u); } @@ -6904,7 +3078,7 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; u.binding = 6; - u.ids.push_back(get_cluster_builder_indices_buffer()); + u.ids.push_back(rb->cluster_builder->get_cluster_buffer()); uniforms.push_back(u); } @@ -6952,8 +3126,8 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::Uniform u; u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; u.binding = 12; - for (int i = 0; i < RenderBuffers::MAX_GIPROBES; i++) { - u.ids.push_back(rb->giprobe_textures[i]); + for (int i = 0; i < RendererSceneGIRD::MAX_GIPROBES; i++) { + u.ids.push_back(rb->gi.giprobe_textures[i]); } uniforms.push_back(u); } @@ -6964,6 +3138,20 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); uniforms.push_back(u); } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 14; + u.ids.push_back(volumetric_fog.params_ubo); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 15; + u.ids.push_back(rb->volumetric_fog->prev_light_density_map); + uniforms.push_back(u); + } rb->volumetric_fog->uniform_set = RD::get_singleton()->uniform_set_create(uniforms, volumetric_fog.shader.version_get_shader(volumetric_fog.shader_version, 0), 0); @@ -7009,7 +3197,7 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e rb->volumetric_fog->length = env->volumetric_fog_length; rb->volumetric_fog->spread = env->volumetric_fog_detail_spread; - VolumetricFogShader::PushConstant push_constant; + VolumetricFogShader::ParamsUBO params; Vector2 frustum_near_size = p_cam_projection.get_viewport_half_extents(); Vector2 frustum_far_size = p_cam_projection.get_far_plane_half_extents(); @@ -7025,51 +3213,78 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e fog_near_size = Vector2(); } - push_constant.fog_frustum_size_begin[0] = fog_near_size.x; - push_constant.fog_frustum_size_begin[1] = fog_near_size.y; + params.fog_frustum_size_begin[0] = fog_near_size.x; + params.fog_frustum_size_begin[1] = fog_near_size.y; - push_constant.fog_frustum_size_end[0] = fog_far_size.x; - push_constant.fog_frustum_size_end[1] = fog_far_size.y; + params.fog_frustum_size_end[0] = fog_far_size.x; + params.fog_frustum_size_end[1] = fog_far_size.y; - push_constant.z_near = z_near; - push_constant.z_far = z_far; + params.z_near = z_near; + params.z_far = z_far; - push_constant.fog_frustum_end = fog_end; + params.fog_frustum_end = fog_end; - push_constant.fog_volume_size[0] = rb->volumetric_fog->width; - push_constant.fog_volume_size[1] = rb->volumetric_fog->height; - push_constant.fog_volume_size[2] = rb->volumetric_fog->depth; + params.fog_volume_size[0] = rb->volumetric_fog->width; + params.fog_volume_size[1] = rb->volumetric_fog->height; + params.fog_volume_size[2] = rb->volumetric_fog->depth; - push_constant.directional_light_count = p_directional_light_count; + params.directional_light_count = p_directional_light_count; Color light = env->volumetric_fog_light.to_linear(); - push_constant.light_energy[0] = light.r * env->volumetric_fog_light_energy; - push_constant.light_energy[1] = light.g * env->volumetric_fog_light_energy; - push_constant.light_energy[2] = light.b * env->volumetric_fog_light_energy; - push_constant.base_density = env->volumetric_fog_density; - - push_constant.detail_spread = env->volumetric_fog_detail_spread; - push_constant.gi_inject = env->volumetric_fog_gi_inject; - - push_constant.cam_rotation[0] = p_cam_transform.basis[0][0]; - push_constant.cam_rotation[1] = p_cam_transform.basis[1][0]; - push_constant.cam_rotation[2] = p_cam_transform.basis[2][0]; - push_constant.cam_rotation[3] = 0; - push_constant.cam_rotation[4] = p_cam_transform.basis[0][1]; - push_constant.cam_rotation[5] = p_cam_transform.basis[1][1]; - push_constant.cam_rotation[6] = p_cam_transform.basis[2][1]; - push_constant.cam_rotation[7] = 0; - push_constant.cam_rotation[8] = p_cam_transform.basis[0][2]; - push_constant.cam_rotation[9] = p_cam_transform.basis[1][2]; - push_constant.cam_rotation[10] = p_cam_transform.basis[2][2]; - push_constant.cam_rotation[11] = 0; - push_constant.filter_axis = 0; - push_constant.max_gi_probes = env->volumetric_fog_gi_inject > 0.001 ? p_gi_probe_count : 0; + params.light_energy[0] = light.r * env->volumetric_fog_light_energy; + params.light_energy[1] = light.g * env->volumetric_fog_light_energy; + params.light_energy[2] = light.b * env->volumetric_fog_light_energy; + params.base_density = env->volumetric_fog_density; + + params.detail_spread = env->volumetric_fog_detail_spread; + params.gi_inject = env->volumetric_fog_gi_inject; + + params.cam_rotation[0] = p_cam_transform.basis[0][0]; + params.cam_rotation[1] = p_cam_transform.basis[1][0]; + params.cam_rotation[2] = p_cam_transform.basis[2][0]; + params.cam_rotation[3] = 0; + params.cam_rotation[4] = p_cam_transform.basis[0][1]; + params.cam_rotation[5] = p_cam_transform.basis[1][1]; + params.cam_rotation[6] = p_cam_transform.basis[2][1]; + params.cam_rotation[7] = 0; + params.cam_rotation[8] = p_cam_transform.basis[0][2]; + params.cam_rotation[9] = p_cam_transform.basis[1][2]; + params.cam_rotation[10] = p_cam_transform.basis[2][2]; + params.cam_rotation[11] = 0; + params.filter_axis = 0; + params.max_gi_probes = env->volumetric_fog_gi_inject > 0.001 ? p_gi_probe_count : 0; + params.temporal_frame = RSG::rasterizer->get_frame_number() % VolumetricFog::MAX_TEMPORAL_FRAMES; + + Transform to_prev_cam_view = rb->volumetric_fog->prev_cam_transform.affine_inverse() * p_cam_transform; + storage->store_transform(to_prev_cam_view, params.to_prev_view); + + params.use_temporal_reprojection = env->volumetric_fog_temporal_reprojection; + params.temporal_blend = env->volumetric_fog_temporal_reprojection_amount; + + { + uint32_t cluster_size = rb->cluster_builder->get_cluster_size(); + params.cluster_shift = get_shift_from_power_of_2(cluster_size); + + uint32_t cluster_screen_width = (rb->width - 1) / cluster_size + 1; + uint32_t cluster_screen_height = (rb->height - 1) / cluster_size + 1; + params.cluster_type_size = cluster_screen_width * cluster_screen_height * (32 + 32); + params.cluster_width = cluster_screen_width; + params.max_cluster_element_count_div_32 = max_cluster_elements / 32; + + params.screen_size[0] = rb->width; + params.screen_size[1] = rb->height; + } /* Vector2 dssize = directional_shadow_get_size(); push_constant.directional_shadow_pixel_size[0] = 1.0 / dssize.x; push_constant.directional_shadow_pixel_size[1] = 1.0 / dssize.y; */ + + RD::get_singleton()->draw_command_begin_label("Render Volumetric Fog"); + + RENDER_TIMESTAMP("Render Fog"); + RD::get_singleton()->buffer_update(volumetric_fog.params_ubo, 0, sizeof(VolumetricFogShader::ParamsUBO), ¶ms, RD::BARRIER_MASK_COMPUTE); + RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); bool use_filter = volumetric_fog_filter_active; @@ -7077,93 +3292,225 @@ void RendererSceneRenderRD::_update_volumetric_fog(RID p_render_buffers, RID p_e RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, volumetric_fog.pipelines[using_sdfgi ? VOLUMETRIC_FOG_SHADER_DENSITY_WITH_SDFGI : VOLUMETRIC_FOG_SHADER_DENSITY]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->uniform_set, 0); + if (using_sdfgi) { RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->sdfgi_uniform_set, 1); } - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(VolumetricFogShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth, 4, 4, 4); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth); + + RD::get_singleton()->draw_command_end_label(); + + RD::get_singleton()->compute_list_end(); - RD::get_singleton()->compute_list_add_barrier(compute_list); + RD::get_singleton()->texture_copy(rb->volumetric_fog->light_density_map, rb->volumetric_fog->prev_light_density_map, Vector3(0, 0, 0), Vector3(0, 0, 0), Vector3(rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth), 0, 0, 0, 0); + + compute_list = RD::get_singleton()->compute_list_begin(); if (use_filter) { + RD::get_singleton()->draw_command_begin_label("Filter Fog"); + + RENDER_TIMESTAMP("Filter Fog"); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, volumetric_fog.pipelines[VOLUMETRIC_FOG_SHADER_FILTER]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(VolumetricFogShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth); - RD::get_singleton()->compute_list_add_barrier(compute_list); + RD::get_singleton()->compute_list_end(); + //need restart for buffer update - push_constant.filter_axis = 1; + params.filter_axis = 1; + RD::get_singleton()->buffer_update(volumetric_fog.params_ubo, 0, sizeof(VolumetricFogShader::ParamsUBO), ¶ms); + compute_list = RD::get_singleton()->compute_list_begin(); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, volumetric_fog.pipelines[VOLUMETRIC_FOG_SHADER_FILTER]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->uniform_set2, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(VolumetricFogShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth, 8, 8, 1); + if (using_sdfgi) { + RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->sdfgi_uniform_set, 1); + } + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, rb->volumetric_fog->depth); RD::get_singleton()->compute_list_add_barrier(compute_list); + RD::get_singleton()->draw_command_end_label(); } + RENDER_TIMESTAMP("Integrate Fog"); + RD::get_singleton()->draw_command_begin_label("Integrate Fog"); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, volumetric_fog.pipelines[VOLUMETRIC_FOG_SHADER_FOG]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->volumetric_fog->uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(VolumetricFogShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, 1, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->volumetric_fog->width, rb->volumetric_fog->height, 1); - RD::get_singleton()->compute_list_end(); + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_RASTER); + + RENDER_TIMESTAMP("<Volumetric Fog"); + RD::get_singleton()->draw_command_end_label(); + + rb->volumetric_fog->prev_cam_transform = p_cam_transform; } -void RendererSceneRenderRD::render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold) { - Color clear_color; - if (p_render_buffers.is_valid()) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(!rb); - clear_color = storage->render_target_get_clear_request_color(rb->render_target); - } else { - clear_color = storage->get_default_clear_color(); +uint32_t RendererSceneRenderRD::_get_render_state_directional_light_count() const { + return render_state.directional_light_count; +} + +bool RendererSceneRenderRD::_needs_post_prepass_render(bool p_use_gi) { + if (render_state.render_buffers.is_valid()) { + RenderBuffers *rb = render_buffers_owner.getornull(render_state.render_buffers); + if (rb->sdfgi != nullptr) { + return true; + } } + return false; +} - //assign render indices to giprobes - for (uint32_t i = 0; i < (uint32_t)p_gi_probes.size(); i++) { - GIProbeInstance *giprobe_inst = gi_probe_instance_owner.getornull(p_gi_probes[i]); - if (giprobe_inst) { - giprobe_inst->render_index = i; +void RendererSceneRenderRD::_post_prepass_render(bool p_use_gi) { + if (render_state.render_buffers.is_valid()) { + if (p_use_gi) { + RenderBuffers *rb = render_buffers_owner.getornull(render_state.render_buffers); + ERR_FAIL_COND(rb == nullptr); + if (rb->sdfgi == nullptr) { + return; + } + + RendererSceneEnvironmentRD *env = environment_owner.getornull(render_state.environment); + rb->sdfgi->update_probes(env, sky.sky_owner.getornull(env->sky)); } } +} - const PagedArray<RID> *lights = &p_lights; - const PagedArray<RID> *reflections = &p_reflection_probes; - const PagedArray<RID> *gi_probes = &p_gi_probes; +void RendererSceneRenderRD::_pre_resolve_render(bool p_use_gi) { + if (render_state.render_buffers.is_valid()) { + if (p_use_gi) { + RD::get_singleton()->compute_list_end(); + } + } +} - PagedArray<RID> empty; +void RendererSceneRenderRD::_pre_opaque_render(bool p_use_ssao, bool p_use_gi, RID p_normal_roughness_buffer, RID p_gi_probe_buffer) { + // Render shadows while GI is rendering, due to how barriers are handled, this should happen at the same time - if (get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_UNSHADED) { - lights = ∅ - reflections = ∅ - gi_probes = ∅ + if (render_state.render_buffers.is_valid() && p_use_gi) { + RenderBuffers *rb = render_buffers_owner.getornull(render_state.render_buffers); + ERR_FAIL_COND(rb == nullptr); + if (rb->sdfgi == nullptr) { + return; + } + + rb->sdfgi->store_probes(); + } + + render_state.cube_shadows.clear(); + render_state.shadows.clear(); + render_state.directional_shadows.clear(); + + Plane camera_plane(render_state.cam_transform.origin, -render_state.cam_transform.basis.get_axis(Vector3::AXIS_Z)); + float lod_distance_multiplier = render_state.cam_projection.get_lod_multiplier(); + + { + for (int i = 0; i < render_state.render_shadow_count; i++) { + LightInstance *li = light_instance_owner.getornull(render_state.render_shadows[i].light); + + if (storage->light_get_type(li->light) == RS::LIGHT_DIRECTIONAL) { + render_state.directional_shadows.push_back(i); + } else if (storage->light_get_type(li->light) == RS::LIGHT_OMNI && storage->light_omni_get_shadow_mode(li->light) == RS::LIGHT_OMNI_SHADOW_CUBE) { + render_state.cube_shadows.push_back(i); + } else { + render_state.shadows.push_back(i); + } + } + + //cube shadows are rendered in their own way + for (uint32_t i = 0; i < render_state.cube_shadows.size(); i++) { + _render_shadow_pass(render_state.render_shadows[render_state.cube_shadows[i]].light, render_state.shadow_atlas, render_state.render_shadows[render_state.cube_shadows[i]].pass, render_state.render_shadows[render_state.cube_shadows[i]].instances, camera_plane, lod_distance_multiplier, render_state.screen_lod_threshold, true, true, true); + } + + if (render_state.directional_shadows.size()) { + //open the pass for directional shadows + _update_directional_shadow_atlas(); + RD::get_singleton()->draw_list_begin(directional_shadow.fb, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_CONTINUE); + RD::get_singleton()->draw_list_end(); + } } - cluster.builder.begin(p_cam_transform.affine_inverse(), p_cam_projection); //prepare cluster + // Render GI + + bool render_shadows = render_state.directional_shadows.size() || render_state.shadows.size(); + bool render_gi = render_state.render_buffers.is_valid() && p_use_gi; + + if (render_shadows && render_gi) { + RENDER_TIMESTAMP("Render GI + Render Shadows (parallel)"); + } else if (render_shadows) { + RENDER_TIMESTAMP("Render Shadows"); + } else if (render_gi) { + RENDER_TIMESTAMP("Render GI"); + } + + //prepare shadow rendering + if (render_shadows) { + _render_shadow_begin(); + + //render directional shadows + for (uint32_t i = 0; i < render_state.directional_shadows.size(); i++) { + _render_shadow_pass(render_state.render_shadows[render_state.directional_shadows[i]].light, render_state.shadow_atlas, render_state.render_shadows[render_state.directional_shadows[i]].pass, render_state.render_shadows[render_state.directional_shadows[i]].instances, camera_plane, lod_distance_multiplier, render_state.screen_lod_threshold, false, i == render_state.directional_shadows.size() - 1, false); + } + //render positional shadows + for (uint32_t i = 0; i < render_state.shadows.size(); i++) { + _render_shadow_pass(render_state.render_shadows[render_state.shadows[i]].light, render_state.shadow_atlas, render_state.render_shadows[render_state.shadows[i]].pass, render_state.render_shadows[render_state.shadows[i]].instances, camera_plane, lod_distance_multiplier, render_state.screen_lod_threshold, i == 0, i == render_state.shadows.size() - 1, true); + } + + _render_shadow_process(); + } + + //start GI + if (render_gi) { + gi.process_gi(render_state.render_buffers, p_normal_roughness_buffer, p_gi_probe_buffer, render_state.environment, render_state.cam_projection, render_state.cam_transform, *render_state.gi_probes, this); + } + + //Do shadow rendering (in parallel with GI) + if (render_shadows) { + _render_shadow_end(RD::BARRIER_MASK_NO_BARRIER); + } + + if (render_gi) { + RD::get_singleton()->compute_list_end(RD::BARRIER_MASK_NO_BARRIER); //use a later barrier + } + + if (render_state.render_buffers.is_valid()) { + if (p_use_ssao) { + _process_ssao(render_state.render_buffers, render_state.environment, p_normal_roughness_buffer, render_state.cam_projection); + } + } + + //full barrier here, we need raster, transfer and compute and it depends from the previous work + RD::get_singleton()->barrier(RD::BARRIER_MASK_ALL, RD::BARRIER_MASK_ALL); + + if (current_cluster_builder) { + current_cluster_builder->begin(render_state.cam_transform, render_state.cam_projection, !render_state.reflection_probe.is_valid()); + } bool using_shadows = true; - if (p_reflection_probe.is_valid()) { - if (!storage->reflection_probe_renders_shadows(reflection_probe_instance_get_probe(p_reflection_probe))) { + if (render_state.reflection_probe.is_valid()) { + if (!storage->reflection_probe_renders_shadows(reflection_probe_instance_get_probe(render_state.reflection_probe))) { using_shadows = false; } } else { //do not render reflections when rendering a reflection probe - _setup_reflections(*reflections, p_cam_transform.affine_inverse(), p_environment); + _setup_reflections(*render_state.reflection_probes, render_state.cam_transform.affine_inverse(), render_state.environment); } uint32_t directional_light_count = 0; uint32_t positional_light_count = 0; - _setup_lights(*lights, p_cam_transform.affine_inverse(), p_shadow_atlas, using_shadows, directional_light_count, positional_light_count); - _setup_decals(p_decals, p_cam_transform.affine_inverse()); - cluster.builder.bake_cluster(); //bake to cluster + _setup_lights(*render_state.lights, render_state.cam_transform, render_state.shadow_atlas, using_shadows, directional_light_count, positional_light_count); + _setup_decals(*render_state.decals, render_state.cam_transform.affine_inverse()); - uint32_t gi_probe_count = 0; - _setup_giprobes(p_render_buffers, p_cam_transform, *gi_probes, gi_probe_count); + render_state.directional_light_count = directional_light_count; - if (p_render_buffers.is_valid()) { + if (current_cluster_builder) { + current_cluster_builder->bake_cluster(); + } + + if (render_state.render_buffers.is_valid()) { bool directional_shadows = false; for (uint32_t i = 0; i < directional_light_count; i++) { if (cluster.directional_lights[i].shadow_enabled) { @@ -7171,43 +3518,166 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const Transform & break; } } - _update_volumetric_fog(p_render_buffers, p_environment, p_cam_projection, p_cam_transform, p_shadow_atlas, directional_light_count, directional_shadows, positional_light_count, gi_probe_count); + if (is_volumetric_supported()) { + _update_volumetric_fog(render_state.render_buffers, render_state.environment, render_state.cam_projection, render_state.cam_transform, render_state.shadow_atlas, directional_light_count, directional_shadows, positional_light_count, render_state.gi_probe_count); + } + } +} + +void RendererSceneRenderRD::render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold, const RenderShadowData *p_render_shadows, int p_render_shadow_count, const RenderSDFGIData *p_render_sdfgi_regions, int p_render_sdfgi_region_count, const RenderSDFGIUpdateData *p_sdfgi_update_data) { + // getting this here now so we can direct call a bunch of things more easily + RenderBuffers *rb = nullptr; + if (p_render_buffers.is_valid()) { + rb = render_buffers_owner.getornull(p_render_buffers); + ERR_FAIL_COND(!rb); // !BAS! Do we fail here or skip the parts that won't work. can't really see a case why we would be rendering without buffers.... } - _render_scene(p_render_buffers, p_cam_transform, p_cam_projection, p_cam_ortogonal, p_instances, directional_light_count, *gi_probes, p_lightmaps, p_environment, p_camera_effects, p_shadow_atlas, p_reflection_atlas, p_reflection_probe, p_reflection_probe_pass, clear_color, p_screen_lod_threshold); + //assign render data + { + render_state.render_buffers = p_render_buffers; + render_state.cam_transform = p_cam_transform; + render_state.cam_projection = p_cam_projection; + render_state.cam_ortogonal = p_cam_projection.is_orthogonal(); + render_state.instances = &p_instances; + render_state.lights = &p_lights; + render_state.reflection_probes = &p_reflection_probes; + render_state.gi_probes = &p_gi_probes; + render_state.decals = &p_decals; + render_state.lightmaps = &p_lightmaps; + render_state.environment = p_environment; + render_state.camera_effects = p_camera_effects; + render_state.shadow_atlas = p_shadow_atlas; + render_state.reflection_atlas = p_reflection_atlas; + render_state.reflection_probe = p_reflection_probe; + render_state.reflection_probe_pass = p_reflection_probe_pass; + render_state.screen_lod_threshold = p_screen_lod_threshold; + + render_state.render_shadows = p_render_shadows; + render_state.render_shadow_count = p_render_shadow_count; + render_state.render_sdfgi_regions = p_render_sdfgi_regions; + render_state.render_sdfgi_region_count = p_render_sdfgi_region_count; + render_state.sdfgi_update_data = p_sdfgi_update_data; + } + + PagedArray<RID> empty; + if (get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_UNSHADED) { + render_state.lights = ∅ + render_state.reflection_probes = ∅ + render_state.gi_probes = ∅ + } + + //sdfgi first + if (rb != nullptr && rb->sdfgi != nullptr) { + for (int i = 0; i < render_state.render_sdfgi_region_count; i++) { + rb->sdfgi->render_region(p_render_buffers, render_state.render_sdfgi_regions[i].region, render_state.render_sdfgi_regions[i].instances, this); + } + if (render_state.sdfgi_update_data->update_static) { + rb->sdfgi->render_static_lights(p_render_buffers, render_state.sdfgi_update_data->static_cascade_count, p_sdfgi_update_data->static_cascade_indices, render_state.sdfgi_update_data->static_positional_lights, this); + } + } + + Color clear_color; if (p_render_buffers.is_valid()) { + clear_color = storage->render_target_get_clear_request_color(rb->render_target); + } else { + clear_color = storage->get_default_clear_color(); + } + + //assign render indices to giprobes + for (uint32_t i = 0; i < (uint32_t)p_gi_probes.size(); i++) { + RendererSceneGIRD::GIProbeInstance *giprobe_inst = gi.gi_probe_instance_owner.getornull(p_gi_probes[i]); + if (giprobe_inst) { + giprobe_inst->render_index = i; + } + } + + if (render_buffers_owner.owns(render_state.render_buffers)) { + // render_state.render_buffers == p_render_buffers so we can use our already retrieved rb + current_cluster_builder = rb->cluster_builder; + } else if (reflection_probe_instance_owner.owns(render_state.reflection_probe)) { + ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(render_state.reflection_probe); + ReflectionAtlas *ra = reflection_atlas_owner.getornull(rpi->atlas); + if (!ra) { + ERR_PRINT("reflection probe has no reflection atlas! Bug?"); + current_cluster_builder = nullptr; + } else { + current_cluster_builder = ra->cluster_builder; + } + } else { + ERR_PRINT("No render buffer nor reflection atlas, bug"); //should never happen, will crash + current_cluster_builder = nullptr; + } + + if (rb != nullptr && rb->sdfgi != nullptr) { + rb->sdfgi->update_cascades(); + + rb->sdfgi->pre_process_gi(p_cam_transform, this); + } + + render_state.gi_probe_count = 0; + if (rb != nullptr && rb->sdfgi != nullptr) { + gi.setup_giprobes(render_state.render_buffers, render_state.cam_transform, *render_state.gi_probes, render_state.gi_probe_count, this); + + rb->sdfgi->update_light(); + } + + render_state.depth_prepass_used = false; + //calls _pre_opaque_render between depth pre-pass and opaque pass + _render_scene(p_render_buffers, p_cam_transform, p_cam_projection, p_cam_ortogonal, p_instances, *render_state.gi_probes, p_lightmaps, p_environment, current_cluster_builder->get_cluster_buffer(), current_cluster_builder->get_cluster_size(), current_cluster_builder->get_max_cluster_elements(), p_camera_effects, p_shadow_atlas, p_reflection_atlas, p_reflection_probe, p_reflection_probe_pass, clear_color, p_screen_lod_threshold); + + if (p_render_buffers.is_valid()) { + if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS || debug_draw == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES) { + ClusterBuilderRD::ElementType elem_type = ClusterBuilderRD::ELEMENT_TYPE_MAX; + switch (debug_draw) { + case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS: + elem_type = ClusterBuilderRD::ELEMENT_TYPE_OMNI_LIGHT; + break; + case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS: + elem_type = ClusterBuilderRD::ELEMENT_TYPE_SPOT_LIGHT; + break; + case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS: + elem_type = ClusterBuilderRD::ELEMENT_TYPE_DECAL; + break; + case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES: + elem_type = ClusterBuilderRD::ELEMENT_TYPE_REFLECTION_PROBE; + break; + default: { + } + } + current_cluster_builder->debug(elem_type); + } + RENDER_TIMESTAMP("Tonemap"); _render_buffers_post_process_and_tonemap(p_render_buffers, p_environment, p_camera_effects, p_cam_projection); _render_buffers_debug_draw(p_render_buffers, p_shadow_atlas); - if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI) { - _sdfgi_debug_draw(p_render_buffers, p_cam_projection, p_cam_transform); + if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI && rb != nullptr && rb->sdfgi != nullptr) { + rb->sdfgi->debug_draw(p_cam_projection, p_cam_transform, rb->width, rb->height, rb->render_target, rb->texture); } } } -void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold) { +void RendererSceneRenderRD::_render_shadow_pass(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane, float p_lod_distance_multiplier, float p_screen_lod_threshold, bool p_open_pass, bool p_close_pass, bool p_clear_region) { LightInstance *light_instance = light_instance_owner.getornull(p_light); ERR_FAIL_COND(!light_instance); Rect2i atlas_rect; - RID atlas_texture; + uint32_t atlas_size; + RID atlas_fb; bool using_dual_paraboloid = false; bool using_dual_paraboloid_flip = false; - float znear = 0; - float zfar = 0; RID render_fb; RID render_texture; - float bias = 0; - float normal_bias = 0; + float zfar; bool use_pancake = false; - bool use_linear_depth = false; bool render_cubemap = false; bool finalize_cubemap = false; + bool flip_y = false; + CameraMatrix light_projection; Transform light_transform; @@ -7240,7 +3710,6 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p atlas_rect.position.x += atlas_rect.size.width; atlas_rect.position.y += atlas_rect.size.height; } - } else if (storage->light_directional_get_shadow_mode(light_instance->light) == RS::LIGHT_DIRECTIONAL_SHADOW_PARALLEL_2_SPLITS) { atlas_rect.size.height /= 2; @@ -7255,15 +3724,11 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p light_instance->shadow_transform[p_pass].atlas_rect.position /= directional_shadow.size; light_instance->shadow_transform[p_pass].atlas_rect.size /= directional_shadow.size; - float bias_mult = light_instance->shadow_transform[p_pass].bias_scale; zfar = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_RANGE); - bias = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_SHADOW_BIAS) * bias_mult; - normal_bias = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS) * bias_mult; - ShadowMap *shadow_map = _get_shadow_map(atlas_rect.size); - render_fb = shadow_map->fb; - render_texture = shadow_map->depth; - atlas_texture = directional_shadow.depth; + render_fb = directional_shadow.fb; + render_texture = RID(); + flip_y = true; } else { //set from shadow atlas @@ -7272,6 +3737,8 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p ERR_FAIL_COND(!shadow_atlas); ERR_FAIL_COND(!shadow_atlas->shadow_owners.has(p_light)); + _update_shadow_atlas(shadow_atlas); + uint32_t key = shadow_atlas->shadow_owners[p_light]; uint32_t quadrant = (key >> ShadowAtlas::QUADRANT_SHIFT) & 0x3; @@ -7290,11 +3757,8 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p atlas_rect.size.width = shadow_size; atlas_rect.size.height = shadow_size; - atlas_texture = shadow_atlas->depth; zfar = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_RANGE); - bias = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_SHADOW_BIAS); - normal_bias = storage->light_get_param(light_instance->light, RS::LIGHT_PARAM_SHADOW_NORMAL_BIAS); if (storage->light_get_type(light_instance->light) == RS::LIGHT_OMNI) { if (storage->light_omni_get_shadow_mode(light_instance->light) == RS::LIGHT_OMNI_SHADOW_CUBE) { @@ -7303,10 +3767,17 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p render_fb = cubemap->side_fb[p_pass]; render_texture = cubemap->cubemap; - light_projection = light_instance->shadow_transform[0].camera; - light_transform = light_instance->shadow_transform[0].transform; + light_projection = light_instance->shadow_transform[p_pass].camera; + light_transform = light_instance->shadow_transform[p_pass].transform; render_cubemap = true; finalize_cubemap = p_pass == 5; + atlas_fb = shadow_atlas->fb; + + atlas_size = shadow_atlas->size; + + if (p_pass == 0) { + _render_shadow_begin(); + } } else { light_projection = light_instance->shadow_transform[0].camera; @@ -7317,49 +3788,44 @@ void RendererSceneRenderRD::render_shadow(RID p_light, RID p_shadow_atlas, int p using_dual_paraboloid = true; using_dual_paraboloid_flip = p_pass == 1; - - ShadowMap *shadow_map = _get_shadow_map(atlas_rect.size); - render_fb = shadow_map->fb; - render_texture = shadow_map->depth; + render_fb = shadow_atlas->fb; + flip_y = true; } } else if (storage->light_get_type(light_instance->light) == RS::LIGHT_SPOT) { light_projection = light_instance->shadow_transform[0].camera; light_transform = light_instance->shadow_transform[0].transform; - ShadowMap *shadow_map = _get_shadow_map(atlas_rect.size); - render_fb = shadow_map->fb; - render_texture = shadow_map->depth; + render_fb = shadow_atlas->fb; - znear = light_instance->shadow_transform[0].camera.get_z_near(); - use_linear_depth = true; + flip_y = true; } } if (render_cubemap) { //rendering to cubemap - _render_shadow(render_fb, p_instances, light_projection, light_transform, zfar, 0, 0, false, false, use_pancake, p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold); + _render_shadow_append(render_fb, p_instances, light_projection, light_transform, zfar, 0, 0, false, false, use_pancake, p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold, Rect2(), false, true, true, true); if (finalize_cubemap) { + _render_shadow_process(); + _render_shadow_end(); //reblit - atlas_rect.size.height /= 2; - storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_texture, atlas_rect, light_projection.get_z_near(), light_projection.get_z_far(), 0.0, false); - atlas_rect.position.y += atlas_rect.size.height; - storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_texture, atlas_rect, light_projection.get_z_near(), light_projection.get_z_far(), 0.0, true); - } - } else { - //render shadow - - _render_shadow(render_fb, p_instances, light_projection, light_transform, zfar, bias, normal_bias, using_dual_paraboloid, using_dual_paraboloid_flip, use_pancake, p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold); + Rect2 atlas_rect_norm = atlas_rect; + atlas_rect_norm.position.x /= float(atlas_size); + atlas_rect_norm.position.y /= float(atlas_size); + atlas_rect_norm.size.x /= float(atlas_size); + atlas_rect_norm.size.y /= float(atlas_size); + atlas_rect_norm.size.height /= 2; + storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, light_projection.get_z_near(), light_projection.get_z_far(), false); + atlas_rect_norm.position.y += atlas_rect_norm.size.height; + storage->get_effects()->copy_cubemap_to_dp(render_texture, atlas_fb, atlas_rect_norm, light_projection.get_z_near(), light_projection.get_z_far(), true); - //copy to atlas - if (use_linear_depth) { - storage->get_effects()->copy_depth_to_rect_and_linearize(render_texture, atlas_texture, atlas_rect, true, znear, zfar); - } else { - storage->get_effects()->copy_depth_to_rect(render_texture, atlas_texture, atlas_rect, true); + //restore transform so it can be properly used + light_instance_set_shadow_transform(p_light, CameraMatrix(), light_instance->transform, zfar, 0, 0, 0); } - //does not work from depth to color - //RD::get_singleton()->texture_copy(render_texture, atlas_texture, Vector3(0, 0, 0), Vector3(atlas_rect.position.x, atlas_rect.position.y, 0), Vector3(atlas_rect.size.x, atlas_rect.size.y, 1), 0, 0, 0, 0, true); + } else { + //render shadow + _render_shadow_append(render_fb, p_instances, light_projection, light_transform, zfar, 0, 0, using_dual_paraboloid, using_dual_paraboloid_flip, use_pancake, p_camera_plane, p_lod_distance_multiplier, p_screen_lod_threshold, atlas_rect, flip_y, p_clear_region, p_open_pass, p_close_pass); } } @@ -7367,343 +3833,6 @@ void RendererSceneRenderRD::render_material(const Transform &p_cam_transform, co _render_material(p_cam_transform, p_cam_projection, p_cam_ortogonal, p_instances, p_framebuffer, p_region); } -void RendererSceneRenderRD::render_sdfgi(RID p_render_buffers, int p_region, const PagedArray<GeometryInstance *> &p_instances) { - //print_line("rendering region " + itos(p_region)); - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(!rb); - ERR_FAIL_COND(!rb->sdfgi); - AABB bounds; - Vector3i from; - Vector3i size; - - int cascade_prev = _sdfgi_get_pending_region_data(p_render_buffers, p_region - 1, from, size, bounds); - int cascade_next = _sdfgi_get_pending_region_data(p_render_buffers, p_region + 1, from, size, bounds); - int cascade = _sdfgi_get_pending_region_data(p_render_buffers, p_region, from, size, bounds); - ERR_FAIL_COND(cascade < 0); - - if (cascade_prev != cascade) { - //initialize render - RD::get_singleton()->texture_clear(rb->sdfgi->render_albedo, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - RD::get_singleton()->texture_clear(rb->sdfgi->render_emission, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - RD::get_singleton()->texture_clear(rb->sdfgi->render_emission_aniso, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - RD::get_singleton()->texture_clear(rb->sdfgi->render_geom_facing, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - } - - //print_line("rendering cascade " + itos(p_region) + " objects: " + itos(p_cull_count) + " bounds: " + bounds + " from: " + from + " size: " + size + " cell size: " + rtos(rb->sdfgi->cascades[cascade].cell_size)); - _render_sdfgi(p_render_buffers, from, size, bounds, p_instances, rb->sdfgi->render_albedo, rb->sdfgi->render_emission, rb->sdfgi->render_emission_aniso, rb->sdfgi->render_geom_facing); - - if (cascade_next != cascade) { - RENDER_TIMESTAMP(">SDFGI Update SDF"); - //done rendering! must update SDF - //clear dispatch indirect data - - SDGIShader::PreprocessPushConstant push_constant; - zeromem(&push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - - RENDER_TIMESTAMP("Scroll SDF"); - - //scroll - if (rb->sdfgi->cascades[cascade].dirty_regions != SDFGI::Cascade::DIRTY_ALL) { - //for scroll - Vector3i dirty = rb->sdfgi->cascades[cascade].dirty_regions; - push_constant.scroll[0] = dirty.x; - push_constant.scroll[1] = dirty.y; - push_constant.scroll[2] = dirty.z; - } else { - //for no scroll - push_constant.scroll[0] = 0; - push_constant.scroll[1] = 0; - push_constant.scroll[2] = 0; - } - push_constant.grid_size = rb->sdfgi->cascade_size; - push_constant.cascade = cascade; - - if (rb->sdfgi->cascades[cascade].dirty_regions != SDFGI::Cascade::DIRTY_ALL) { - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - - //must pre scroll existing data because not all is dirty - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_SCROLL]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[cascade].scroll_uniform_set, 0); - - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_indirect(compute_list, rb->sdfgi->cascades[cascade].solid_cell_dispatch_buffer, 0); - // no barrier do all together - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_SCROLL_OCCLUSION]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[cascade].scroll_occlusion_uniform_set, 0); - - Vector3i dirty = rb->sdfgi->cascades[cascade].dirty_regions; - Vector3i groups; - groups.x = rb->sdfgi->cascade_size - ABS(dirty.x); - groups.y = rb->sdfgi->cascade_size - ABS(dirty.y); - groups.z = rb->sdfgi->cascade_size - ABS(dirty.z); - - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, groups.x, groups.y, groups.z, 4, 4, 4); - - //no barrier, continue together - - { - //scroll probes and their history also - - SDGIShader::IntegratePushConstant ipush_constant; - ipush_constant.grid_size[1] = rb->sdfgi->cascade_size; - ipush_constant.grid_size[2] = rb->sdfgi->cascade_size; - ipush_constant.grid_size[0] = rb->sdfgi->cascade_size; - ipush_constant.max_cascades = rb->sdfgi->cascades.size(); - ipush_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - ipush_constant.history_index = 0; - ipush_constant.history_size = rb->sdfgi->history_size; - ipush_constant.ray_count = 0; - ipush_constant.ray_bias = 0; - ipush_constant.sky_mode = 0; - ipush_constant.sky_energy = 0; - ipush_constant.sky_color[0] = 0; - ipush_constant.sky_color[1] = 0; - ipush_constant.sky_color[2] = 0; - ipush_constant.y_mult = rb->sdfgi->y_mult; - ipush_constant.store_ambient_texture = false; - - ipush_constant.image_size[0] = rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count; - ipush_constant.image_size[1] = rb->sdfgi->probe_axis_count; - - int32_t probe_divisor = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - ipush_constant.cascade = cascade; - ipush_constant.world_offset[0] = rb->sdfgi->cascades[cascade].position.x / probe_divisor; - ipush_constant.world_offset[1] = rb->sdfgi->cascades[cascade].position.y / probe_divisor; - ipush_constant.world_offset[2] = rb->sdfgi->cascades[cascade].position.z / probe_divisor; - - ipush_constant.scroll[0] = dirty.x / probe_divisor; - ipush_constant.scroll[1] = dirty.y / probe_divisor; - ipush_constant.scroll[2] = dirty.z / probe_divisor; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.integrate_pipeline[SDGIShader::INTEGRATE_MODE_SCROLL]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[cascade].integrate_uniform_set, 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sdfgi_shader.integrate_default_sky_uniform_set, 1); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &ipush_constant, sizeof(SDGIShader::IntegratePushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count, rb->sdfgi->probe_axis_count, 1, 8, 8, 1); - - RD::get_singleton()->compute_list_add_barrier(compute_list); - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.integrate_pipeline[SDGIShader::INTEGRATE_MODE_SCROLL_STORE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[cascade].integrate_uniform_set, 0); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, sdfgi_shader.integrate_default_sky_uniform_set, 1); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &ipush_constant, sizeof(SDGIShader::IntegratePushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->probe_axis_count * rb->sdfgi->probe_axis_count, rb->sdfgi->probe_axis_count, 1, 8, 8, 1); - } - - //ok finally barrier - RD::get_singleton()->compute_list_end(); - } - - //clear dispatch indirect data - uint32_t dispatch_indirct_data[4] = { 0, 0, 0, 0 }; - RD::get_singleton()->buffer_update(rb->sdfgi->cascades[cascade].solid_cell_dispatch_buffer, 0, sizeof(uint32_t) * 4, dispatch_indirct_data, true); - - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - - bool half_size = true; //much faster, very little difference - static const int optimized_jf_group_size = 8; - - if (half_size) { - push_constant.grid_size >>= 1; - - uint32_t cascade_half_size = rb->sdfgi->cascade_size >> 1; - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->sdf_initialize_half_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size, 4, 4, 4); - RD::get_singleton()->compute_list_add_barrier(compute_list); - - //must start with regular jumpflood - - push_constant.half_size = true; - { - RENDER_TIMESTAMP("SDFGI Jump Flood (Half Size)"); - - uint32_t s = cascade_half_size; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD]); - - int jf_us = 0; - //start with regular jump flood for very coarse reads, as this is impossible to optimize - while (s > 1) { - s /= 2; - push_constant.step_size = s; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->jump_flood_half_uniform_set[jf_us], 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size, 4, 4, 4); - RD::get_singleton()->compute_list_add_barrier(compute_list); - jf_us = jf_us == 0 ? 1 : 0; - - if (cascade_half_size / (s / 2) >= optimized_jf_group_size) { - break; - } - } - - RENDER_TIMESTAMP("SDFGI Jump Flood Optimized (Half Size)"); - - //continue with optimized jump flood for smaller reads - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); - while (s > 1) { - s /= 2; - push_constant.step_size = s; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->jump_flood_half_uniform_set[jf_us], 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, cascade_half_size, cascade_half_size, cascade_half_size, optimized_jf_group_size, optimized_jf_group_size, optimized_jf_group_size); - RD::get_singleton()->compute_list_add_barrier(compute_list); - jf_us = jf_us == 0 ? 1 : 0; - } - } - - // restore grid size for last passes - push_constant.grid_size = rb->sdfgi->cascade_size; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_UPSCALE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->sdf_upscale_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, 4, 4, 4); - RD::get_singleton()->compute_list_add_barrier(compute_list); - - //run one pass of fullsize jumpflood to fix up half size arctifacts - - push_constant.half_size = false; - push_constant.step_size = 1; - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->jump_flood_uniform_set[rb->sdfgi->upscale_jfa_uniform_set_index], 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, optimized_jf_group_size, optimized_jf_group_size, optimized_jf_group_size); - RD::get_singleton()->compute_list_add_barrier(compute_list); - - } else { - //full size jumpflood - RENDER_TIMESTAMP("SDFGI Jump Flood"); - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_INITIALIZE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->sdf_initialize_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, 4, 4, 4); - - RD::get_singleton()->compute_list_add_barrier(compute_list); - - push_constant.half_size = false; - { - uint32_t s = rb->sdfgi->cascade_size; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD]); - - int jf_us = 0; - //start with regular jump flood for very coarse reads, as this is impossible to optimize - while (s > 1) { - s /= 2; - push_constant.step_size = s; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->jump_flood_uniform_set[jf_us], 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, 4, 4, 4); - RD::get_singleton()->compute_list_add_barrier(compute_list); - jf_us = jf_us == 0 ? 1 : 0; - - if (rb->sdfgi->cascade_size / (s / 2) >= optimized_jf_group_size) { - break; - } - } - - RENDER_TIMESTAMP("SDFGI Jump Flood Optimized"); - - //continue with optimized jump flood for smaller reads - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_JUMP_FLOOD_OPTIMIZED]); - while (s > 1) { - s /= 2; - push_constant.step_size = s; - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->jump_flood_uniform_set[jf_us], 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, optimized_jf_group_size, optimized_jf_group_size, optimized_jf_group_size); - RD::get_singleton()->compute_list_add_barrier(compute_list); - jf_us = jf_us == 0 ? 1 : 0; - } - } - } - - RENDER_TIMESTAMP("SDFGI Occlusion"); - - // occlusion - { - uint32_t probe_size = rb->sdfgi->cascade_size / SDFGI::PROBE_DIVISOR; - Vector3i probe_global_pos = rb->sdfgi->cascades[cascade].position / probe_size; - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_OCCLUSION]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->occlusion_uniform_set, 0); - for (int i = 0; i < 8; i++) { - //dispatch all at once for performance - Vector3i offset(i & 1, (i >> 1) & 1, (i >> 2) & 1); - - if ((probe_global_pos.x & 1) != 0) { - offset.x = (offset.x + 1) & 1; - } - if ((probe_global_pos.y & 1) != 0) { - offset.y = (offset.y + 1) & 1; - } - if ((probe_global_pos.z & 1) != 0) { - offset.z = (offset.z + 1) & 1; - } - push_constant.probe_offset[0] = offset.x; - push_constant.probe_offset[1] = offset.y; - push_constant.probe_offset[2] = offset.z; - push_constant.occlusion_index = i; - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - - Vector3i groups = Vector3i(probe_size + 1, probe_size + 1, probe_size + 1) - offset; //if offset, it's one less probe per axis to compute - RD::get_singleton()->compute_list_dispatch(compute_list, groups.x, groups.y, groups.z); - } - RD::get_singleton()->compute_list_add_barrier(compute_list); - } - - RENDER_TIMESTAMP("SDFGI Store"); - - // store - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.preprocess_pipeline[SDGIShader::PRE_PROCESS_STORE]); - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rb->sdfgi->cascades[cascade].sdf_store_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SDGIShader::PreprocessPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, 4, 4, 4); - - RD::get_singleton()->compute_list_end(); - - //clear these textures, as they will have previous garbage on next draw - RD::get_singleton()->texture_clear(rb->sdfgi->cascades[cascade].light_tex, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - RD::get_singleton()->texture_clear(rb->sdfgi->cascades[cascade].light_aniso_0_tex, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - RD::get_singleton()->texture_clear(rb->sdfgi->cascades[cascade].light_aniso_1_tex, Color(0, 0, 0, 0), 0, 1, 0, 1, true); - -#if 0 - Vector<uint8_t> data = RD::get_singleton()->texture_get_data(rb->sdfgi->cascades[cascade].sdf, 0); - Ref<Image> img; - img.instance(); - for (uint32_t i = 0; i < rb->sdfgi->cascade_size; i++) { - Vector<uint8_t> subarr = data.subarray(128 * 128 * i, 128 * 128 * (i + 1) - 1); - img->create(rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, false, Image::FORMAT_L8, subarr); - img->save_png("res://cascade_sdf_" + itos(cascade) + "_" + itos(i) + ".png"); - } - - //finalize render and update sdf -#endif - -#if 0 - Vector<uint8_t> data = RD::get_singleton()->texture_get_data(rb->sdfgi->render_albedo, 0); - Ref<Image> img; - img.instance(); - for (uint32_t i = 0; i < rb->sdfgi->cascade_size; i++) { - Vector<uint8_t> subarr = data.subarray(128 * 128 * i * 2, 128 * 128 * (i + 1) * 2 - 1); - img->create(rb->sdfgi->cascade_size, rb->sdfgi->cascade_size, false, Image::FORMAT_RGB565, subarr); - img->convert(Image::FORMAT_RGBA8); - img->save_png("res://cascade_" + itos(cascade) + "_" + itos(i) + ".png"); - } - - //finalize render and update sdf -#endif - - RENDER_TIMESTAMP("<SDFGI Update SDF"); - } -} - void RendererSceneRenderRD::render_particle_collider_heightfield(RID p_collider, const Transform &p_transform, const PagedArray<GeometryInstance *> &p_instances) { ERR_FAIL_COND(!storage->particles_collision_is_heightfield(p_collider)); Vector3 extents = storage->particles_collision_get_extents(p_collider) * p_transform.basis.get_scale(); @@ -7721,122 +3850,22 @@ void RendererSceneRenderRD::render_particle_collider_heightfield(RID p_collider, _render_particle_collider_heightfield(fb, cam_xform, cm, p_instances); } -void RendererSceneRenderRD::render_sdfgi_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_light_cull_result) { - RenderBuffers *rb = render_buffers_owner.getornull(p_render_buffers); - ERR_FAIL_COND(!rb); - ERR_FAIL_COND(!rb->sdfgi); - - _sdfgi_update_cascades(p_render_buffers); //need cascades updated for this - - RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, sdfgi_shader.direct_light_pipeline[SDGIShader::DIRECT_LIGHT_MODE_STATIC]); - - SDGIShader::DirectLightPushConstant dl_push_constant; - - dl_push_constant.grid_size[0] = rb->sdfgi->cascade_size; - dl_push_constant.grid_size[1] = rb->sdfgi->cascade_size; - dl_push_constant.grid_size[2] = rb->sdfgi->cascade_size; - dl_push_constant.max_cascades = rb->sdfgi->cascades.size(); - dl_push_constant.probe_axis_size = rb->sdfgi->probe_axis_count; - dl_push_constant.multibounce = false; // this is static light, do not multibounce yet - dl_push_constant.y_mult = rb->sdfgi->y_mult; - - //all must be processed - dl_push_constant.process_offset = 0; - dl_push_constant.process_increment = 1; - - SDGIShader::Light lights[SDFGI::MAX_STATIC_LIGHTS]; - - for (uint32_t i = 0; i < p_cascade_count; i++) { - ERR_CONTINUE(p_cascade_indices[i] >= rb->sdfgi->cascades.size()); - - SDFGI::Cascade &cc = rb->sdfgi->cascades[p_cascade_indices[i]]; - - { //fill light buffer - - AABB cascade_aabb; - cascade_aabb.position = Vector3((Vector3i(1, 1, 1) * -int32_t(rb->sdfgi->cascade_size >> 1) + cc.position)) * cc.cell_size; - cascade_aabb.size = Vector3(1, 1, 1) * rb->sdfgi->cascade_size * cc.cell_size; - - int idx = 0; - - for (uint32_t j = 0; j < (uint32_t)p_positional_light_cull_result[i].size(); j++) { - if (idx == SDFGI::MAX_STATIC_LIGHTS) { - break; - } - - LightInstance *li = light_instance_owner.getornull(p_positional_light_cull_result[i][j]); - ERR_CONTINUE(!li); - - uint32_t max_sdfgi_cascade = storage->light_get_max_sdfgi_cascade(li->light); - if (p_cascade_indices[i] > max_sdfgi_cascade) { - continue; - } - - if (!cascade_aabb.intersects(li->aabb)) { - continue; - } - - lights[idx].type = storage->light_get_type(li->light); - - Vector3 dir = -li->transform.basis.get_axis(Vector3::AXIS_Z); - if (lights[idx].type == RS::LIGHT_DIRECTIONAL) { - dir.y *= rb->sdfgi->y_mult; //only makes sense for directional - dir.normalize(); - } - lights[idx].direction[0] = dir.x; - lights[idx].direction[1] = dir.y; - lights[idx].direction[2] = dir.z; - Vector3 pos = li->transform.origin; - pos.y *= rb->sdfgi->y_mult; - lights[idx].position[0] = pos.x; - lights[idx].position[1] = pos.y; - lights[idx].position[2] = pos.z; - Color color = storage->light_get_color(li->light); - color = color.to_linear(); - lights[idx].color[0] = color.r; - lights[idx].color[1] = color.g; - lights[idx].color[2] = color.b; - lights[idx].energy = storage->light_get_param(li->light, RS::LIGHT_PARAM_ENERGY); - lights[idx].has_shadow = storage->light_has_shadow(li->light); - lights[idx].attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_ATTENUATION); - lights[idx].radius = storage->light_get_param(li->light, RS::LIGHT_PARAM_RANGE); - lights[idx].spot_angle = Math::deg2rad(storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ANGLE)); - lights[idx].spot_attenuation = storage->light_get_param(li->light, RS::LIGHT_PARAM_SPOT_ATTENUATION); - - idx++; - } - - if (idx > 0) { - RD::get_singleton()->buffer_update(cc.lights_buffer, 0, idx * sizeof(SDGIShader::Light), lights, true); - } - dl_push_constant.light_count = idx; - } - - dl_push_constant.cascade = p_cascade_indices[i]; - - if (dl_push_constant.light_count > 0) { - RD::get_singleton()->compute_list_bind_uniform_set(compute_list, cc.sdf_direct_light_uniform_set, 0); - RD::get_singleton()->compute_list_set_push_constant(compute_list, &dl_push_constant, sizeof(SDGIShader::DirectLightPushConstant)); - RD::get_singleton()->compute_list_dispatch_indirect(compute_list, cc.solid_cell_dispatch_buffer, 0); - } - } - - RD::get_singleton()->compute_list_end(); -} - bool RendererSceneRenderRD::free(RID p_rid) { if (render_buffers_owner.owns(p_rid)) { RenderBuffers *rb = render_buffers_owner.getornull(p_rid); _free_render_buffer_data(rb); memdelete(rb->data); if (rb->sdfgi) { - _sdfgi_erase(rb); + rb->sdfgi->erase(); + memdelete(rb->sdfgi); + rb->sdfgi = nullptr; } if (rb->volumetric_fog) { _volumetric_fog_erase(rb); } + if (rb->cluster_builder) { + memdelete(rb->cluster_builder); + } render_buffers_owner.free(p_rid); } else if (environment_owner.owns(p_rid)) { //not much to delete, just free it @@ -7846,6 +3875,10 @@ bool RendererSceneRenderRD::free(RID p_rid) { camera_effects_owner.free(p_rid); } else if (reflection_atlas_owner.owns(p_rid)) { reflection_atlas_set_size(p_rid, 0, 0); + ReflectionAtlas *ra = reflection_atlas_owner.getornull(p_rid); + if (ra->cluster_builder) { + memdelete(ra->cluster_builder); + } reflection_atlas_owner.free(p_rid); } else if (reflection_probe_instance_owner.owns(p_rid)) { //not much to delete, just free it @@ -7856,8 +3889,8 @@ bool RendererSceneRenderRD::free(RID p_rid) { decal_instance_owner.free(p_rid); } else if (lightmap_instance_owner.owns(p_rid)) { lightmap_instance_owner.free(p_rid); - } else if (gi_probe_instance_owner.owns(p_rid)) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_rid); + } else if (gi.gi_probe_instance_owner.owns(p_rid)) { + RendererSceneGIRD::GIProbeInstance *gi_probe = gi.gi_probe_instance_owner.getornull(p_rid); if (gi_probe->texture.is_valid()) { RD::get_singleton()->free(gi_probe->texture); RD::get_singleton()->free(gi_probe->write_buffer); @@ -7868,37 +3901,10 @@ bool RendererSceneRenderRD::free(RID p_rid) { RD::get_singleton()->free(gi_probe->dynamic_maps[i].depth); } - gi_probe_instance_owner.free(p_rid); - } else if (sky_owner.owns(p_rid)) { - _update_dirty_skys(); - Sky *sky = sky_owner.getornull(p_rid); - - if (sky->radiance.is_valid()) { - RD::get_singleton()->free(sky->radiance); - sky->radiance = RID(); - } - _clear_reflection_data(sky->reflection); - - if (sky->uniform_buffer.is_valid()) { - RD::get_singleton()->free(sky->uniform_buffer); - sky->uniform_buffer = RID(); - } - - if (sky->half_res_pass.is_valid()) { - RD::get_singleton()->free(sky->half_res_pass); - sky->half_res_pass = RID(); - } - - if (sky->quarter_res_pass.is_valid()) { - RD::get_singleton()->free(sky->quarter_res_pass); - sky->quarter_res_pass = RID(); - } - - if (sky->material.is_valid()) { - storage->free(sky->material); - } - - sky_owner.free(p_rid); + gi.gi_probe_instance_owner.free(p_rid); + } else if (sky.sky_owner.owns(p_rid)) { + sky.update_dirty_skys(); + sky.free_sky(p_rid); } else if (light_instance_owner.owns(p_rid)) { LightInstance *light_instance = light_instance_owner.getornull(p_rid); @@ -7932,7 +3938,7 @@ void RendererSceneRenderRD::set_debug_draw_mode(RS::ViewportDebugDraw p_debug_dr } void RendererSceneRenderRD::update() { - _update_dirty_skys(); + sky.update_dirty_skys(); } void RendererSceneRenderRD::set_time(double p_time, double p_step) { @@ -8058,26 +4064,23 @@ TypedArray<Image> RendererSceneRenderRD::bake_render_uv2(RID p_base, const Vecto } void RendererSceneRenderRD::sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir) { - sdfgi_debug_probe_pos = p_position; - sdfgi_debug_probe_dir = p_dir; + gi.sdfgi_debug_probe_pos = p_position; + gi.sdfgi_debug_probe_dir = p_dir; } RendererSceneRenderRD *RendererSceneRenderRD::singleton = nullptr; -RID RendererSceneRenderRD::get_cluster_builder_texture() { - return cluster.builder.get_cluster_texture(); -} - -RID RendererSceneRenderRD::get_cluster_builder_indices_buffer() { - return cluster.builder.get_cluster_indices_buffer(); -} - RID RendererSceneRenderRD::get_reflection_probe_buffer() { return cluster.reflection_buffer; } -RID RendererSceneRenderRD::get_positional_light_buffer() { - return cluster.light_buffer; +RID RendererSceneRenderRD::get_omni_light_buffer() { + return cluster.omni_light_buffer; +} + +RID RendererSceneRenderRD::get_spot_light_buffer() { + return cluster.spot_light_buffer; } + RID RendererSceneRenderRD::get_directional_light_buffer() { return cluster.directional_light_buffer; } @@ -8089,417 +4092,79 @@ int RendererSceneRenderRD::get_max_directional_lights() const { } bool RendererSceneRenderRD::is_low_end() const { - return low_end; + // by default we switch this on this (may be ignored in some implementations) + return GLOBAL_GET("rendering/driver/rd_renderer/use_low_end_renderer"); +} + +bool RendererSceneRenderRD::is_dynamic_gi_supported() const { + // usable by default (unless low end = true) + return true; +} + +bool RendererSceneRenderRD::is_clustered_enabled() const { + // used by default. + return true; +} + +bool RendererSceneRenderRD::is_volumetric_supported() const { + // usable by default (unless low end = true) + return true; } RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { + max_cluster_elements = GLOBAL_GET("rendering/limits/cluster_builder/max_clustered_elements"); + storage = p_storage; singleton = this; - roughness_layers = GLOBAL_GET("rendering/quality/reflections/roughness_layers"); - sky_ggx_samples_quality = GLOBAL_GET("rendering/quality/reflections/ggx_samples"); - sky_use_cubemap_array = GLOBAL_GET("rendering/quality/reflections/texture_array_reflections"); - // sky_use_cubemap_array = false; + directional_shadow.size = GLOBAL_GET("rendering/shadows/directional_shadow/size"); + directional_shadow.use_16_bits = GLOBAL_GET("rendering/shadows/directional_shadow/16_bits"); uint32_t textures_per_stage = RD::get_singleton()->limit_get(RD::LIMIT_MAX_TEXTURES_PER_SHADER_STAGE); - low_end = GLOBAL_GET("rendering/quality/rd_renderer/use_low_end_renderer"); + low_end = is_low_end(); if (textures_per_stage < 48) { low_end = true; } - if (!low_end) { - //kinda complicated to compute the amount of slots, we try to use as many as we can - - gi_probe_max_lights = 32; - - gi_probe_lights = memnew_arr(GIProbeLight, gi_probe_max_lights); - gi_probe_lights_uniform = RD::get_singleton()->uniform_buffer_create(gi_probe_max_lights * sizeof(GIProbeLight)); - gi_probe_quality = RS::GIProbeQuality(CLAMP(int(GLOBAL_GET("rendering/quality/gi_probes/quality")), 0, 1)); - - String defines = "\n#define MAX_LIGHTS " + itos(gi_probe_max_lights) + "\n"; - - Vector<String> versions; - versions.push_back("\n#define MODE_COMPUTE_LIGHT\n"); - versions.push_back("\n#define MODE_SECOND_BOUNCE\n"); - versions.push_back("\n#define MODE_UPDATE_MIPMAPS\n"); - versions.push_back("\n#define MODE_WRITE_TEXTURE\n"); - versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_LIGHTING\n"); - versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_WRITE\n"); - versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_PLOT\n"); - versions.push_back("\n#define MODE_DYNAMIC\n#define MODE_DYNAMIC_SHRINK\n#define MODE_DYNAMIC_SHRINK_PLOT\n#define MODE_DYNAMIC_SHRINK_WRITE\n"); - - giprobe_shader.initialize(versions, defines); - giprobe_lighting_shader_version = giprobe_shader.version_create(); - for (int i = 0; i < GI_PROBE_SHADER_VERSION_MAX; i++) { - giprobe_lighting_shader_version_shaders[i] = giprobe_shader.version_get_shader(giprobe_lighting_shader_version, i); - giprobe_lighting_shader_version_pipelines[i] = RD::get_singleton()->compute_pipeline_create(giprobe_lighting_shader_version_shaders[i]); - } - } - - if (!low_end) { - String defines; - Vector<String> versions; - versions.push_back("\n#define MODE_DEBUG_COLOR\n"); - versions.push_back("\n#define MODE_DEBUG_LIGHT\n"); - versions.push_back("\n#define MODE_DEBUG_EMISSION\n"); - versions.push_back("\n#define MODE_DEBUG_LIGHT\n#define MODE_DEBUG_LIGHT_FULL\n"); - - giprobe_debug_shader.initialize(versions, defines); - giprobe_debug_shader_version = giprobe_debug_shader.version_create(); - for (int i = 0; i < GI_PROBE_DEBUG_MAX; i++) { - giprobe_debug_shader_version_shaders[i] = giprobe_debug_shader.version_get_shader(giprobe_debug_shader_version, i); - - RD::PipelineRasterizationState rs; - rs.cull_mode = RD::POLYGON_CULL_FRONT; - RD::PipelineDepthStencilState ds; - ds.enable_depth_test = true; - ds.enable_depth_write = true; - ds.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; - - giprobe_debug_shader_version_pipelines[i].setup(giprobe_debug_shader_version_shaders[i], RD::RENDER_PRIMITIVE_TRIANGLES, rs, RD::PipelineMultisampleState(), ds, RD::PipelineColorBlendState::create_disabled(), 0); - } - } - /* SKY SHADER */ - { - // Start with the directional lights for the sky - sky_scene_state.max_directional_lights = 4; - uint32_t directional_light_buffer_size = sky_scene_state.max_directional_lights * sizeof(SkyDirectionalLightData); - sky_scene_state.directional_lights = memnew_arr(SkyDirectionalLightData, sky_scene_state.max_directional_lights); - sky_scene_state.last_frame_directional_lights = memnew_arr(SkyDirectionalLightData, sky_scene_state.max_directional_lights); - sky_scene_state.last_frame_directional_light_count = sky_scene_state.max_directional_lights + 1; - sky_scene_state.directional_light_buffer = RD::get_singleton()->uniform_buffer_create(directional_light_buffer_size); - - String defines = "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(sky_scene_state.max_directional_lights) + "\n"; - - // Initialize sky - Vector<String> sky_modes; - sky_modes.push_back(""); // Full size - sky_modes.push_back("\n#define USE_HALF_RES_PASS\n"); // Half Res - sky_modes.push_back("\n#define USE_QUARTER_RES_PASS\n"); // Quarter res - sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n"); // Cubemap - sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n#define USE_HALF_RES_PASS\n"); // Half Res Cubemap - sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n#define USE_QUARTER_RES_PASS\n"); // Quarter res Cubemap - sky_shader.shader.initialize(sky_modes, defines); - } - - // register our shader funds - storage->shader_set_data_request_function(RendererStorageRD::SHADER_TYPE_SKY, _create_sky_shader_funcs); - storage->material_set_data_request_function(RendererStorageRD::SHADER_TYPE_SKY, _create_sky_material_funcs); + sky.init(storage); - { - ShaderCompilerRD::DefaultIdentifierActions actions; - - actions.renames["COLOR"] = "color"; - actions.renames["ALPHA"] = "alpha"; - actions.renames["EYEDIR"] = "cube_normal"; - actions.renames["POSITION"] = "params.position_multiplier.xyz"; - actions.renames["SKY_COORDS"] = "panorama_coords"; - actions.renames["SCREEN_UV"] = "uv"; - actions.renames["TIME"] = "params.time"; - actions.renames["HALF_RES_COLOR"] = "half_res_color"; - actions.renames["QUARTER_RES_COLOR"] = "quarter_res_color"; - actions.renames["RADIANCE"] = "radiance"; - actions.renames["FOG"] = "custom_fog"; - actions.renames["LIGHT0_ENABLED"] = "directional_lights.data[0].enabled"; - actions.renames["LIGHT0_DIRECTION"] = "directional_lights.data[0].direction_energy.xyz"; - actions.renames["LIGHT0_ENERGY"] = "directional_lights.data[0].direction_energy.w"; - actions.renames["LIGHT0_COLOR"] = "directional_lights.data[0].color_size.xyz"; - actions.renames["LIGHT0_SIZE"] = "directional_lights.data[0].color_size.w"; - actions.renames["LIGHT1_ENABLED"] = "directional_lights.data[1].enabled"; - actions.renames["LIGHT1_DIRECTION"] = "directional_lights.data[1].direction_energy.xyz"; - actions.renames["LIGHT1_ENERGY"] = "directional_lights.data[1].direction_energy.w"; - actions.renames["LIGHT1_COLOR"] = "directional_lights.data[1].color_size.xyz"; - actions.renames["LIGHT1_SIZE"] = "directional_lights.data[1].color_size.w"; - actions.renames["LIGHT2_ENABLED"] = "directional_lights.data[2].enabled"; - actions.renames["LIGHT2_DIRECTION"] = "directional_lights.data[2].direction_energy.xyz"; - actions.renames["LIGHT2_ENERGY"] = "directional_lights.data[2].direction_energy.w"; - actions.renames["LIGHT2_COLOR"] = "directional_lights.data[2].color_size.xyz"; - actions.renames["LIGHT2_SIZE"] = "directional_lights.data[2].color_size.w"; - actions.renames["LIGHT3_ENABLED"] = "directional_lights.data[3].enabled"; - actions.renames["LIGHT3_DIRECTION"] = "directional_lights.data[3].direction_energy.xyz"; - actions.renames["LIGHT3_ENERGY"] = "directional_lights.data[3].direction_energy.w"; - actions.renames["LIGHT3_COLOR"] = "directional_lights.data[3].color_size.xyz"; - actions.renames["LIGHT3_SIZE"] = "directional_lights.data[3].color_size.w"; - actions.renames["AT_CUBEMAP_PASS"] = "AT_CUBEMAP_PASS"; - actions.renames["AT_HALF_RES_PASS"] = "AT_HALF_RES_PASS"; - actions.renames["AT_QUARTER_RES_PASS"] = "AT_QUARTER_RES_PASS"; - actions.custom_samplers["RADIANCE"] = "material_samplers[3]"; - actions.usage_defines["HALF_RES_COLOR"] = "\n#define USES_HALF_RES_COLOR\n"; - actions.usage_defines["QUARTER_RES_COLOR"] = "\n#define USES_QUARTER_RES_COLOR\n"; - actions.render_mode_defines["disable_fog"] = "#define DISABLE_FOG\n"; - - actions.sampler_array_name = "material_samplers"; - actions.base_texture_binding_index = 1; - actions.texture_layout_set = 1; - actions.base_uniform_string = "material."; - actions.base_varying_index = 10; - - actions.default_filter = ShaderLanguage::FILTER_LINEAR_MIPMAP; - actions.default_repeat = ShaderLanguage::REPEAT_ENABLE; - actions.global_buffer_array_variable = "global_variables.data"; - - sky_shader.compiler.initialize(actions); - } - - { - // default material and shader for sky shader - sky_shader.default_shader = storage->shader_create(); - storage->shader_set_code(sky_shader.default_shader, "shader_type sky; void fragment() { COLOR = vec3(0.0); } \n"); - sky_shader.default_material = storage->material_create(); - storage->material_set_shader(sky_shader.default_material, sky_shader.default_shader); - - SkyMaterialData *md = (SkyMaterialData *)storage->material_get_data(sky_shader.default_material, RendererStorageRD::SHADER_TYPE_SKY); - sky_shader.default_shader_rd = sky_shader.shader.version_get_shader(md->shader_data->version, SKY_VERSION_BACKGROUND); - - sky_scene_state.uniform_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(SkySceneState::UBO)); - - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 0; - u.ids.resize(12); - RID *ids_ptr = u.ids.ptrw(); - ids_ptr[0] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[1] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[2] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[3] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[4] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[5] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); - ids_ptr[6] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - ids_ptr[7] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - ids_ptr[8] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - ids_ptr[9] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - ids_ptr[10] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - ids_ptr[11] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 1; - u.ids.push_back(storage->global_variables_get_storage_buffer()); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.binding = 2; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(sky_scene_state.uniform_buffer); - uniforms.push_back(u); - } - - { - RD::Uniform u; - u.binding = 3; - u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; - u.ids.push_back(sky_scene_state.directional_light_buffer); - uniforms.push_back(u); - } - - sky_scene_state.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_UNIFORMS); - } - - { - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.binding = 0; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - RID vfog = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); - u.ids.push_back(vfog); - uniforms.push_back(u); - } - - sky_scene_state.default_fog_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_FOG); - } - - { - // Need defaults for using fog with clear color - sky_scene_state.fog_shader = storage->shader_create(); - storage->shader_set_code(sky_scene_state.fog_shader, "shader_type sky; uniform vec4 clear_color; void fragment() { COLOR = clear_color.rgb; } \n"); - sky_scene_state.fog_material = storage->material_create(); - storage->material_set_shader(sky_scene_state.fog_material, sky_scene_state.fog_shader); - - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 0; - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 1; - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 2; - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); - uniforms.push_back(u); - } + /* GI */ - sky_scene_state.fog_only_texture_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_TEXTURES); + if (!low_end && is_dynamic_gi_supported()) { + gi.init(storage, &sky); } - if (!low_end) { - //SDFGI - { - Vector<String> preprocess_modes; - preprocess_modes.push_back("\n#define MODE_SCROLL\n"); - preprocess_modes.push_back("\n#define MODE_SCROLL_OCCLUSION\n"); - preprocess_modes.push_back("\n#define MODE_INITIALIZE_JUMP_FLOOD\n"); - preprocess_modes.push_back("\n#define MODE_INITIALIZE_JUMP_FLOOD_HALF\n"); - preprocess_modes.push_back("\n#define MODE_JUMPFLOOD\n"); - preprocess_modes.push_back("\n#define MODE_JUMPFLOOD_OPTIMIZED\n"); - preprocess_modes.push_back("\n#define MODE_UPSCALE_JUMP_FLOOD\n"); - preprocess_modes.push_back("\n#define MODE_OCCLUSION\n"); - preprocess_modes.push_back("\n#define MODE_STORE\n"); - String defines = "\n#define OCCLUSION_SIZE " + itos(SDFGI::CASCADE_SIZE / SDFGI::PROBE_DIVISOR) + "\n"; - sdfgi_shader.preprocess.initialize(preprocess_modes, defines); - sdfgi_shader.preprocess_shader = sdfgi_shader.preprocess.version_create(); - for (int i = 0; i < SDGIShader::PRE_PROCESS_MAX; i++) { - sdfgi_shader.preprocess_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.preprocess.version_get_shader(sdfgi_shader.preprocess_shader, i)); - } - } - - { - //calculate tables - String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; - - Vector<String> direct_light_modes; - direct_light_modes.push_back("\n#define MODE_PROCESS_STATIC\n"); - direct_light_modes.push_back("\n#define MODE_PROCESS_DYNAMIC\n"); - sdfgi_shader.direct_light.initialize(direct_light_modes, defines); - sdfgi_shader.direct_light_shader = sdfgi_shader.direct_light.version_create(); - for (int i = 0; i < SDGIShader::DIRECT_LIGHT_MODE_MAX; i++) { - sdfgi_shader.direct_light_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.direct_light.version_get_shader(sdfgi_shader.direct_light_shader, i)); - } - } - - { - //calculate tables - String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; - defines += "\n#define SH_SIZE " + itos(SDFGI::SH_SIZE) + "\n"; - - Vector<String> integrate_modes; - integrate_modes.push_back("\n#define MODE_PROCESS\n"); - integrate_modes.push_back("\n#define MODE_STORE\n"); - integrate_modes.push_back("\n#define MODE_SCROLL\n"); - integrate_modes.push_back("\n#define MODE_SCROLL_STORE\n"); - sdfgi_shader.integrate.initialize(integrate_modes, defines); - sdfgi_shader.integrate_shader = sdfgi_shader.integrate.version_create(); - - for (int i = 0; i < SDGIShader::INTEGRATE_MODE_MAX; i++) { - sdfgi_shader.integrate_pipeline[i] = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, i)); - } - - { - Vector<RD::Uniform> uniforms; - - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; - u.binding = 0; - u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_WHITE)); - uniforms.push_back(u); - } - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; - u.binding = 1; - u.ids.push_back(storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED)); - uniforms.push_back(u); - } - - sdfgi_shader.integrate_default_sky_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sdfgi_shader.integrate.version_get_shader(sdfgi_shader.integrate_shader, 0), 1); - } - } - { - //calculate tables - String defines = "\n#define SDFGI_OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; - Vector<String> gi_modes; - gi_modes.push_back(""); - gi.shader.initialize(gi_modes, defines); - gi.shader_version = gi.shader.version_create(); - for (int i = 0; i < GI::MODE_MAX; i++) { - gi.pipelines[i] = RD::get_singleton()->compute_pipeline_create(gi.shader.version_get_shader(gi.shader_version, i)); - } - - gi.sdfgi_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(GI::SDFGIData)); - } - { - String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; - Vector<String> debug_modes; - debug_modes.push_back(""); - sdfgi_shader.debug.initialize(debug_modes, defines); - sdfgi_shader.debug_shader = sdfgi_shader.debug.version_create(); - sdfgi_shader.debug_shader_version = sdfgi_shader.debug.version_get_shader(sdfgi_shader.debug_shader, 0); - sdfgi_shader.debug_pipeline = RD::get_singleton()->compute_pipeline_create(sdfgi_shader.debug_shader_version); - } - { - String defines = "\n#define OCT_SIZE " + itos(SDFGI::LIGHTPROBE_OCT_SIZE) + "\n"; - - Vector<String> versions; - versions.push_back("\n#define MODE_PROBES\n"); - versions.push_back("\n#define MODE_VISIBILITY\n"); - - sdfgi_shader.debug_probes.initialize(versions, defines); - sdfgi_shader.debug_probes_shader = sdfgi_shader.debug_probes.version_create(); - - { - RD::PipelineRasterizationState rs; - rs.cull_mode = RD::POLYGON_CULL_DISABLED; - RD::PipelineDepthStencilState ds; - ds.enable_depth_test = true; - ds.enable_depth_write = true; - ds.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; - for (int i = 0; i < SDGIShader::PROBE_DEBUG_MAX; i++) { - RID debug_probes_shader_version = sdfgi_shader.debug_probes.version_get_shader(sdfgi_shader.debug_probes_shader, i); - sdfgi_shader.debug_probes_pipeline[i].setup(debug_probes_shader_version, RD::RENDER_PRIMITIVE_TRIANGLE_STRIPS, rs, RD::PipelineMultisampleState(), ds, RD::PipelineColorBlendState::create_disabled(), 0); - } - } - } - default_giprobe_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(GI::GIProbeData) * RenderBuffers::MAX_GIPROBES); + { //decals + cluster.max_decals = max_cluster_elements; + uint32_t decal_buffer_size = cluster.max_decals * sizeof(Cluster::DecalData); + cluster.decals = memnew_arr(Cluster::DecalData, cluster.max_decals); + cluster.decal_sort = memnew_arr(Cluster::InstanceSort<DecalInstance>, cluster.max_decals); + cluster.decal_buffer = RD::get_singleton()->storage_buffer_create(decal_buffer_size); } - //cluster setup - uint32_t uniform_max_size = RD::get_singleton()->limit_get(RD::LIMIT_MAX_UNIFORM_BUFFER_SIZE); - { //reflections - uint32_t reflection_buffer_size; - if (uniform_max_size < 65536) { - //Yes, you guessed right, ARM again - reflection_buffer_size = uniform_max_size; - } else { - reflection_buffer_size = 65536; - } - cluster.max_reflections = reflection_buffer_size / sizeof(Cluster::ReflectionData); + cluster.max_reflections = max_cluster_elements; cluster.reflections = memnew_arr(Cluster::ReflectionData, cluster.max_reflections); - cluster.reflection_buffer = RD::get_singleton()->storage_buffer_create(reflection_buffer_size); + cluster.reflection_sort = memnew_arr(Cluster::InstanceSort<ReflectionProbeInstance>, cluster.max_reflections); + cluster.reflection_buffer = RD::get_singleton()->storage_buffer_create(sizeof(Cluster::ReflectionData) * cluster.max_reflections); } { //lights - cluster.max_lights = MIN(1024 * 1024, uniform_max_size) / sizeof(Cluster::LightData); //1mb of lights + cluster.max_lights = max_cluster_elements; + uint32_t light_buffer_size = cluster.max_lights * sizeof(Cluster::LightData); - cluster.lights = memnew_arr(Cluster::LightData, cluster.max_lights); - cluster.light_buffer = RD::get_singleton()->storage_buffer_create(light_buffer_size); + cluster.omni_lights = memnew_arr(Cluster::LightData, cluster.max_lights); + cluster.omni_light_buffer = RD::get_singleton()->storage_buffer_create(light_buffer_size); + cluster.omni_light_sort = memnew_arr(Cluster::InstanceSort<LightInstance>, cluster.max_lights); + cluster.spot_lights = memnew_arr(Cluster::LightData, cluster.max_lights); + cluster.spot_light_buffer = RD::get_singleton()->storage_buffer_create(light_buffer_size); + cluster.spot_light_sort = memnew_arr(Cluster::InstanceSort<LightInstance>, cluster.max_lights); //defines += "\n#define MAX_LIGHT_DATA_STRUCTS " + itos(cluster.max_lights) + "\n"; - cluster.lights_instances = memnew_arr(RID, cluster.max_lights); - cluster.lights_shadow_rect_cache = memnew_arr(Rect2i, cluster.max_lights); cluster.max_directional_lights = MAX_DIRECTIONAL_LIGHTS; uint32_t directional_light_buffer_size = cluster.max_directional_lights * sizeof(Cluster::DirectionalLightData); @@ -8507,16 +4172,7 @@ RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { cluster.directional_light_buffer = RD::get_singleton()->uniform_buffer_create(directional_light_buffer_size); } - { //decals - cluster.max_decals = MIN(1024 * 1024, uniform_max_size) / sizeof(Cluster::DecalData); //1mb of decals - uint32_t decal_buffer_size = cluster.max_decals * sizeof(Cluster::DecalData); - cluster.decals = memnew_arr(Cluster::DecalData, cluster.max_decals); - cluster.decal_buffer = RD::get_singleton()->storage_buffer_create(decal_buffer_size); - } - - cluster.builder.setup(16, 8, 24); - - if (!low_end) { + if (!low_end && is_volumetric_supported()) { String defines = "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(cluster.max_directional_lights) + "\n"; Vector<String> volumetric_fog_modes; volumetric_fog_modes.push_back("\n#define MODE_DENSITY\n"); @@ -8528,6 +4184,7 @@ RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { for (int i = 0; i < VOLUMETRIC_FOG_SHADER_MAX; i++) { volumetric_fog.pipelines[i] = RD::get_singleton()->compute_pipeline_create(volumetric_fog.shader.version_get_shader(volumetric_fog.shader_version, i)); } + volumetric_fog.params_ubo = RD::get_singleton()->uniform_buffer_create(sizeof(VolumetricFogShader::ParamsUBO)); } { @@ -8539,74 +4196,57 @@ RendererSceneRenderRD::RendererSceneRenderRD(RendererStorageRD *p_storage) { shadow_sampler = RD::get_singleton()->sampler_create(sampler); } - camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokehShape(int(GLOBAL_GET("rendering/quality/depth_of_field/depth_of_field_bokeh_shape")))); - camera_effects_set_dof_blur_quality(RS::DOFBlurQuality(int(GLOBAL_GET("rendering/quality/depth_of_field/depth_of_field_bokeh_quality"))), GLOBAL_GET("rendering/quality/depth_of_field/depth_of_field_use_jitter")); - environment_set_ssao_quality(RS::EnvironmentSSAOQuality(int(GLOBAL_GET("rendering/quality/ssao/quality"))), GLOBAL_GET("rendering/quality/ssao/half_size"), GLOBAL_GET("rendering/quality/ssao/adaptive_target"), GLOBAL_GET("rendering/quality/ssao/blur_passes"), GLOBAL_GET("rendering/quality/ssao/fadeout_from"), GLOBAL_GET("rendering/quality/ssao/fadeout_to")); - screen_space_roughness_limiter = GLOBAL_GET("rendering/quality/screen_filters/screen_space_roughness_limiter_enabled"); - screen_space_roughness_limiter_amount = GLOBAL_GET("rendering/quality/screen_filters/screen_space_roughness_limiter_amount"); - screen_space_roughness_limiter_limit = GLOBAL_GET("rendering/quality/screen_filters/screen_space_roughness_limiter_limit"); - glow_bicubic_upscale = int(GLOBAL_GET("rendering/quality/glow/upscale_mode")) > 0; - glow_high_quality = GLOBAL_GET("rendering/quality/glow/use_high_quality"); - ssr_roughness_quality = RS::EnvironmentSSRRoughnessQuality(int(GLOBAL_GET("rendering/quality/screen_space_reflection/roughness_quality"))); - sss_quality = RS::SubSurfaceScatteringQuality(int(GLOBAL_GET("rendering/quality/subsurface_scattering/subsurface_scattering_quality"))); - sss_scale = GLOBAL_GET("rendering/quality/subsurface_scattering/subsurface_scattering_scale"); - sss_depth_scale = GLOBAL_GET("rendering/quality/subsurface_scattering/subsurface_scattering_depth_scale"); + camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokehShape(int(GLOBAL_GET("rendering/camera/depth_of_field/depth_of_field_bokeh_shape")))); + camera_effects_set_dof_blur_quality(RS::DOFBlurQuality(int(GLOBAL_GET("rendering/camera/depth_of_field/depth_of_field_bokeh_quality"))), GLOBAL_GET("rendering/camera/depth_of_field/depth_of_field_use_jitter")); + environment_set_ssao_quality(RS::EnvironmentSSAOQuality(int(GLOBAL_GET("rendering/environment/ssao/quality"))), GLOBAL_GET("rendering/environment/ssao/half_size"), GLOBAL_GET("rendering/environment/ssao/adaptive_target"), GLOBAL_GET("rendering/environment/ssao/blur_passes"), GLOBAL_GET("rendering/environment/ssao/fadeout_from"), GLOBAL_GET("rendering/environment/ssao/fadeout_to")); + screen_space_roughness_limiter = GLOBAL_GET("rendering/anti_aliasing/screen_space_roughness_limiter/enabled"); + screen_space_roughness_limiter_amount = GLOBAL_GET("rendering/anti_aliasing/screen_space_roughness_limiter/amount"); + screen_space_roughness_limiter_limit = GLOBAL_GET("rendering/anti_aliasing/screen_space_roughness_limiter/limit"); + glow_bicubic_upscale = int(GLOBAL_GET("rendering/environment/glow/upscale_mode")) > 0; + glow_high_quality = GLOBAL_GET("rendering/environment/glow/use_high_quality"); + ssr_roughness_quality = RS::EnvironmentSSRRoughnessQuality(int(GLOBAL_GET("rendering/environment/screen_space_reflection/roughness_quality"))); + sss_quality = RS::SubSurfaceScatteringQuality(int(GLOBAL_GET("rendering/environment/subsurface_scattering/subsurface_scattering_quality"))); + sss_scale = GLOBAL_GET("rendering/environment/subsurface_scattering/subsurface_scattering_scale"); + sss_depth_scale = GLOBAL_GET("rendering/environment/subsurface_scattering/subsurface_scattering_depth_scale"); directional_penumbra_shadow_kernel = memnew_arr(float, 128); directional_soft_shadow_kernel = memnew_arr(float, 128); penumbra_shadow_kernel = memnew_arr(float, 128); soft_shadow_kernel = memnew_arr(float, 128); - shadows_quality_set(RS::ShadowQuality(int(GLOBAL_GET("rendering/quality/shadows/soft_shadow_quality")))); - directional_shadow_quality_set(RS::ShadowQuality(int(GLOBAL_GET("rendering/quality/directional_shadow/soft_shadow_quality")))); + shadows_quality_set(RS::ShadowQuality(int(GLOBAL_GET("rendering/shadows/shadows/soft_shadow_quality")))); + directional_shadow_quality_set(RS::ShadowQuality(int(GLOBAL_GET("rendering/shadows/directional_shadow/soft_shadow_quality")))); - environment_set_volumetric_fog_volume_size(GLOBAL_GET("rendering/volumetric_fog/volume_size"), GLOBAL_GET("rendering/volumetric_fog/volume_depth")); - environment_set_volumetric_fog_filter_active(GLOBAL_GET("rendering/volumetric_fog/use_filter")); - environment_set_volumetric_fog_directional_shadow_shrink_size(GLOBAL_GET("rendering/volumetric_fog/directional_shadow_shrink")); - environment_set_volumetric_fog_positional_shadow_shrink_size(GLOBAL_GET("rendering/volumetric_fog/positional_shadow_shrink")); + environment_set_volumetric_fog_volume_size(GLOBAL_GET("rendering/environment/volumetric_fog/volume_size"), GLOBAL_GET("rendering/environment/volumetric_fog/volume_depth")); + environment_set_volumetric_fog_filter_active(GLOBAL_GET("rendering/environment/volumetric_fog/use_filter")); cull_argument.set_page_pool(&cull_argument_pool); } RendererSceneRenderRD::~RendererSceneRenderRD() { - for (Map<Vector2i, ShadowMap>::Element *E = shadow_maps.front(); E; E = E->next()) { - RD::get_singleton()->free(E->get().depth); - } for (Map<int, ShadowCubemap>::Element *E = shadow_cubemaps.front(); E; E = E->next()) { RD::get_singleton()->free(E->get().cubemap); } - if (sky_scene_state.uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(sky_scene_state.uniform_set)) { - RD::get_singleton()->free(sky_scene_state.uniform_set); + if (sky.sky_scene_state.uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(sky.sky_scene_state.uniform_set)) { + RD::get_singleton()->free(sky.sky_scene_state.uniform_set); } if (!low_end) { - RD::get_singleton()->free(default_giprobe_buffer); - RD::get_singleton()->free(gi_probe_lights_uniform); - RD::get_singleton()->free(gi.sdfgi_ubo); - - giprobe_debug_shader.version_free(giprobe_debug_shader_version); - giprobe_shader.version_free(giprobe_lighting_shader_version); - gi.shader.version_free(gi.shader_version); - sdfgi_shader.debug_probes.version_free(sdfgi_shader.debug_probes_shader); - sdfgi_shader.debug.version_free(sdfgi_shader.debug_shader); - sdfgi_shader.direct_light.version_free(sdfgi_shader.direct_light_shader); - sdfgi_shader.integrate.version_free(sdfgi_shader.integrate_shader); - sdfgi_shader.preprocess.version_free(sdfgi_shader.preprocess_shader); + gi.free(); volumetric_fog.shader.version_free(volumetric_fog.shader_version); - - memdelete_arr(gi_probe_lights); - } - - SkyMaterialData *md = (SkyMaterialData *)storage->material_get_data(sky_shader.default_material, RendererStorageRD::SHADER_TYPE_SKY); - sky_shader.shader.version_free(md->shader_data->version); - RD::get_singleton()->free(sky_scene_state.directional_light_buffer); - RD::get_singleton()->free(sky_scene_state.uniform_buffer); - memdelete_arr(sky_scene_state.directional_lights); - memdelete_arr(sky_scene_state.last_frame_directional_lights); - storage->free(sky_shader.default_shader); - storage->free(sky_shader.default_material); - storage->free(sky_scene_state.fog_shader); - storage->free(sky_scene_state.fog_material); + RD::get_singleton()->free(volumetric_fog.params_ubo); + } + + RendererSceneSkyRD::SkyMaterialData *md = (RendererSceneSkyRD::SkyMaterialData *)storage->material_get_data(sky.sky_shader.default_material, RendererStorageRD::SHADER_TYPE_SKY); + sky.sky_shader.shader.version_free(md->shader_data->version); + RD::get_singleton()->free(sky.sky_scene_state.directional_light_buffer); + RD::get_singleton()->free(sky.sky_scene_state.uniform_buffer); + memdelete_arr(sky.sky_scene_state.directional_lights); + memdelete_arr(sky.sky_scene_state.last_frame_directional_lights); + storage->free(sky.sky_shader.default_shader); + storage->free(sky.sky_shader.default_material); + storage->free(sky.sky_scene_state.fog_shader); + storage->free(sky.sky_scene_state.fog_material); memdelete_arr(directional_penumbra_shadow_kernel); memdelete_arr(directional_soft_shadow_kernel); memdelete_arr(penumbra_shadow_kernel); @@ -8614,15 +4254,19 @@ RendererSceneRenderRD::~RendererSceneRenderRD() { { RD::get_singleton()->free(cluster.directional_light_buffer); - RD::get_singleton()->free(cluster.light_buffer); + RD::get_singleton()->free(cluster.omni_light_buffer); + RD::get_singleton()->free(cluster.spot_light_buffer); RD::get_singleton()->free(cluster.reflection_buffer); RD::get_singleton()->free(cluster.decal_buffer); memdelete_arr(cluster.directional_lights); - memdelete_arr(cluster.lights); - memdelete_arr(cluster.lights_shadow_rect_cache); - memdelete_arr(cluster.lights_instances); + memdelete_arr(cluster.omni_lights); + memdelete_arr(cluster.spot_lights); + memdelete_arr(cluster.omni_light_sort); + memdelete_arr(cluster.spot_light_sort); memdelete_arr(cluster.reflections); + memdelete_arr(cluster.reflection_sort); memdelete_arr(cluster.decals); + memdelete_arr(cluster.decal_sort); } RD::get_singleton()->free(shadow_sampler); diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.h b/servers/rendering/renderer_rd/renderer_scene_render_rd.h index f81a35f025..264c0e4276 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.h @@ -34,69 +34,24 @@ #include "core/templates/local_vector.h" #include "core/templates/rid_owner.h" #include "servers/rendering/renderer_compositor.h" -#include "servers/rendering/renderer_rd/light_cluster_builder.h" +#include "servers/rendering/renderer_rd/cluster_builder_rd.h" +#include "servers/rendering/renderer_rd/renderer_scene_environment_rd.h" +#include "servers/rendering/renderer_rd/renderer_scene_gi_rd.h" +#include "servers/rendering/renderer_rd/renderer_scene_sky_rd.h" #include "servers/rendering/renderer_rd/renderer_storage_rd.h" -#include "servers/rendering/renderer_rd/shaders/gi.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/giprobe.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/giprobe_debug.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sdfgi_debug_probes.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sdfgi_preprocess.glsl.gen.h" -#include "servers/rendering/renderer_rd/shaders/sky.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/volumetric_fog.glsl.gen.h" #include "servers/rendering/renderer_scene_render.h" #include "servers/rendering/rendering_device.h" class RendererSceneRenderRD : public RendererSceneRender { + friend RendererSceneSkyRD; + friend RendererSceneGIRD; + protected: + RendererStorageRD *storage; double time; - - // Skys need less info from Directional Lights than the normal shaders - struct SkyDirectionalLightData { - float direction[3]; - float energy; - float color[3]; - float size; - uint32_t enabled; - uint32_t pad[3]; - }; - - struct SkySceneState { - struct UBO { - uint32_t volumetric_fog_enabled; - float volumetric_fog_inv_length; - float volumetric_fog_detail_spread; - - float fog_aerial_perspective; - - float fog_light_color[3]; - float fog_sun_scatter; - - uint32_t fog_enabled; - float fog_density; - - float z_far; - uint32_t directional_light_count; - }; - - UBO ubo; - - SkyDirectionalLightData *directional_lights; - SkyDirectionalLightData *last_frame_directional_lights; - uint32_t max_directional_lights; - uint32_t last_frame_directional_light_count; - RID directional_light_buffer; - RID uniform_set; - RID uniform_buffer; - RID fog_uniform_set; - RID default_fog_uniform_set; - - RID fog_shader; - RID fog_material; - RID fog_only_texture_uniform_set; - } sky_scene_state; + double time_step = 0; + bool low_end = false; // If true GI and Volumetric fog are disabled struct RenderBufferData { virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa) = 0; @@ -104,19 +59,22 @@ protected: }; virtual RenderBufferData *_create_render_buffer_data() = 0; - void _setup_lights(const PagedArray<RID> &p_lights, const Transform &p_camera_inverse_transform, RID p_shadow_atlas, bool p_using_shadows, uint32_t &r_directional_light_count, uint32_t &r_positional_light_count); + void _setup_lights(const PagedArray<RID> &p_lights, const Transform &p_camera_transform, RID p_shadow_atlas, bool p_using_shadows, uint32_t &r_directional_light_count, uint32_t &r_positional_light_count); void _setup_decals(const PagedArray<RID> &p_decals, const Transform &p_camera_inverse_xform); void _setup_reflections(const PagedArray<RID> &p_reflections, const Transform &p_camera_inverse_transform, RID p_environment); - void _setup_giprobes(RID p_render_buffers, const Transform &p_transform, const PagedArray<RID> &p_gi_probes, uint32_t &r_gi_probes_used); - virtual void _render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, int p_directional_light_count, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_color, float p_screen_lod_threshold) = 0; - virtual void _render_shadow(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0) = 0; + virtual void _render_scene(RID p_render_buffer, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_cluster_buffer, uint32_t p_cluster_size, uint32_t p_cluster_max_elements, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, const Color &p_default_color, float p_screen_lod_threshold) = 0; + + virtual void _render_shadow_begin() = 0; + virtual void _render_shadow_append(RID p_framebuffer, const PagedArray<GeometryInstance *> &p_instances, const CameraMatrix &p_projection, const Transform &p_transform, float p_zfar, float p_bias, float p_normal_bias, bool p_use_dp, bool p_use_dp_flip, bool p_use_pancake, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0.0, float p_screen_lod_threshold = 0.0, const Rect2i &p_rect = Rect2i(), bool p_flip_y = false, bool p_clear_region = true, bool p_begin = true, bool p_end = true) = 0; + virtual void _render_shadow_process() = 0; + virtual void _render_shadow_end(uint32_t p_barrier = RD::BARRIER_MASK_ALL) = 0; + virtual void _render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) = 0; virtual void _render_uv2(const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) = 0; virtual void _render_sdfgi(RID p_render_buffers, const Vector3i &p_from, const Vector3i &p_size, const AABB &p_bounds, const PagedArray<GeometryInstance *> &p_instances, const RID &p_albedo_texture, const RID &p_emission_texture, const RID &p_emission_aniso_texture, const RID &p_geom_facing_texture) = 0; virtual void _render_particle_collider_heightfield(RID p_fb, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, const PagedArray<GeometryInstance *> &p_instances) = 0; - virtual void _debug_giprobe(RID p_gi_probe, RenderingDevice::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform, bool p_lighting, bool p_emission, float p_alpha); void _debug_sdfgi_probes(RID p_render_buffers, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform); RenderBufferData *render_buffers_get_data(RID p_render_buffers); @@ -124,204 +82,36 @@ protected: virtual void _base_uniforms_changed() = 0; virtual void _render_buffers_uniform_set_changed(RID p_render_buffers) = 0; virtual RID _render_buffers_get_normal_texture(RID p_render_buffers) = 0; - virtual RID _render_buffers_get_ambient_texture(RID p_render_buffers) = 0; - virtual RID _render_buffers_get_reflection_texture(RID p_render_buffers) = 0; void _process_ssao(RID p_render_buffers, RID p_environment, RID p_normal_buffer, const CameraMatrix &p_projection); void _process_ssr(RID p_render_buffers, RID p_dest_framebuffer, RID p_normal_buffer, RID p_specular_buffer, RID p_metallic, const Color &p_metallic_mask, RID p_environment, const CameraMatrix &p_projection, bool p_use_additive); void _process_sss(RID p_render_buffers, const CameraMatrix &p_camera); - void _setup_sky(RID p_environment, RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform, const Size2i p_screen_size); - void _update_sky(RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform); - void _draw_sky(bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform); - void _process_gi(RID p_render_buffers, RID p_normal_roughness_buffer, RID p_ambient_buffer, RID p_reflection_buffer, RID p_gi_probe_buffer, RID p_environment, const CameraMatrix &p_projection, const Transform &p_transform, const PagedArray<RID> &p_gi_probes); + bool _needs_post_prepass_render(bool p_use_gi); + void _post_prepass_render(bool p_use_gi); + void _pre_resolve_render(bool p_use_gi); + + void _pre_opaque_render(bool p_use_ssao, bool p_use_gi, RID p_normal_roughness_buffer, RID p_gi_probe_buffer); + uint32_t _get_render_state_directional_light_count() const; // needed for a single argument calls (material and uv2) PagedArrayPool<GeometryInstance *> cull_argument_pool; PagedArray<GeometryInstance *> cull_argument; //need this to exist -private: - RS::ViewportDebugDraw debug_draw = RS::VIEWPORT_DEBUG_DRAW_DISABLED; - double time_step = 0; - static RendererSceneRenderRD *singleton; - - int roughness_layers; - RendererStorageRD *storage; - - struct ReflectionData { - struct Layer { - struct Mipmap { - RID framebuffers[6]; - RID views[6]; - Size2i size; - }; - Vector<Mipmap> mipmaps; //per-face view - Vector<RID> views; // per-cubemap view - }; + RendererSceneGIRD gi; + RendererSceneSkyRD sky; - struct DownsampleLayer { - struct Mipmap { - RID view; - Size2i size; - }; - Vector<Mipmap> mipmaps; - }; - - RID radiance_base_cubemap; //cubemap for first layer, first cubemap - RID downsampled_radiance_cubemap; - DownsampleLayer downsampled_layer; - RID coefficient_buffer; - - bool dirty = true; - - Vector<Layer> layers; - }; - - void _clear_reflection_data(ReflectionData &rd); - void _update_reflection_data(ReflectionData &rd, int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality); - void _create_reflection_fast_filter(ReflectionData &rd, bool p_use_arrays); - void _create_reflection_importance_sample(ReflectionData &rd, bool p_use_arrays, int p_cube_side, int p_base_layer); - void _update_reflection_mipmaps(ReflectionData &rd, int p_start, int p_end); - - /* Sky shader */ - - enum SkyVersion { - SKY_VERSION_BACKGROUND, - SKY_VERSION_HALF_RES, - SKY_VERSION_QUARTER_RES, - SKY_VERSION_CUBEMAP, - SKY_VERSION_CUBEMAP_HALF_RES, - SKY_VERSION_CUBEMAP_QUARTER_RES, - SKY_VERSION_MAX - }; - - struct SkyShader { - SkyShaderRD shader; - ShaderCompilerRD compiler; - - RID default_shader; - RID default_material; - RID default_shader_rd; - } sky_shader; - - struct SkyShaderData : public RendererStorageRD::ShaderData { - bool valid; - RID version; - - PipelineCacheRD pipelines[SKY_VERSION_MAX]; - Map<StringName, ShaderLanguage::ShaderNode::Uniform> uniforms; - Vector<ShaderCompilerRD::GeneratedCode::Texture> texture_uniforms; - - Vector<uint32_t> ubo_offsets; - uint32_t ubo_size; - - String path; - String code; - Map<StringName, RID> default_texture_params; - - bool uses_time; - bool uses_position; - bool uses_half_res; - bool uses_quarter_res; - bool uses_light; - - virtual void set_code(const String &p_Code); - virtual void set_default_texture_param(const StringName &p_name, RID p_texture); - virtual void get_param_list(List<PropertyInfo> *p_param_list) const; - virtual void get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const; - virtual bool is_param_texture(const StringName &p_param) const; - virtual bool is_animated() const; - virtual bool casts_shadows() const; - virtual Variant get_default_parameter(const StringName &p_parameter) const; - SkyShaderData(); - virtual ~SkyShaderData(); - }; - - RendererStorageRD::ShaderData *_create_sky_shader_func(); - static RendererStorageRD::ShaderData *_create_sky_shader_funcs() { - return static_cast<RendererSceneRenderRD *>(singleton)->_create_sky_shader_func(); - }; - - struct SkyMaterialData : public RendererStorageRD::MaterialData { - uint64_t last_frame; - SkyShaderData *shader_data; - RID uniform_buffer; - RID uniform_set; - Vector<RID> texture_cache; - Vector<uint8_t> ubo_data; - bool uniform_set_updated; - - virtual void set_render_priority(int p_priority) {} - virtual void set_next_pass(RID p_pass) {} - virtual void update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty); - virtual ~SkyMaterialData(); - }; - - RendererStorageRD::MaterialData *_create_sky_material_func(SkyShaderData *p_shader); - static RendererStorageRD::MaterialData *_create_sky_material_funcs(RendererStorageRD::ShaderData *p_shader) { - return static_cast<RendererSceneRenderRD *>(singleton)->_create_sky_material_func(static_cast<SkyShaderData *>(p_shader)); - }; - - enum SkyTextureSetVersion { - SKY_TEXTURE_SET_BACKGROUND, - SKY_TEXTURE_SET_HALF_RES, - SKY_TEXTURE_SET_QUARTER_RES, - SKY_TEXTURE_SET_CUBEMAP, - SKY_TEXTURE_SET_CUBEMAP_HALF_RES, - SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES, - SKY_TEXTURE_SET_MAX - }; - - enum SkySet { - SKY_SET_UNIFORMS, - SKY_SET_MATERIAL, - SKY_SET_TEXTURES, - SKY_SET_FOG, - SKY_SET_MAX - }; - - /* SKY */ - struct Sky { - RID radiance; - RID half_res_pass; - RID half_res_framebuffer; - RID quarter_res_pass; - RID quarter_res_framebuffer; - Size2i screen_size; - - RID texture_uniform_sets[SKY_TEXTURE_SET_MAX]; - RID uniform_set; - - RID material; - RID uniform_buffer; - - int radiance_size = 256; - - RS::SkyMode mode = RS::SKY_MODE_AUTOMATIC; - - ReflectionData reflection; - bool dirty = false; - int processing_layer = 0; - Sky *dirty_list = nullptr; - - //State to track when radiance cubemap needs updating - SkyMaterialData *prev_material; - Vector3 prev_position; - float prev_time; - - RID sdfgi_integrate_sky_uniform_set; - }; - - Sky *dirty_sky_list = nullptr; - - void _sky_invalidate(Sky *p_sky); - void _update_dirty_skys(); - RID _get_sky_textures(Sky *p_sky, SkyTextureSetVersion p_version); - - uint32_t sky_ggx_samples_quality; - bool sky_use_cubemap_array; + RendererSceneEnvironmentRD *get_environment(RID p_environment) { + if (p_environment.is_valid()) { + return environment_owner.getornull(p_environment); + } else { + return nullptr; + } + } - mutable RID_Owner<Sky> sky_owner; +private: + RS::ViewportDebugDraw debug_draw = RS::VIEWPORT_DEBUG_DRAW_DISABLED; + static RendererSceneRenderRD *singleton; /* REFLECTION ATLAS */ @@ -335,11 +125,13 @@ private: struct Reflection { RID owner; - ReflectionData data; + RendererSceneSkyRD::ReflectionData data; RID fbs[6]; }; Vector<Reflection> reflections; + + ClusterBuilderRD *cluster_builder = nullptr; }; mutable RID_Owner<ReflectionAtlas> reflection_atlas_owner; @@ -383,151 +175,6 @@ private: mutable RID_Owner<LightmapInstance> lightmap_instance_owner; - /* GIPROBE INSTANCE */ - - struct GIProbeLight { - uint32_t type; - float energy; - float radius; - float attenuation; - - float color[3]; - float spot_angle_radians; - - float position[3]; - float spot_attenuation; - - float direction[3]; - uint32_t has_shadow; - }; - - struct GIProbePushConstant { - int32_t limits[3]; - uint32_t stack_size; - - float emission_scale; - float propagation; - float dynamic_range; - uint32_t light_count; - - uint32_t cell_offset; - uint32_t cell_count; - float aniso_strength; - uint32_t pad; - }; - - struct GIProbeDynamicPushConstant { - int32_t limits[3]; - uint32_t light_count; - int32_t x_dir[3]; - float z_base; - int32_t y_dir[3]; - float z_sign; - int32_t z_dir[3]; - float pos_multiplier; - uint32_t rect_pos[2]; - uint32_t rect_size[2]; - uint32_t prev_rect_ofs[2]; - uint32_t prev_rect_size[2]; - uint32_t flip_x; - uint32_t flip_y; - float dynamic_range; - uint32_t on_mipmap; - float propagation; - float pad[3]; - }; - - struct GIProbeInstance { - RID probe; - RID texture; - RID write_buffer; - - struct Mipmap { - RID texture; - RID uniform_set; - RID second_bounce_uniform_set; - RID write_uniform_set; - uint32_t level; - uint32_t cell_offset; - uint32_t cell_count; - }; - Vector<Mipmap> mipmaps; - - struct DynamicMap { - RID texture; //color normally, or emission on first pass - RID fb_depth; //actual depth buffer for the first pass, float depth for later passes - RID depth; //actual depth buffer for the first pass, float depth for later passes - RID normal; //normal buffer for the first pass - RID albedo; //emission buffer for the first pass - RID orm; //orm buffer for the first pass - RID fb; //used for rendering, only valid on first map - RID uniform_set; - uint32_t size; - int mipmap; // mipmap to write to, -1 if no mipmap assigned - }; - - Vector<DynamicMap> dynamic_maps; - - int slot = -1; - uint32_t last_probe_version = 0; - uint32_t last_probe_data_version = 0; - - //uint64_t last_pass = 0; - uint32_t render_index = 0; - - bool has_dynamic_object_data = false; - - Transform transform; - }; - - GIProbeLight *gi_probe_lights; - uint32_t gi_probe_max_lights; - RID gi_probe_lights_uniform; - - enum { - GI_PROBE_SHADER_VERSION_COMPUTE_LIGHT, - GI_PROBE_SHADER_VERSION_COMPUTE_SECOND_BOUNCE, - GI_PROBE_SHADER_VERSION_COMPUTE_MIPMAP, - GI_PROBE_SHADER_VERSION_WRITE_TEXTURE, - GI_PROBE_SHADER_VERSION_DYNAMIC_OBJECT_LIGHTING, - GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE, - GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_PLOT, - GI_PROBE_SHADER_VERSION_DYNAMIC_SHRINK_WRITE_PLOT, - GI_PROBE_SHADER_VERSION_MAX - }; - GiprobeShaderRD giprobe_shader; - RID giprobe_lighting_shader_version; - RID giprobe_lighting_shader_version_shaders[GI_PROBE_SHADER_VERSION_MAX]; - RID giprobe_lighting_shader_version_pipelines[GI_PROBE_SHADER_VERSION_MAX]; - - mutable RID_Owner<GIProbeInstance> gi_probe_instance_owner; - - RS::GIProbeQuality gi_probe_quality = RS::GI_PROBE_QUALITY_HIGH; - - enum { - GI_PROBE_DEBUG_COLOR, - GI_PROBE_DEBUG_LIGHT, - GI_PROBE_DEBUG_EMISSION, - GI_PROBE_DEBUG_LIGHT_FULL, - GI_PROBE_DEBUG_MAX - }; - - struct GIProbeDebugPushConstant { - float projection[16]; - uint32_t cell_offset; - float dynamic_range; - float alpha; - uint32_t level; - int32_t bounds[3]; - uint32_t pad; - }; - - GiprobeDebugShaderRD giprobe_debug_shader; - RID giprobe_debug_shader_version; - RID giprobe_debug_shader_version_shaders[GI_PROBE_DEBUG_MAX]; - PipelineCacheRD giprobe_debug_shader_version_pipelines[GI_PROBE_DEBUG_MAX]; - RID giprobe_debug_uniform_set; - /* SHADOW ATLAS */ struct ShadowShrinkStage { @@ -571,17 +218,18 @@ private: uint32_t smallest_subdiv = 0; int size = 0; + bool use_16_bits = false; RID depth; RID fb; //for copying Map<RID, uint32_t> shadow_owners; - - Vector<ShadowShrinkStage> shrink_stages; }; RID_Owner<ShadowAtlas> shadow_atlas_owner; + void _update_shadow_atlas(ShadowAtlas *shadow_atlas); + bool _shadow_atlas_find_shadow(ShadowAtlas *shadow_atlas, int *p_in_quadrants, int p_quadrant_count, int p_current_subdiv, uint64_t p_tick, int &r_quadrant, int &r_shadow); RS::ShadowQuality shadows_quality = RS::SHADOW_QUALITY_MAX; //So it always updates when first set @@ -602,17 +250,16 @@ private: struct DirectionalShadow { RID depth; + RID fb; //when renderign direct int light_count = 0; int size = 0; + bool use_16_bits = false; int current_light = 0; - Vector<ShadowShrinkStage> shrink_stages; - } directional_shadow; - void _allocate_shadow_shrink_stages(RID p_base, int p_base_size, Vector<ShadowShrinkStage> &shrink_stages, uint32_t p_target_size); - void _clear_shadow_shrink_stages(Vector<ShadowShrinkStage> &shrink_stages); + void _update_directional_shadow_atlas(); /* SHADOW CUBEMAPS */ @@ -624,14 +271,6 @@ private: Map<int, ShadowCubemap> shadow_cubemaps; ShadowCubemap *_get_shadow_cubemap(int p_size); - struct ShadowMap { - RID depth; - RID fb; - }; - - Map<Vector2i, ShadowMap> shadow_maps; - ShadowMap *_get_shadow_map(const Size2i &p_size); - void _create_shadow_cubemaps(); /* LIGHT INSTANCE */ @@ -651,7 +290,7 @@ private: RS::LightType light_type = RS::LIGHT_DIRECTIONAL; - ShadowTransform shadow_transform[4]; + ShadowTransform shadow_transform[6]; AABB aabb; RID self; @@ -684,110 +323,6 @@ private: /* ENVIRONMENT */ - struct Environment { - // BG - RS::EnvironmentBG background = RS::ENV_BG_CLEAR_COLOR; - RID sky; - float sky_custom_fov = 0.0; - Basis sky_orientation; - Color bg_color; - float bg_energy = 1.0; - int canvas_max_layer = 0; - RS::EnvironmentAmbientSource ambient_source = RS::ENV_AMBIENT_SOURCE_BG; - Color ambient_light; - float ambient_light_energy = 1.0; - float ambient_sky_contribution = 1.0; - RS::EnvironmentReflectionSource reflection_source = RS::ENV_REFLECTION_SOURCE_BG; - Color ao_color; - - /// Tonemap - - RS::EnvironmentToneMapper tone_mapper; - float exposure = 1.0; - float white = 1.0; - bool auto_exposure = false; - float min_luminance = 0.2; - float max_luminance = 8.0; - float auto_exp_speed = 0.2; - float auto_exp_scale = 0.5; - uint64_t auto_exposure_version = 0; - - // Fog - bool fog_enabled = false; - Color fog_light_color = Color(0.5, 0.6, 0.7); - float fog_light_energy = 1.0; - float fog_sun_scatter = 0.0; - float fog_density = 0.001; - float fog_height = 0.0; - float fog_height_density = 0.0; //can be negative to invert effect - float fog_aerial_perspective = 0.0; - - /// Volumetric Fog - /// - bool volumetric_fog_enabled = false; - float volumetric_fog_density = 0.01; - Color volumetric_fog_light = Color(0, 0, 0); - float volumetric_fog_light_energy = 0.0; - float volumetric_fog_length = 64.0; - float volumetric_fog_detail_spread = 2.0; - RS::EnvVolumetricFogShadowFilter volumetric_fog_shadow_filter = RS::ENV_VOLUMETRIC_FOG_SHADOW_FILTER_LOW; - float volumetric_fog_gi_inject = 0.0; - - /// Glow - - bool glow_enabled = false; - Vector<float> glow_levels; - float glow_intensity = 0.8; - float glow_strength = 1.0; - float glow_bloom = 0.0; - float glow_mix = 0.01; - RS::EnvironmentGlowBlendMode glow_blend_mode = RS::ENV_GLOW_BLEND_MODE_SOFTLIGHT; - float glow_hdr_bleed_threshold = 1.0; - float glow_hdr_luminance_cap = 12.0; - float glow_hdr_bleed_scale = 2.0; - - /// SSAO - - bool ssao_enabled = false; - float ssao_radius = 1.0; - float ssao_intensity = 2.0; - float ssao_power = 1.5; - float ssao_detail = 0.5; - float ssao_horizon = 0.06; - float ssao_sharpness = 0.98; - float ssao_direct_light_affect = 0.0; - float ssao_ao_channel_affect = 0.0; - - /// SSR - /// - bool ssr_enabled = false; - int ssr_max_steps = 64; - float ssr_fade_in = 0.15; - float ssr_fade_out = 2.0; - float ssr_depth_tolerance = 0.2; - - /// SDFGI - bool sdfgi_enabled = false; - RS::EnvironmentSDFGICascades sdfgi_cascades; - float sdfgi_min_cell_size = 0.2; - bool sdfgi_use_occlusion = false; - bool sdfgi_use_multibounce = false; - bool sdfgi_read_sky_light = false; - float sdfgi_energy = 1.0; - float sdfgi_normal_bias = 1.1; - float sdfgi_probe_bias = 1.1; - RS::EnvironmentSDFGIYScale sdfgi_y_scale = RS::ENV_SDFGI_Y_SCALE_DISABLED; - - /// Adjustments - - bool adjustments_enabled = false; - float adjustments_brightness = 1.0f; - float adjustments_contrast = 1.0f; - float adjustments_saturation = 1.0f; - bool use_1d_color_correction = false; - RID color_correction = RID(); - }; - RS::EnvironmentSSAOQuality ssao_quality = RS::ENV_SSAO_QUALITY_MEDIUM; bool ssao_half_size = false; bool ssao_using_half_size = false; @@ -800,9 +335,7 @@ private: bool glow_high_quality = false; RS::EnvironmentSSRRoughnessQuality ssr_roughness_quality = RS::ENV_SSR_ROUGNESS_QUALITY_LOW; - static uint64_t auto_exposure_counter; - - mutable RID_Owner<Environment> environment_owner; + mutable RID_Owner<RendererSceneEnvironmentRD, true> environment_owner; /* CAMERA EFFECTS */ @@ -828,18 +361,16 @@ private: float sss_scale = 0.05; float sss_depth_scale = 0.01; - mutable RID_Owner<CameraEffects> camera_effects_owner; + mutable RID_Owner<CameraEffects, true> camera_effects_owner; /* RENDER BUFFERS */ - struct SDFGI; + ClusterBuilderSharedDataRD cluster_builder_shared; + ClusterBuilderRD *current_cluster_builder = nullptr; + struct VolumetricFog; struct RenderBuffers { - enum { - MAX_GIPROBES = 8 - }; - RenderBufferData *data = nullptr; int width = 0, height = 0; RS::ViewportMSAA msaa = RS::VIEWPORT_MSAA_DISABLED; @@ -853,9 +384,11 @@ private: RID texture; //main texture for rendering to, must be filled after done rendering RID depth_texture; //main depth texture - RID gi_uniform_set; - SDFGI *sdfgi = nullptr; + RendererSceneGIRD::SDFGI *sdfgi = nullptr; VolumetricFog *volumetric_fog = nullptr; + RendererSceneGIRD::RenderBuffersGI gi; + + ClusterBuilderRD *cluster_builder = nullptr; //built-in textures used for ping pong image processing and blurring struct Blur { @@ -894,389 +427,11 @@ private: RID blur_radius[2]; } ssr; - RID giprobe_textures[MAX_GIPROBES]; - RID giprobe_buffer; - }; - - RID default_giprobe_buffer; - - /* SDFGI */ - - struct SDFGI { - enum { - MAX_CASCADES = 8, - CASCADE_SIZE = 128, - PROBE_DIVISOR = 16, - ANISOTROPY_SIZE = 6, - MAX_DYNAMIC_LIGHTS = 128, - MAX_STATIC_LIGHTS = 1024, - LIGHTPROBE_OCT_SIZE = 6, - SH_SIZE = 16 - }; - - struct Cascade { - struct UBO { - float offset[3]; - float to_cell; - int32_t probe_offset[3]; - uint32_t pad; - }; - - //cascade blocks are full-size for volume (128^3), half size for albedo/emission - RID sdf_tex; - RID light_tex; - RID light_aniso_0_tex; - RID light_aniso_1_tex; - - RID light_data; - RID light_aniso_0_data; - RID light_aniso_1_data; - - struct SolidCell { // this struct is unused, but remains as reference for size - uint32_t position; - uint32_t albedo; - uint32_t static_light; - uint32_t static_light_aniso; - }; - - RID solid_cell_dispatch_buffer; //buffer for indirect compute dispatch - RID solid_cell_buffer; - - RID lightprobe_history_tex; - RID lightprobe_average_tex; - - float cell_size; - Vector3i position; - - static const Vector3i DIRTY_ALL; - Vector3i dirty_regions; //(0,0,0 is not dirty, negative is refresh from the end, DIRTY_ALL is refresh all. - - RID sdf_store_uniform_set; - RID sdf_direct_light_uniform_set; - RID scroll_uniform_set; - RID scroll_occlusion_uniform_set; - RID integrate_uniform_set; - RID lights_buffer; - }; - - //used for rendering (voxelization) - RID render_albedo; - RID render_emission; - RID render_emission_aniso; - RID render_occlusion[8]; - RID render_geom_facing; - - RID render_sdf[2]; - RID render_sdf_half[2]; - - //used for ping pong processing in cascades - RID sdf_initialize_uniform_set; - RID sdf_initialize_half_uniform_set; - RID jump_flood_uniform_set[2]; - RID jump_flood_half_uniform_set[2]; - RID sdf_upscale_uniform_set; - int upscale_jfa_uniform_set_index; - RID occlusion_uniform_set; - - uint32_t cascade_size = 128; - - LocalVector<Cascade> cascades; - - RID lightprobe_texture; - RID lightprobe_data; - RID occlusion_texture; - RID occlusion_data; - RID ambient_texture; //integrates with volumetric fog - - RID lightprobe_history_scroll; //used for scrolling lightprobes - RID lightprobe_average_scroll; //used for scrolling lightprobes - - uint32_t history_size = 0; - float solid_cell_ratio = 0; - uint32_t solid_cell_count = 0; - - RS::EnvironmentSDFGICascades cascade_mode; - float min_cell_size = 0; - uint32_t probe_axis_count = 0; //amount of probes per axis, this is an odd number because it encloses endpoints - - RID debug_uniform_set; - RID debug_probes_uniform_set; - RID cascades_ubo; - - bool uses_occlusion = false; - bool uses_multibounce = false; - bool reads_sky = false; - float energy = 1.0; - float normal_bias = 1.1; - float probe_bias = 1.1; - RS::EnvironmentSDFGIYScale y_scale_mode = RS::ENV_SDFGI_Y_SCALE_DISABLED; - - float y_mult = 1.0; - - uint32_t render_pass = 0; + RID ambient_buffer; + RID reflection_buffer; }; - RS::EnvironmentSDFGIRayCount sdfgi_ray_count = RS::ENV_SDFGI_RAY_COUNT_16; - RS::EnvironmentSDFGIFramesToConverge sdfgi_frames_to_converge = RS::ENV_SDFGI_CONVERGE_IN_10_FRAMES; - float sdfgi_solid_cell_ratio = 0.25; - Vector3 sdfgi_debug_probe_pos; - Vector3 sdfgi_debug_probe_dir; - bool sdfgi_debug_probe_enabled = false; - Vector3i sdfgi_debug_probe_index; - - struct SDGIShader { - enum SDFGIPreprocessShaderVersion { - PRE_PROCESS_SCROLL, - PRE_PROCESS_SCROLL_OCCLUSION, - PRE_PROCESS_JUMP_FLOOD_INITIALIZE, - PRE_PROCESS_JUMP_FLOOD_INITIALIZE_HALF, - PRE_PROCESS_JUMP_FLOOD, - PRE_PROCESS_JUMP_FLOOD_OPTIMIZED, - PRE_PROCESS_JUMP_FLOOD_UPSCALE, - PRE_PROCESS_OCCLUSION, - PRE_PROCESS_STORE, - PRE_PROCESS_MAX - }; - - struct PreprocessPushConstant { - int32_t scroll[3]; - int32_t grid_size; - - int32_t probe_offset[3]; - int32_t step_size; - - int32_t half_size; - uint32_t occlusion_index; - int32_t cascade; - uint32_t pad; - }; - - SdfgiPreprocessShaderRD preprocess; - RID preprocess_shader; - RID preprocess_pipeline[PRE_PROCESS_MAX]; - - struct DebugPushConstant { - float grid_size[3]; - uint32_t max_cascades; - - int32_t screen_size[2]; - uint32_t use_occlusion; - float y_mult; - - float cam_extent[3]; - uint32_t probe_axis_size; - - float cam_transform[16]; - }; - - SdfgiDebugShaderRD debug; - RID debug_shader; - RID debug_shader_version; - RID debug_pipeline; - - enum ProbeDebugMode { - PROBE_DEBUG_PROBES, - PROBE_DEBUG_VISIBILITY, - PROBE_DEBUG_MAX - }; - - struct DebugProbesPushConstant { - float projection[16]; - - uint32_t band_power; - uint32_t sections_in_band; - uint32_t band_mask; - float section_arc; - - float grid_size[3]; - uint32_t cascade; - - uint32_t pad; - float y_mult; - int32_t probe_debug_index; - int32_t probe_axis_size; - }; - - SdfgiDebugProbesShaderRD debug_probes; - RID debug_probes_shader; - RID debug_probes_shader_version; - - PipelineCacheRD debug_probes_pipeline[PROBE_DEBUG_MAX]; - - struct Light { - float color[3]; - float energy; - - float direction[3]; - uint32_t has_shadow; - - float position[3]; - float attenuation; - - uint32_t type; - float spot_angle; - float spot_attenuation; - float radius; - - float shadow_color[4]; - }; - - struct DirectLightPushConstant { - float grid_size[3]; - uint32_t max_cascades; - - uint32_t cascade; - uint32_t light_count; - uint32_t process_offset; - uint32_t process_increment; - - int32_t probe_axis_size; - uint32_t multibounce; - float y_mult; - uint32_t pad; - }; - - enum { - DIRECT_LIGHT_MODE_STATIC, - DIRECT_LIGHT_MODE_DYNAMIC, - DIRECT_LIGHT_MODE_MAX - }; - SdfgiDirectLightShaderRD direct_light; - RID direct_light_shader; - RID direct_light_pipeline[DIRECT_LIGHT_MODE_MAX]; - - enum { - INTEGRATE_MODE_PROCESS, - INTEGRATE_MODE_STORE, - INTEGRATE_MODE_SCROLL, - INTEGRATE_MODE_SCROLL_STORE, - INTEGRATE_MODE_MAX - }; - struct IntegratePushConstant { - enum { - SKY_MODE_DISABLED, - SKY_MODE_COLOR, - SKY_MODE_SKY, - }; - - float grid_size[3]; - uint32_t max_cascades; - - uint32_t probe_axis_size; - uint32_t cascade; - uint32_t history_index; - uint32_t history_size; - - uint32_t ray_count; - float ray_bias; - int32_t image_size[2]; - - int32_t world_offset[3]; - uint32_t sky_mode; - - int32_t scroll[3]; - float sky_energy; - - float sky_color[3]; - float y_mult; - - uint32_t store_ambient_texture; - uint32_t pad[3]; - }; - - SdfgiIntegrateShaderRD integrate; - RID integrate_shader; - RID integrate_pipeline[INTEGRATE_MODE_MAX]; - - RID integrate_default_sky_uniform_set; - - } sdfgi_shader; - - void _sdfgi_erase(RenderBuffers *rb); - int _sdfgi_get_pending_region_data(RID p_render_buffers, int p_region, Vector3i &r_local_offset, Vector3i &r_local_size, AABB &r_bounds) const; - void _sdfgi_update_cascades(RID p_render_buffers); - /* GI */ - - struct GI { - struct SDFGIData { - float grid_size[3]; - uint32_t max_cascades; - - uint32_t use_occlusion; - int32_t probe_axis_size; - float probe_to_uvw; - float normal_bias; - - float lightprobe_tex_pixel_size[3]; - float energy; - - float lightprobe_uv_offset[3]; - float y_mult; - - float occlusion_clamp[3]; - uint32_t pad3; - - float occlusion_renormalize[3]; - uint32_t pad4; - - float cascade_probe_size[3]; - uint32_t pad5; - - struct ProbeCascadeData { - float position[3]; //offset of (0,0,0) in world coordinates - float to_probe; // 1/bounds * grid_size - int32_t probe_world_offset[3]; - float to_cell; // 1/bounds * grid_size - }; - - ProbeCascadeData cascades[SDFGI::MAX_CASCADES]; - }; - - struct GIProbeData { - float xform[16]; - float bounds[3]; - float dynamic_range; - - float bias; - float normal_bias; - uint32_t blend_ambient; - uint32_t texture_slot; - - float anisotropy_strength; - float ao; - float ao_size; - uint32_t mipmaps; - }; - - struct PushConstant { - int32_t screen_size[2]; - float z_near; - float z_far; - - float proj_info[4]; - - uint32_t max_giprobes; - uint32_t high_quality_vct; - uint32_t use_sdfgi; - uint32_t orthogonal; - - float ao_color[3]; - uint32_t pad; - - float cam_rotation[12]; - }; - - RID sdfgi_ubo; - enum { - MODE_MAX = 1 - }; - - GiShaderRD shader; - RID shader_version; - RID pipelines[MODE_MAX]; - } gi; - bool screen_space_roughness_limiter = false; float screen_space_roughness_limiter_amount = 0.25; float screen_space_roughness_limiter_limit = 0.18; @@ -1289,21 +444,29 @@ private: void _render_buffers_debug_draw(RID p_render_buffers, RID p_shadow_atlas); void _render_buffers_post_process_and_tonemap(RID p_render_buffers, RID p_environment, RID p_camera_effects, const CameraMatrix &p_projection); - void _sdfgi_debug_draw(RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform); /* Cluster */ struct Cluster { /* Scene State UBO */ - struct ReflectionData { //should always be 128 bytes + enum { + REFLECTION_AMBIENT_DISABLED = 0, + REFLECTION_AMBIENT_ENVIRONMENT = 1, + REFLECTION_AMBIENT_COLOR = 2, + }; + + struct ReflectionData { float box_extents[3]; float index; float box_offset[3]; uint32_t mask; - float params[4]; // intensity, 0, interior , boxproject float ambient[3]; // ambient color, + float intensity; + bool exterior; + bool box_project; uint32_t ambient_mode; + uint32_t pad; float local_matrix[16]; // up to here for spot and omni, rest is for directional }; @@ -1312,10 +475,15 @@ private: float inv_radius; float direction[3]; float size; - uint16_t attenuation_energy[2]; //16 bits attenuation, then energy - uint8_t color_specular[4]; //rgb color, a specular (8 bit unorm) - uint16_t cone_attenuation_angle[2]; // attenuation and angle, (16bit float) - uint8_t shadow_color_enabled[4]; //shadow rgb color, a>0.5 enabled (8bit unorm) + + float color[3]; + float attenuation; + + float inv_spot_attenuation; + float cos_spot_angle; + float specular_amount; + uint32_t shadow_enabled; + float atlas_rect[4]; // in omni, used for atlas uv, in spot, used for projector uv float shadow_matrix[16]; float shadow_bias; @@ -1379,31 +547,85 @@ private: float normal_fade; }; + template <class T> + struct InstanceSort { + float depth; + T *instance; + bool operator<(const InstanceSort &p_sort) const { + return depth < p_sort.depth; + } + }; + ReflectionData *reflections; + InstanceSort<ReflectionProbeInstance> *reflection_sort; uint32_t max_reflections; RID reflection_buffer; uint32_t max_reflection_probes_per_instance; + uint32_t reflection_count = 0; DecalData *decals; + InstanceSort<DecalInstance> *decal_sort; uint32_t max_decals; RID decal_buffer; + uint32_t decal_count; + + LightData *omni_lights; + LightData *spot_lights; - LightData *lights; + InstanceSort<LightInstance> *omni_light_sort; + InstanceSort<LightInstance> *spot_light_sort; uint32_t max_lights; - RID light_buffer; - RID *lights_instances; - Rect2i *lights_shadow_rect_cache; - uint32_t lights_shadow_rect_cache_count = 0; + RID omni_light_buffer; + RID spot_light_buffer; + uint32_t omni_light_count = 0; + uint32_t spot_light_count = 0; DirectionalLightData *directional_lights; uint32_t max_directional_lights; RID directional_light_buffer; - LightClusterBuilder builder; - } cluster; + struct RenderState { + RID render_buffers; + Transform cam_transform; + CameraMatrix cam_projection; + bool cam_ortogonal = false; + const PagedArray<GeometryInstance *> *instances = nullptr; + const PagedArray<RID> *lights = nullptr; + const PagedArray<RID> *reflection_probes = nullptr; + const PagedArray<RID> *gi_probes = nullptr; + const PagedArray<RID> *decals = nullptr; + const PagedArray<RID> *lightmaps = nullptr; + RID environment; + RID camera_effects; + RID shadow_atlas; + RID reflection_atlas; + RID reflection_probe; + int reflection_probe_pass = 0; + float screen_lod_threshold = 0.0; + + const RenderShadowData *render_shadows = nullptr; + int render_shadow_count = 0; + const RenderSDFGIData *render_sdfgi_regions = nullptr; + int render_sdfgi_region_count = 0; + const RenderSDFGIUpdateData *sdfgi_update_data = nullptr; + + uint32_t directional_light_count = 0; + uint32_t gi_probe_count = 0; + + LocalVector<int> cube_shadows; + LocalVector<int> shadows; + LocalVector<int> directional_shadows; + + bool depth_prepass_used; + } render_state; + struct VolumetricFog { + enum { + MAX_TEMPORAL_FRAMES = 16 + }; + uint32_t width = 0; uint32_t height = 0; uint32_t depth = 0; @@ -1412,6 +634,8 @@ private: float spread; RID light_density_map; + RID prev_light_density_map; + RID fog_map; RID uniform_set; RID uniform_set2; @@ -1419,6 +643,8 @@ private: RID sky_uniform_set; int last_shadow_filter = -1; + + Transform prev_cam_transform; }; enum { @@ -1430,7 +656,7 @@ private: }; struct VolumetricFogShader { - struct PushConstant { + struct ParamsUBO { float fog_frustum_size_begin[2]; float fog_frustum_size_end[2]; @@ -1448,13 +674,24 @@ private: float detail_spread; float gi_inject; uint32_t max_gi_probes; - uint32_t pad; + uint32_t cluster_type_size; + + float screen_size[2]; + uint32_t cluster_shift; + uint32_t cluster_width; + + uint32_t max_cluster_element_count_div_32; + uint32_t use_temporal_reprojection; + uint32_t temporal_frame; + float temporal_blend; float cam_rotation[12]; + float to_prev_view[16]; }; VolumetricFogShaderRD shader; + RID params_ubo; RID shader_version; RID pipelines[VOLUMETRIC_FOG_SHADER_MAX]; @@ -1462,9 +699,7 @@ private: uint32_t volumetric_fog_depth = 128; uint32_t volumetric_fog_size = 128; - bool volumetric_fog_filter_active = false; - uint32_t volumetric_fog_directional_shadow_shrink = 512; - uint32_t volumetric_fog_positional_shadow_shrink = 512; + bool volumetric_fog_filter_active = true; void _volumetric_fog_erase(RenderBuffers *rb); void _update_volumetric_fog(RID p_render_buffers, RID p_environment, const CameraMatrix &p_cam_projection, const Transform &p_cam_transform, RID p_shadow_atlas, int p_directional_light_count, bool p_use_directional_shadows, int p_positional_light_count, int p_gi_probe_count); @@ -1474,12 +709,16 @@ private: uint64_t scene_pass = 0; uint64_t shadow_atlas_realloc_tolerance_msec = 500; + /* !BAS! is this used anywhere? struct SDFGICosineNeighbour { uint32_t neighbour; float weight; }; + */ + + uint32_t max_cluster_elements = 512; - bool low_end = false; + void _render_shadow_pass(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0, float p_screen_lod_threshold = 0.0, bool p_open_pass = true, bool p_close_pass = true, bool p_clear_region = true); public: virtual Transform geometry_instance_get_transform(GeometryInstance *p_instance) = 0; @@ -1488,7 +727,7 @@ public: /* SHADOW ATLAS API */ RID shadow_atlas_create(); - void shadow_atlas_set_size(RID p_atlas, int p_size); + void shadow_atlas_set_size(RID p_atlas, int p_size, bool p_16_bits = false); void shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision); bool shadow_atlas_update_light(RID p_atlas, RID p_light_intance, float p_coverage, uint64_t p_light_version); _FORCE_INLINE_ bool shadow_atlas_owns_light_instance(RID p_atlas, RID p_light_intance) { @@ -1509,7 +748,7 @@ public: return Size2(atlas->size, atlas->size); } - void directional_shadow_atlas_set_size(int p_size); + void directional_shadow_atlas_set_size(int p_size, bool p_16_bits = false); int get_directional_light_shadow_size(RID p_light_intance); void set_directional_shadow_count(int p_count); @@ -1523,28 +762,26 @@ public: /* SDFGI UPDATE */ - int sdfgi_get_lightprobe_octahedron_size() const { return SDFGI::LIGHTPROBE_OCT_SIZE; } virtual void sdfgi_update(RID p_render_buffers, RID p_environment, const Vector3 &p_world_position); virtual int sdfgi_get_pending_region_count(RID p_render_buffers) const; virtual AABB sdfgi_get_pending_region_bounds(RID p_render_buffers, int p_region) const; virtual uint32_t sdfgi_get_pending_region_cascade(RID p_render_buffers, int p_region) const; - virtual void sdfgi_update_probes(RID p_render_buffers, RID p_environment, const Vector<RID> &p_directional_lights, const RID *p_positional_light_instances, uint32_t p_positional_light_count); RID sdfgi_get_ubo() const { return gi.sdfgi_ubo; } + /* SKY API */ - RID sky_create(); + virtual RID sky_allocate(); + virtual void sky_initialize(RID p_rid); + void sky_set_radiance_size(RID p_sky, int p_radiance_size); void sky_set_mode(RID p_sky, RS::SkyMode p_mode); void sky_set_material(RID p_sky, RID p_material); Ref<Image> sky_bake_panorama(RID p_sky, float p_energy, bool p_bake_irradiance, const Size2i &p_size); - RID sky_get_radiance_texture_rd(RID p_sky) const; - RID sky_get_radiance_uniform_set_rd(RID p_sky, RID p_shader, int p_set) const; - RID sky_get_material(RID p_sky) const; - /* ENVIRONMENT API */ - RID environment_create(); + virtual RID environment_allocate(); + virtual void environment_initialize(RID p_rid); void environment_set_background(RID p_env, RS::EnvironmentBG p_bg); void environment_set_sky(RID p_env, RID p_sky); @@ -1585,12 +822,10 @@ public: float environment_get_fog_height_density(RID p_env) const; float environment_get_fog_aerial_perspective(RID p_env) const; - void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, RS::EnvVolumetricFogShadowFilter p_shadow_filter); + void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount); virtual void environment_set_volumetric_fog_volume_size(int p_size, int p_depth); virtual void environment_set_volumetric_fog_filter_active(bool p_enable); - virtual void environment_set_volumetric_fog_directional_shadow_shrink_size(int p_shrink_size); - virtual void environment_set_volumetric_fog_positional_shadow_shrink_size(int p_shrink_size); void environment_set_ssr(RID p_env, bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance); void environment_set_ssao(RID p_env, bool p_enable, float p_radius, float p_intensity, float p_power, float p_detail, float p_horizon, float p_sharpness, float p_light_affect, float p_ao_channel_affect); @@ -1601,9 +836,10 @@ public: bool environment_is_ssr_enabled(RID p_env) const; bool environment_is_sdfgi_enabled(RID p_env) const; - virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, bool p_use_multibounce, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias); + virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias); virtual void environment_set_sdfgi_ray_count(RS::EnvironmentSDFGIRayCount p_ray_count); virtual void environment_set_sdfgi_frames_to_converge(RS::EnvironmentSDFGIFramesToConverge p_frames); + virtual void environment_set_sdfgi_frames_to_update_light(RS::EnvironmentSDFGIFramesToUpdateLight p_update); void environment_set_ssr_roughness_quality(RS::EnvironmentSSRRoughnessQuality p_quality); RS::EnvironmentSSRRoughnessQuality environment_get_ssr_roughness_quality() const; @@ -1613,7 +849,8 @@ public: virtual Ref<Image> environment_bake_panorama(RID p_env, bool p_bake_irradiance, const Size2i &p_size); - virtual RID camera_effects_create(); + virtual RID camera_effects_allocate(); + virtual void camera_effects_initialize(RID p_rid); virtual void camera_effects_set_dof_blur_quality(RS::DOFBlurQuality p_quality, bool p_use_jitter); virtual void camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokehShape p_shape); @@ -1849,64 +1086,26 @@ public: return li->transform; } + /* gi light probes */ + RID gi_probe_instance_create(RID p_base); void gi_probe_instance_set_transform_to_data(RID p_probe, const Transform &p_xform); bool gi_probe_needs_update(RID p_probe) const; void gi_probe_update(RID p_probe, bool p_update_light_instances, const Vector<RID> &p_light_instances, const PagedArray<RendererSceneRender::GeometryInstance *> &p_dynamic_objects); + void gi_probe_set_quality(RS::GIProbeQuality p_quality) { gi.gi_probe_quality = p_quality; } - void gi_probe_set_quality(RS::GIProbeQuality p_quality) { gi_probe_quality = p_quality; } + /* render buffers */ - _FORCE_INLINE_ uint32_t gi_probe_instance_get_slot(RID p_probe) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - return gi_probe->slot; - } - _FORCE_INLINE_ RID gi_probe_instance_get_base_probe(RID p_probe) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - return gi_probe->probe; - } - _FORCE_INLINE_ Transform gi_probe_instance_get_transform_to_cell(RID p_probe) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - return storage->gi_probe_get_to_cell_xform(gi_probe->probe) * gi_probe->transform.affine_inverse(); - } - - _FORCE_INLINE_ RID gi_probe_instance_get_texture(RID p_probe) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_probe); - return gi_probe->texture; - } - - _FORCE_INLINE_ void gi_probe_instance_set_render_index(RID p_instance, uint32_t p_render_index) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_instance); - ERR_FAIL_COND(!gi_probe); - gi_probe->render_index = p_render_index; - } - - _FORCE_INLINE_ uint32_t gi_probe_instance_get_render_index(RID p_instance) { - GIProbeInstance *gi_probe = gi_probe_instance_owner.getornull(p_instance); - ERR_FAIL_COND_V(!gi_probe, 0); - - return gi_probe->render_index; - } - /* - _FORCE_INLINE_ void gi_probe_instance_set_render_pass(RID p_instance, uint32_t p_render_pass) { - GIProbeInstance *g_probe = gi_probe_instance_owner.getornull(p_instance); - ERR_FAIL_COND(!g_probe); - g_probe->last_pass = p_render_pass; - } - - _FORCE_INLINE_ uint32_t gi_probe_instance_get_render_pass(RID p_instance) { - GIProbeInstance *g_probe = gi_probe_instance_owner.getornull(p_instance); - ERR_FAIL_COND_V(!g_probe, 0); - - return g_probe->last_pass; - } -*/ RID render_buffers_create(); void render_buffers_configure(RID p_render_buffers, RID p_render_target, int p_width, int p_height, RS::ViewportMSAA p_msaa, RS::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_debanding); + void gi_set_use_half_resolution(bool p_enable); RID render_buffers_get_ao_texture(RID p_render_buffers); RID render_buffers_get_back_buffer_texture(RID p_render_buffers); RID render_buffers_get_gi_probe_buffer(RID p_render_buffers); RID render_buffers_get_default_gi_probe_buffer(); + RID render_buffers_get_gi_ambient_texture(RID p_render_buffers); + RID render_buffers_get_gi_reflection_texture(RID p_render_buffers); uint32_t render_buffers_get_sdfgi_cascade_count(RID p_render_buffers) const; bool render_buffers_is_sdfgi_enabled(RID p_render_buffers) const; @@ -1927,15 +1126,10 @@ public: float render_buffers_get_volumetric_fog_end(RID p_render_buffers); float render_buffers_get_volumetric_fog_detail_spread(RID p_render_buffers); - void render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold); - - void render_shadow(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0, float p_screen_lod_threshold = 0.0); + void render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold, const RenderShadowData *p_render_shadows, int p_render_shadow_count, const RenderSDFGIData *p_render_sdfgi_regions, int p_render_sdfgi_region_count, const RenderSDFGIUpdateData *p_sdfgi_update_data = nullptr); void render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region); - void render_sdfgi(RID p_render_buffers, int p_region, const PagedArray<GeometryInstance *> &p_instances); - void render_sdfgi_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_light_cull_result); - void render_particle_collider_heightfield(RID p_collider, const Transform &p_transform, const PagedArray<GeometryInstance *> &p_instances); virtual void set_scene_pass(uint64_t p_pass) { @@ -1985,19 +1179,22 @@ public: return debug_draw; } - virtual void set_time(double p_time, double p_step); + void set_time(double p_time, double p_step); - RID get_cluster_builder_texture(); - RID get_cluster_builder_indices_buffer(); RID get_reflection_probe_buffer(); - RID get_positional_light_buffer(); + RID get_omni_light_buffer(); + RID get_spot_light_buffer(); RID get_directional_light_buffer(); RID get_decal_buffer(); int get_max_directional_lights() const; void sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir); - bool is_low_end() const; + virtual bool is_low_end() const; + + virtual bool is_dynamic_gi_supported() const; + virtual bool is_clustered_enabled() const; + virtual bool is_volumetric_supported() const; RendererSceneRenderRD(RendererStorageRD *p_storage); ~RendererSceneRenderRD(); diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp new file mode 100644 index 0000000000..769335ac16 --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp @@ -0,0 +1,1491 @@ +/*************************************************************************/ +/* renderer_scene_sky_rd.cpp */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#include "renderer_scene_sky_rd.h" +#include "core/config/project_settings.h" +#include "renderer_scene_render_rd.h" +#include "servers/rendering/rendering_server_default.h" + +//////////////////////////////////////////////////////////////////////////////// +// SKY SHADER + +void RendererSceneSkyRD::SkyShaderData::set_code(const String &p_code) { + //compile + + code = p_code; + valid = false; + ubo_size = 0; + uniforms.clear(); + + if (code == String()) { + return; //just invalid, but no error + } + + ShaderCompilerRD::GeneratedCode gen_code; + ShaderCompilerRD::IdentifierActions actions; + + uses_time = false; + uses_half_res = false; + uses_quarter_res = false; + uses_position = false; + uses_light = false; + + actions.render_mode_flags["use_half_res_pass"] = &uses_half_res; + actions.render_mode_flags["use_quarter_res_pass"] = &uses_quarter_res; + + actions.usage_flag_pointers["TIME"] = &uses_time; + actions.usage_flag_pointers["POSITION"] = &uses_position; + actions.usage_flag_pointers["LIGHT0_ENABLED"] = &uses_light; + actions.usage_flag_pointers["LIGHT0_ENERGY"] = &uses_light; + actions.usage_flag_pointers["LIGHT0_DIRECTION"] = &uses_light; + actions.usage_flag_pointers["LIGHT0_COLOR"] = &uses_light; + actions.usage_flag_pointers["LIGHT0_SIZE"] = &uses_light; + actions.usage_flag_pointers["LIGHT1_ENABLED"] = &uses_light; + actions.usage_flag_pointers["LIGHT1_ENERGY"] = &uses_light; + actions.usage_flag_pointers["LIGHT1_DIRECTION"] = &uses_light; + actions.usage_flag_pointers["LIGHT1_COLOR"] = &uses_light; + actions.usage_flag_pointers["LIGHT1_SIZE"] = &uses_light; + actions.usage_flag_pointers["LIGHT2_ENABLED"] = &uses_light; + actions.usage_flag_pointers["LIGHT2_ENERGY"] = &uses_light; + actions.usage_flag_pointers["LIGHT2_DIRECTION"] = &uses_light; + actions.usage_flag_pointers["LIGHT2_COLOR"] = &uses_light; + actions.usage_flag_pointers["LIGHT2_SIZE"] = &uses_light; + actions.usage_flag_pointers["LIGHT3_ENABLED"] = &uses_light; + actions.usage_flag_pointers["LIGHT3_ENERGY"] = &uses_light; + actions.usage_flag_pointers["LIGHT3_DIRECTION"] = &uses_light; + actions.usage_flag_pointers["LIGHT3_COLOR"] = &uses_light; + actions.usage_flag_pointers["LIGHT3_SIZE"] = &uses_light; + + actions.uniforms = &uniforms; + + // !BAS! Contemplate making `SkyShader sky` accessible from this struct or even part of this struct. + RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; + + Error err = scene_singleton->sky.sky_shader.compiler.compile(RS::SHADER_SKY, code, &actions, path, gen_code); + + ERR_FAIL_COND(err != OK); + + if (version.is_null()) { + version = scene_singleton->sky.sky_shader.shader.version_create(); + } + +#if 0 + print_line("**compiling shader:"); + print_line("**defines:\n"); + for (int i = 0; i < gen_code.defines.size(); i++) { + print_line(gen_code.defines[i]); + } + print_line("\n**uniforms:\n" + gen_code.uniforms); + // print_line("\n**vertex_globals:\n" + gen_code.vertex_global); + // print_line("\n**vertex_code:\n" + gen_code.vertex); + print_line("\n**fragment_globals:\n" + gen_code.fragment_global); + print_line("\n**fragment_code:\n" + gen_code.fragment); + print_line("\n**light_code:\n" + gen_code.light); +#endif + + scene_singleton->sky.sky_shader.shader.version_set_code(version, gen_code.uniforms, gen_code.vertex_global, gen_code.vertex, gen_code.fragment_global, gen_code.light, gen_code.fragment, gen_code.defines); + ERR_FAIL_COND(!scene_singleton->sky.sky_shader.shader.version_is_valid(version)); + + ubo_size = gen_code.uniform_total_size; + ubo_offsets = gen_code.uniform_offsets; + texture_uniforms = gen_code.texture_uniforms; + + //update pipelines + + for (int i = 0; i < SKY_VERSION_MAX; i++) { + RD::PipelineDepthStencilState depth_stencil_state; + depth_stencil_state.enable_depth_test = true; + depth_stencil_state.depth_compare_operator = RD::COMPARE_OP_LESS_OR_EQUAL; + + RID shader_variant = scene_singleton->sky.sky_shader.shader.version_get_shader(version, i); + pipelines[i].setup(shader_variant, RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), depth_stencil_state, RD::PipelineColorBlendState::create_disabled(), 0); + } + + valid = true; +} + +void RendererSceneSkyRD::SkyShaderData::set_default_texture_param(const StringName &p_name, RID p_texture) { + if (!p_texture.is_valid()) { + default_texture_params.erase(p_name); + } else { + default_texture_params[p_name] = p_texture; + } +} + +void RendererSceneSkyRD::SkyShaderData::get_param_list(List<PropertyInfo> *p_param_list) const { + Map<int, StringName> order; + + for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { + if (E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_GLOBAL || E->get().scope == ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + continue; + } + + if (E->get().texture_order >= 0) { + order[E->get().texture_order + 100000] = E->key(); + } else { + order[E->get().order] = E->key(); + } + } + + for (Map<int, StringName>::Element *E = order.front(); E; E = E->next()) { + PropertyInfo pi = ShaderLanguage::uniform_to_property_info(uniforms[E->get()]); + pi.name = E->get(); + p_param_list->push_back(pi); + } +} + +void RendererSceneSkyRD::SkyShaderData::get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const { + for (Map<StringName, ShaderLanguage::ShaderNode::Uniform>::Element *E = uniforms.front(); E; E = E->next()) { + if (E->get().scope != ShaderLanguage::ShaderNode::Uniform::SCOPE_INSTANCE) { + continue; + } + + RendererStorage::InstanceShaderParam p; + p.info = ShaderLanguage::uniform_to_property_info(E->get()); + p.info.name = E->key(); //supply name + p.index = E->get().instance_index; + p.default_value = ShaderLanguage::constant_value_to_variant(E->get().default_value, E->get().type, E->get().hint); + p_param_list->push_back(p); + } +} + +bool RendererSceneSkyRD::SkyShaderData::is_param_texture(const StringName &p_param) const { + if (!uniforms.has(p_param)) { + return false; + } + + return uniforms[p_param].texture_order >= 0; +} + +bool RendererSceneSkyRD::SkyShaderData::is_animated() const { + return false; +} + +bool RendererSceneSkyRD::SkyShaderData::casts_shadows() const { + return false; +} + +Variant RendererSceneSkyRD::SkyShaderData::get_default_parameter(const StringName &p_parameter) const { + if (uniforms.has(p_parameter)) { + ShaderLanguage::ShaderNode::Uniform uniform = uniforms[p_parameter]; + Vector<ShaderLanguage::ConstantNode::Value> default_value = uniform.default_value; + return ShaderLanguage::constant_value_to_variant(default_value, uniform.type, uniform.hint); + } + return Variant(); +} + +RS::ShaderNativeSourceCode RendererSceneSkyRD::SkyShaderData::get_native_source_code() const { + RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; + + return scene_singleton->sky.sky_shader.shader.version_get_native_source_code(version); +} + +RendererSceneSkyRD::SkyShaderData::SkyShaderData() { + valid = false; +} + +RendererSceneSkyRD::SkyShaderData::~SkyShaderData() { + RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; + ERR_FAIL_COND(!scene_singleton); + //pipeline variants will clear themselves if shader is gone + if (version.is_valid()) { + scene_singleton->sky.sky_shader.shader.version_free(version); + } +} + +//////////////////////////////////////////////////////////////////////////////// +// Sky material + +void RendererSceneSkyRD::SkyMaterialData::update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty) { + RendererSceneRenderRD *scene_singleton = (RendererSceneRenderRD *)RendererSceneRenderRD::singleton; + + uniform_set_updated = true; + + if ((uint32_t)ubo_data.size() != shader_data->ubo_size) { + p_uniform_dirty = true; + if (uniform_buffer.is_valid()) { + RD::get_singleton()->free(uniform_buffer); + uniform_buffer = RID(); + } + + ubo_data.resize(shader_data->ubo_size); + if (ubo_data.size()) { + uniform_buffer = RD::get_singleton()->uniform_buffer_create(ubo_data.size()); + memset(ubo_data.ptrw(), 0, ubo_data.size()); //clear + } + + //clear previous uniform set + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + uniform_set = RID(); + } + } + + //check whether buffer changed + if (p_uniform_dirty && ubo_data.size()) { + update_uniform_buffer(shader_data->uniforms, shader_data->ubo_offsets.ptr(), p_parameters, ubo_data.ptrw(), ubo_data.size(), false); + RD::get_singleton()->buffer_update(uniform_buffer, 0, ubo_data.size(), ubo_data.ptrw()); + } + + uint32_t tex_uniform_count = shader_data->texture_uniforms.size(); + + if ((uint32_t)texture_cache.size() != tex_uniform_count) { + texture_cache.resize(tex_uniform_count); + p_textures_dirty = true; + + //clear previous uniform set + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + uniform_set = RID(); + } + } + + if (p_textures_dirty && tex_uniform_count) { + update_textures(p_parameters, shader_data->default_texture_params, shader_data->texture_uniforms, texture_cache.ptrw(), true); + } + + if (shader_data->ubo_size == 0 && shader_data->texture_uniforms.size() == 0) { + // This material does not require an uniform set, so don't create it. + return; + } + + if (!p_textures_dirty && uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + //no reason to update uniform set, only UBO (or nothing) was needed to update + return; + } + + Vector<RD::Uniform> uniforms; + + { + if (shader_data->ubo_size) { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.binding = 0; + u.ids.push_back(uniform_buffer); + uniforms.push_back(u); + } + + const RID *textures = texture_cache.ptrw(); + for (uint32_t i = 0; i < tex_uniform_count; i++) { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 1 + i; + u.ids.push_back(textures[i]); + uniforms.push_back(u); + } + } + + uniform_set = RD::get_singleton()->uniform_set_create(uniforms, scene_singleton->sky.sky_shader.shader.version_get_shader(shader_data->version, 0), SKY_SET_MATERIAL); +} + +RendererSceneSkyRD::SkyMaterialData::~SkyMaterialData() { + if (uniform_set.is_valid() && RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + RD::get_singleton()->free(uniform_set); + } + + if (uniform_buffer.is_valid()) { + RD::get_singleton()->free(uniform_buffer); + } +} + +//////////////////////////////////////////////////////////////////////////////// +// ReflectionData + +void RendererSceneSkyRD::ReflectionData::clear_reflection_data() { + layers.clear(); + radiance_base_cubemap = RID(); + if (downsampled_radiance_cubemap.is_valid()) { + RD::get_singleton()->free(downsampled_radiance_cubemap); + } + downsampled_radiance_cubemap = RID(); + downsampled_layer.mipmaps.clear(); + coefficient_buffer = RID(); +} + +void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers) { + //recreate radiance and all data + + int mipmaps = p_mipmaps; + uint32_t w = p_size, h = p_size; + + if (p_use_array) { + int num_layers = p_low_quality ? 8 : p_roughness_layers; + + for (int i = 0; i < num_layers; i++) { + ReflectionData::Layer layer; + uint32_t mmw = w; + uint32_t mmh = h; + layer.mipmaps.resize(mipmaps); + layer.views.resize(mipmaps); + for (int j = 0; j < mipmaps; j++) { + ReflectionData::Layer::Mipmap &mm = layer.mipmaps.write[j]; + mm.size.width = mmw; + mm.size.height = mmh; + for (int k = 0; k < 6; k++) { + mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + i * 6 + k, j); + Vector<RID> fbtex; + fbtex.push_back(mm.views[k]); + mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); + } + + layer.views.write[j] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + i * 6, j, RD::TEXTURE_SLICE_CUBEMAP); + + mmw = MAX(1, mmw >> 1); + mmh = MAX(1, mmh >> 1); + } + + layers.push_back(layer); + } + + } else { + mipmaps = p_low_quality ? 8 : mipmaps; + //regular cubemap, lower quality (aliasing, less memory) + ReflectionData::Layer layer; + uint32_t mmw = w; + uint32_t mmh = h; + layer.mipmaps.resize(mipmaps); + layer.views.resize(mipmaps); + for (int j = 0; j < mipmaps; j++) { + ReflectionData::Layer::Mipmap &mm = layer.mipmaps.write[j]; + mm.size.width = mmw; + mm.size.height = mmh; + for (int k = 0; k < 6; k++) { + mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer + k, j); + Vector<RID> fbtex; + fbtex.push_back(mm.views[k]); + mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); + } + + layer.views.write[j] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer, j, RD::TEXTURE_SLICE_CUBEMAP); + + mmw = MAX(1, mmw >> 1); + mmh = MAX(1, mmh >> 1); + } + + layers.push_back(layer); + } + + radiance_base_cubemap = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), p_base_cube, p_base_layer, 0, RD::TEXTURE_SLICE_CUBEMAP); + + RD::TextureFormat tf; + tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.width = 64; // Always 64x64 + tf.height = 64; + tf.texture_type = RD::TEXTURE_TYPE_CUBE; + tf.array_layers = 6; + tf.mipmaps = 7; + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + + downsampled_radiance_cubemap = RD::get_singleton()->texture_create(tf, RD::TextureView()); + { + uint32_t mmw = 64; + uint32_t mmh = 64; + downsampled_layer.mipmaps.resize(7); + for (int j = 0; j < downsampled_layer.mipmaps.size(); j++) { + ReflectionData::DownsampleLayer::Mipmap &mm = downsampled_layer.mipmaps.write[j]; + mm.size.width = mmw; + mm.size.height = mmh; + mm.view = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), downsampled_radiance_cubemap, 0, j, RD::TEXTURE_SLICE_CUBEMAP); + + mmw = MAX(1, mmw >> 1); + mmh = MAX(1, mmh >> 1); + } + } +} + +void RendererSceneSkyRD::ReflectionData::create_reflection_fast_filter(RendererStorageRD *p_storage, bool p_use_arrays) { + p_storage->get_effects()->cubemap_downsample(radiance_base_cubemap, downsampled_layer.mipmaps[0].view, downsampled_layer.mipmaps[0].size); + + for (int i = 1; i < downsampled_layer.mipmaps.size(); i++) { + p_storage->get_effects()->cubemap_downsample(downsampled_layer.mipmaps[i - 1].view, downsampled_layer.mipmaps[i].view, downsampled_layer.mipmaps[i].size); + } + + Vector<RID> views; + if (p_use_arrays) { + for (int i = 1; i < layers.size(); i++) { + views.push_back(layers[i].views[0]); + } + } else { + for (int i = 1; i < layers[0].views.size(); i++) { + views.push_back(layers[0].views[i]); + } + } + + p_storage->get_effects()->cubemap_filter(downsampled_radiance_cubemap, views, p_use_arrays); +} + +void RendererSceneSkyRD::ReflectionData::create_reflection_importance_sample(RendererStorageRD *p_storage, bool p_use_arrays, int p_cube_side, int p_base_layer, uint32_t p_sky_ggx_samples_quality) { + if (p_use_arrays) { + //render directly to the layers + p_storage->get_effects()->cubemap_roughness(radiance_base_cubemap, layers[p_base_layer].views[0], p_cube_side, p_sky_ggx_samples_quality, float(p_base_layer) / (layers.size() - 1.0), layers[p_base_layer].mipmaps[0].size.x); + } else { + p_storage->get_effects()->cubemap_roughness( + layers[0].views[p_base_layer - 1], + layers[0].views[p_base_layer], + p_cube_side, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers[0].mipmaps.size() - 1.0), + layers[0].mipmaps[p_base_layer].size.x); + } +} + +void RendererSceneSkyRD::ReflectionData::update_reflection_mipmaps(RendererStorageRD *p_storage, int p_start, int p_end) { + for (int i = p_start; i < p_end; i++) { + for (int j = 0; j < layers[i].views.size() - 1; j++) { + RID view = layers[i].views[j]; + RID texture = layers[i].views[j + 1]; + Size2i size = layers[i].mipmaps[j + 1].size; + p_storage->get_effects()->cubemap_downsample(view, texture, size); + } + } +} + +//////////////////////////////////////////////////////////////////////////////// +// RendererSceneSkyRD::Sky + +void RendererSceneSkyRD::Sky::free(RendererStorageRD *p_storage) { + if (radiance.is_valid()) { + RD::get_singleton()->free(radiance); + radiance = RID(); + } + reflection.clear_reflection_data(); + + if (uniform_buffer.is_valid()) { + RD::get_singleton()->free(uniform_buffer); + uniform_buffer = RID(); + } + + if (half_res_pass.is_valid()) { + RD::get_singleton()->free(half_res_pass); + half_res_pass = RID(); + } + + if (quarter_res_pass.is_valid()) { + RD::get_singleton()->free(quarter_res_pass); + quarter_res_pass = RID(); + } + + if (material.is_valid()) { + p_storage->free(material); + } +} + +RID RendererSceneSkyRD::Sky::get_textures(RendererStorageRD *p_storage, SkyTextureSetVersion p_version, RID p_default_shader_rd) { + if (texture_uniform_sets[p_version].is_valid() && RD::get_singleton()->uniform_set_is_valid(texture_uniform_sets[p_version])) { + return texture_uniform_sets[p_version]; + } + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 0; + if (radiance.is_valid() && p_version <= SKY_TEXTURE_SET_QUARTER_RES) { + u.ids.push_back(radiance); + } else { + u.ids.push_back(p_storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 1; // half res + if (half_res_pass.is_valid() && p_version != SKY_TEXTURE_SET_HALF_RES && p_version != SKY_TEXTURE_SET_CUBEMAP_HALF_RES) { + if (p_version >= SKY_TEXTURE_SET_CUBEMAP) { + u.ids.push_back(reflection.layers[0].views[1]); + } else { + u.ids.push_back(half_res_pass); + } + } else { + if (p_version < SKY_TEXTURE_SET_CUBEMAP) { + u.ids.push_back(p_storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); + } else { + u.ids.push_back(p_storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); + } + } + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 2; // quarter res + if (quarter_res_pass.is_valid() && p_version != SKY_TEXTURE_SET_QUARTER_RES && p_version != SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES) { + if (p_version >= SKY_TEXTURE_SET_CUBEMAP) { + u.ids.push_back(reflection.layers[0].views[2]); + } else { + u.ids.push_back(quarter_res_pass); + } + } else { + if (p_version < SKY_TEXTURE_SET_CUBEMAP) { + u.ids.push_back(p_storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); + } else { + u.ids.push_back(p_storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); + } + } + uniforms.push_back(u); + } + + texture_uniform_sets[p_version] = RD::get_singleton()->uniform_set_create(uniforms, p_default_shader_rd, SKY_SET_TEXTURES); + return texture_uniform_sets[p_version]; +} + +bool RendererSceneSkyRD::Sky::set_radiance_size(int p_radiance_size) { + ERR_FAIL_COND_V(p_radiance_size < 32 || p_radiance_size > 2048, false); + if (radiance_size == p_radiance_size) { + return false; + } + radiance_size = p_radiance_size; + + if (mode == RS::SKY_MODE_REALTIME && radiance_size != 256) { + WARN_PRINT("Realtime Skies can only use a radiance size of 256. Radiance size will be set to 256 internally."); + radiance_size = 256; + } + + if (radiance.is_valid()) { + RD::get_singleton()->free(radiance); + radiance = RID(); + } + reflection.clear_reflection_data(); + + return true; +} + +bool RendererSceneSkyRD::Sky::set_mode(RS::SkyMode p_mode) { + if (mode == p_mode) { + return false; + } + + mode = p_mode; + + if (mode == RS::SKY_MODE_REALTIME && radiance_size != 256) { + WARN_PRINT("Realtime Skies can only use a radiance size of 256. Radiance size will be set to 256 internally."); + set_radiance_size(256); + } + + if (radiance.is_valid()) { + RD::get_singleton()->free(radiance); + radiance = RID(); + } + reflection.clear_reflection_data(); + + return true; +} + +bool RendererSceneSkyRD::Sky::set_material(RID p_material) { + if (material == p_material) { + return false; + } + + material = p_material; + return true; +} + +Ref<Image> RendererSceneSkyRD::Sky::bake_panorama(RendererStorageRD *p_storage, float p_energy, int p_roughness_layers, const Size2i &p_size) { + if (radiance.is_valid()) { + RD::TextureFormat tf; + tf.format = RD::DATA_FORMAT_R32G32B32A32_SFLOAT; + tf.width = p_size.width; + tf.height = p_size.height; + tf.usage_bits = RD::TEXTURE_USAGE_STORAGE_BIT | RD::TEXTURE_USAGE_CAN_COPY_FROM_BIT; + + RID rad_tex = RD::get_singleton()->texture_create(tf, RD::TextureView()); + p_storage->get_effects()->copy_cubemap_to_panorama(radiance, rad_tex, p_size, p_roughness_layers, reflection.layers.size() > 1); + Vector<uint8_t> data = RD::get_singleton()->texture_get_data(rad_tex, 0); + RD::get_singleton()->free(rad_tex); + + Ref<Image> img; + img.instance(); + img->create(p_size.width, p_size.height, false, Image::FORMAT_RGBAF, data); + for (int i = 0; i < p_size.width; i++) { + for (int j = 0; j < p_size.height; j++) { + Color c = img->get_pixel(i, j); + c.r *= p_energy; + c.g *= p_energy; + c.b *= p_energy; + img->set_pixel(i, j, c); + } + } + return img; + } + + return Ref<Image>(); +} + +//////////////////////////////////////////////////////////////////////////////// +// RendererSceneSkyRD + +RendererStorageRD::ShaderData *RendererSceneSkyRD::_create_sky_shader_func() { + SkyShaderData *shader_data = memnew(SkyShaderData); + return shader_data; +} + +RendererStorageRD::ShaderData *RendererSceneSkyRD::_create_sky_shader_funcs() { + // !BAS! Why isn't _create_sky_shader_func not just static too? + return static_cast<RendererSceneRenderRD *>(RendererSceneRenderRD::singleton)->sky._create_sky_shader_func(); +}; + +RendererStorageRD::MaterialData *RendererSceneSkyRD::_create_sky_material_func(SkyShaderData *p_shader) { + SkyMaterialData *material_data = memnew(SkyMaterialData); + material_data->shader_data = p_shader; + material_data->last_frame = false; + //update will happen later anyway so do nothing. + return material_data; +} + +RendererStorageRD::MaterialData *RendererSceneSkyRD::_create_sky_material_funcs(RendererStorageRD::ShaderData *p_shader) { + // !BAS! same here, we could just make _create_sky_material_func static? + return static_cast<RendererSceneRenderRD *>(RendererSceneRenderRD::singleton)->sky._create_sky_material_func(static_cast<SkyShaderData *>(p_shader)); +}; + +RendererSceneSkyRD::RendererSceneSkyRD() { + roughness_layers = GLOBAL_GET("rendering/reflections/sky_reflections/roughness_layers"); + sky_ggx_samples_quality = GLOBAL_GET("rendering/reflections/sky_reflections/ggx_samples"); + sky_use_cubemap_array = GLOBAL_GET("rendering/reflections/sky_reflections/texture_array_reflections"); +} + +void RendererSceneSkyRD::init(RendererStorageRD *p_storage) { + storage = p_storage; + + { + // Start with the directional lights for the sky + sky_scene_state.max_directional_lights = 4; + uint32_t directional_light_buffer_size = sky_scene_state.max_directional_lights * sizeof(SkyDirectionalLightData); + sky_scene_state.directional_lights = memnew_arr(SkyDirectionalLightData, sky_scene_state.max_directional_lights); + sky_scene_state.last_frame_directional_lights = memnew_arr(SkyDirectionalLightData, sky_scene_state.max_directional_lights); + sky_scene_state.last_frame_directional_light_count = sky_scene_state.max_directional_lights + 1; + sky_scene_state.directional_light_buffer = RD::get_singleton()->uniform_buffer_create(directional_light_buffer_size); + + String defines = "\n#define MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS " + itos(sky_scene_state.max_directional_lights) + "\n"; + + // Initialize sky + Vector<String> sky_modes; + sky_modes.push_back(""); // Full size + sky_modes.push_back("\n#define USE_HALF_RES_PASS\n"); // Half Res + sky_modes.push_back("\n#define USE_QUARTER_RES_PASS\n"); // Quarter res + sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n"); // Cubemap + sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n#define USE_HALF_RES_PASS\n"); // Half Res Cubemap + sky_modes.push_back("\n#define USE_CUBEMAP_PASS\n#define USE_QUARTER_RES_PASS\n"); // Quarter res Cubemap + sky_shader.shader.initialize(sky_modes, defines); + } + + // register our shader funds + storage->shader_set_data_request_function(RendererStorageRD::SHADER_TYPE_SKY, _create_sky_shader_funcs); + storage->material_set_data_request_function(RendererStorageRD::SHADER_TYPE_SKY, _create_sky_material_funcs); + + { + ShaderCompilerRD::DefaultIdentifierActions actions; + + actions.renames["COLOR"] = "color"; + actions.renames["ALPHA"] = "alpha"; + actions.renames["EYEDIR"] = "cube_normal"; + actions.renames["POSITION"] = "params.position_multiplier.xyz"; + actions.renames["SKY_COORDS"] = "panorama_coords"; + actions.renames["SCREEN_UV"] = "uv"; + actions.renames["TIME"] = "params.time"; + actions.renames["HALF_RES_COLOR"] = "half_res_color"; + actions.renames["QUARTER_RES_COLOR"] = "quarter_res_color"; + actions.renames["RADIANCE"] = "radiance"; + actions.renames["FOG"] = "custom_fog"; + actions.renames["LIGHT0_ENABLED"] = "directional_lights.data[0].enabled"; + actions.renames["LIGHT0_DIRECTION"] = "directional_lights.data[0].direction_energy.xyz"; + actions.renames["LIGHT0_ENERGY"] = "directional_lights.data[0].direction_energy.w"; + actions.renames["LIGHT0_COLOR"] = "directional_lights.data[0].color_size.xyz"; + actions.renames["LIGHT0_SIZE"] = "directional_lights.data[0].color_size.w"; + actions.renames["LIGHT1_ENABLED"] = "directional_lights.data[1].enabled"; + actions.renames["LIGHT1_DIRECTION"] = "directional_lights.data[1].direction_energy.xyz"; + actions.renames["LIGHT1_ENERGY"] = "directional_lights.data[1].direction_energy.w"; + actions.renames["LIGHT1_COLOR"] = "directional_lights.data[1].color_size.xyz"; + actions.renames["LIGHT1_SIZE"] = "directional_lights.data[1].color_size.w"; + actions.renames["LIGHT2_ENABLED"] = "directional_lights.data[2].enabled"; + actions.renames["LIGHT2_DIRECTION"] = "directional_lights.data[2].direction_energy.xyz"; + actions.renames["LIGHT2_ENERGY"] = "directional_lights.data[2].direction_energy.w"; + actions.renames["LIGHT2_COLOR"] = "directional_lights.data[2].color_size.xyz"; + actions.renames["LIGHT2_SIZE"] = "directional_lights.data[2].color_size.w"; + actions.renames["LIGHT3_ENABLED"] = "directional_lights.data[3].enabled"; + actions.renames["LIGHT3_DIRECTION"] = "directional_lights.data[3].direction_energy.xyz"; + actions.renames["LIGHT3_ENERGY"] = "directional_lights.data[3].direction_energy.w"; + actions.renames["LIGHT3_COLOR"] = "directional_lights.data[3].color_size.xyz"; + actions.renames["LIGHT3_SIZE"] = "directional_lights.data[3].color_size.w"; + actions.renames["AT_CUBEMAP_PASS"] = "AT_CUBEMAP_PASS"; + actions.renames["AT_HALF_RES_PASS"] = "AT_HALF_RES_PASS"; + actions.renames["AT_QUARTER_RES_PASS"] = "AT_QUARTER_RES_PASS"; + actions.custom_samplers["RADIANCE"] = "material_samplers[3]"; + actions.usage_defines["HALF_RES_COLOR"] = "\n#define USES_HALF_RES_COLOR\n"; + actions.usage_defines["QUARTER_RES_COLOR"] = "\n#define USES_QUARTER_RES_COLOR\n"; + actions.render_mode_defines["disable_fog"] = "#define DISABLE_FOG\n"; + + actions.sampler_array_name = "material_samplers"; + actions.base_texture_binding_index = 1; + actions.texture_layout_set = 1; + actions.base_uniform_string = "material."; + actions.base_varying_index = 10; + + actions.default_filter = ShaderLanguage::FILTER_LINEAR_MIPMAP; + actions.default_repeat = ShaderLanguage::REPEAT_ENABLE; + actions.global_buffer_array_variable = "global_variables.data"; + + sky_shader.compiler.initialize(actions); + } + + { + // default material and shader for sky shader + sky_shader.default_shader = storage->shader_allocate(); + storage->shader_initialize(sky_shader.default_shader); + + storage->shader_set_code(sky_shader.default_shader, "shader_type sky; void fragment() { COLOR = vec3(0.0); } \n"); + + sky_shader.default_material = storage->material_allocate(); + storage->material_initialize(sky_shader.default_material); + + storage->material_set_shader(sky_shader.default_material, sky_shader.default_shader); + + SkyMaterialData *md = (SkyMaterialData *)storage->material_get_data(sky_shader.default_material, RendererStorageRD::SHADER_TYPE_SKY); + sky_shader.default_shader_rd = sky_shader.shader.version_get_shader(md->shader_data->version, SKY_VERSION_BACKGROUND); + + sky_scene_state.uniform_buffer = RD::get_singleton()->uniform_buffer_create(sizeof(SkySceneState::UBO)); + + Vector<RD::Uniform> uniforms; + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_SAMPLER; + u.binding = 0; + u.ids.resize(12); + RID *ids_ptr = u.ids.ptrw(); + ids_ptr[0] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[1] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[2] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[3] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[4] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[5] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_DISABLED); + ids_ptr[6] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + ids_ptr[7] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + ids_ptr[8] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + ids_ptr[9] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + ids_ptr[10] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_NEAREST_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + ids_ptr[11] = storage->sampler_rd_get_default(RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC, RS::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 1; + u.ids.push_back(storage->global_variables_get_storage_buffer()); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.binding = 2; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(sky_scene_state.uniform_buffer); + uniforms.push_back(u); + } + + { + RD::Uniform u; + u.binding = 3; + u.uniform_type = RD::UNIFORM_TYPE_UNIFORM_BUFFER; + u.ids.push_back(sky_scene_state.directional_light_buffer); + uniforms.push_back(u); + } + + sky_scene_state.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_UNIFORMS); + } + + { + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.binding = 0; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + RID vfog = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); + u.ids.push_back(vfog); + uniforms.push_back(u); + } + + sky_scene_state.default_fog_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_FOG); + } + + { + // Need defaults for using fog with clear color + sky_scene_state.fog_shader = storage->shader_allocate(); + storage->shader_initialize(sky_scene_state.fog_shader); + + storage->shader_set_code(sky_scene_state.fog_shader, "shader_type sky; uniform vec4 clear_color; void fragment() { COLOR = clear_color.rgb; } \n"); + sky_scene_state.fog_material = storage->material_allocate(); + storage->material_initialize(sky_scene_state.fog_material); + + storage->material_set_shader(sky_scene_state.fog_material, sky_scene_state.fog_shader); + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 0; + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_CUBEMAP_BLACK)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 1; + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); + uniforms.push_back(u); + } + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_TEXTURE; + u.binding = 2; + u.ids.push_back(storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE)); + uniforms.push_back(u); + } + + sky_scene_state.fog_only_texture_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, sky_shader.default_shader_rd, SKY_SET_TEXTURES); + } +} + +void RendererSceneSkyRD::setup(RendererSceneEnvironmentRD *p_env, RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform, const Size2i p_screen_size, RendererSceneRenderRD *p_scene_render) { + ERR_FAIL_COND(!p_env); // I guess without an environment we also can't have a sky... + + SkyMaterialData *material = nullptr; + Sky *sky = get_sky(p_env->sky); + + RID sky_material; + + SkyShaderData *shader_data = nullptr; + + RS::EnvironmentBG background = p_env->background; + + if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { + // !BAS! Possibly silently fail here, we now get error spam when you select sky as the background but haven't setup the sky yet. + ERR_FAIL_COND(!sky); + sky_material = sky_get_material(p_env->sky); + + if (sky_material.is_valid()) { + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + if (!material || !material->shader_data->valid) { + material = nullptr; + } + } + + if (!material) { + sky_material = sky_shader.default_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + + ERR_FAIL_COND(!material); + + shader_data = material->shader_data; + + ERR_FAIL_COND(!shader_data); + } + + if (sky) { + // Invalidate supbass buffers if screen size changes + if (sky->screen_size != p_screen_size) { + sky->screen_size = p_screen_size; + sky->screen_size.x = sky->screen_size.x < 4 ? 4 : sky->screen_size.x; + sky->screen_size.y = sky->screen_size.y < 4 ? 4 : sky->screen_size.y; + if (shader_data->uses_half_res) { + if (sky->half_res_pass.is_valid()) { + RD::get_singleton()->free(sky->half_res_pass); + sky->half_res_pass = RID(); + } + invalidate_sky(sky); + } + if (shader_data->uses_quarter_res) { + if (sky->quarter_res_pass.is_valid()) { + RD::get_singleton()->free(sky->quarter_res_pass); + sky->quarter_res_pass = RID(); + } + invalidate_sky(sky); + } + } + + // Create new subpass buffers if necessary + if ((shader_data->uses_half_res && sky->half_res_pass.is_null()) || + (shader_data->uses_quarter_res && sky->quarter_res_pass.is_null()) || + sky->radiance.is_null()) { + invalidate_sky(sky); + update_dirty_skys(); + } + + if (shader_data->uses_time && p_scene_render->time - sky->prev_time > 0.00001) { + sky->prev_time = p_scene_render->time; + sky->reflection.dirty = true; + RenderingServerDefault::redraw_request(); + } + + if (material != sky->prev_material) { + sky->prev_material = material; + sky->reflection.dirty = true; + } + + if (material->uniform_set_updated) { + material->uniform_set_updated = false; + sky->reflection.dirty = true; + } + + if (!p_transform.origin.is_equal_approx(sky->prev_position) && shader_data->uses_position) { + sky->prev_position = p_transform.origin; + sky->reflection.dirty = true; + } + + if (shader_data->uses_light) { + // Check whether the directional_light_buffer changes + bool light_data_dirty = false; + + if (sky_scene_state.ubo.directional_light_count != sky_scene_state.last_frame_directional_light_count) { + light_data_dirty = true; + for (uint32_t i = sky_scene_state.ubo.directional_light_count; i < sky_scene_state.max_directional_lights; i++) { + sky_scene_state.directional_lights[i].enabled = false; + } + } + if (!light_data_dirty) { + for (uint32_t i = 0; i < sky_scene_state.ubo.directional_light_count; i++) { + if (sky_scene_state.directional_lights[i].direction[0] != sky_scene_state.last_frame_directional_lights[i].direction[0] || + sky_scene_state.directional_lights[i].direction[1] != sky_scene_state.last_frame_directional_lights[i].direction[1] || + sky_scene_state.directional_lights[i].direction[2] != sky_scene_state.last_frame_directional_lights[i].direction[2] || + sky_scene_state.directional_lights[i].energy != sky_scene_state.last_frame_directional_lights[i].energy || + sky_scene_state.directional_lights[i].color[0] != sky_scene_state.last_frame_directional_lights[i].color[0] || + sky_scene_state.directional_lights[i].color[1] != sky_scene_state.last_frame_directional_lights[i].color[1] || + sky_scene_state.directional_lights[i].color[2] != sky_scene_state.last_frame_directional_lights[i].color[2] || + sky_scene_state.directional_lights[i].enabled != sky_scene_state.last_frame_directional_lights[i].enabled || + sky_scene_state.directional_lights[i].size != sky_scene_state.last_frame_directional_lights[i].size) { + light_data_dirty = true; + break; + } + } + } + + if (light_data_dirty) { + RD::get_singleton()->buffer_update(sky_scene_state.directional_light_buffer, 0, sizeof(SkyDirectionalLightData) * sky_scene_state.max_directional_lights, sky_scene_state.directional_lights); + + SkyDirectionalLightData *temp = sky_scene_state.last_frame_directional_lights; + sky_scene_state.last_frame_directional_lights = sky_scene_state.directional_lights; + sky_scene_state.directional_lights = temp; + sky_scene_state.last_frame_directional_light_count = sky_scene_state.ubo.directional_light_count; + sky->reflection.dirty = true; + } + } + } + + //setup fog variables + sky_scene_state.ubo.volumetric_fog_enabled = false; + if (p_render_buffers.is_valid()) { + if (p_scene_render->render_buffers_has_volumetric_fog(p_render_buffers)) { + sky_scene_state.ubo.volumetric_fog_enabled = true; + + float fog_end = p_scene_render->render_buffers_get_volumetric_fog_end(p_render_buffers); + if (fog_end > 0.0) { + sky_scene_state.ubo.volumetric_fog_inv_length = 1.0 / fog_end; + } else { + sky_scene_state.ubo.volumetric_fog_inv_length = 1.0; + } + + float fog_detail_spread = p_scene_render->render_buffers_get_volumetric_fog_detail_spread(p_render_buffers); //reverse lookup + if (fog_detail_spread > 0.0) { + sky_scene_state.ubo.volumetric_fog_detail_spread = 1.0 / fog_detail_spread; + } else { + sky_scene_state.ubo.volumetric_fog_detail_spread = 1.0; + } + } + + RID fog_uniform_set = p_scene_render->render_buffers_get_volumetric_fog_sky_uniform_set(p_render_buffers); + + if (fog_uniform_set != RID()) { + sky_scene_state.fog_uniform_set = fog_uniform_set; + } else { + sky_scene_state.fog_uniform_set = sky_scene_state.default_fog_uniform_set; + } + } + + sky_scene_state.ubo.z_far = p_projection.get_z_far(); + sky_scene_state.ubo.fog_enabled = p_env->fog_enabled; + sky_scene_state.ubo.fog_density = p_env->fog_density; + sky_scene_state.ubo.fog_aerial_perspective = p_env->fog_aerial_perspective; + Color fog_color = p_env->fog_light_color.to_linear(); + float fog_energy = p_env->fog_light_energy; + sky_scene_state.ubo.fog_light_color[0] = fog_color.r * fog_energy; + sky_scene_state.ubo.fog_light_color[1] = fog_color.g * fog_energy; + sky_scene_state.ubo.fog_light_color[2] = fog_color.b * fog_energy; + sky_scene_state.ubo.fog_sun_scatter = p_env->fog_sun_scatter; + + RD::get_singleton()->buffer_update(sky_scene_state.uniform_buffer, 0, sizeof(SkySceneState::UBO), &sky_scene_state.ubo); +} + +void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform &p_transform, double p_time) { + ERR_FAIL_COND(!p_env); + + Sky *sky = get_sky(p_env->sky); + ERR_FAIL_COND(!sky); + + RID sky_material = sky_get_material(p_env->sky); + + SkyMaterialData *material = nullptr; + + if (sky_material.is_valid()) { + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + if (!material || !material->shader_data->valid) { + material = nullptr; + } + } + + if (!material) { + sky_material = sky_shader.default_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + + ERR_FAIL_COND(!material); + + SkyShaderData *shader_data = material->shader_data; + + ERR_FAIL_COND(!shader_data); + + float multiplier = p_env->bg_energy; + + bool update_single_frame = sky->mode == RS::SKY_MODE_REALTIME || sky->mode == RS::SKY_MODE_QUALITY; + RS::SkyMode sky_mode = sky->mode; + + if (sky_mode == RS::SKY_MODE_AUTOMATIC) { + if (shader_data->uses_time || shader_data->uses_position) { + update_single_frame = true; + sky_mode = RS::SKY_MODE_REALTIME; + } else if (shader_data->uses_light || shader_data->ubo_size > 0) { + update_single_frame = false; + sky_mode = RS::SKY_MODE_INCREMENTAL; + } else { + update_single_frame = true; + sky_mode = RS::SKY_MODE_QUALITY; + } + } + + if (sky->processing_layer == 0 && sky_mode == RS::SKY_MODE_INCREMENTAL) { + // On the first frame after creating sky, rebuild in single frame + update_single_frame = true; + sky_mode = RS::SKY_MODE_QUALITY; + } + + int max_processing_layer = sky_use_cubemap_array ? sky->reflection.layers.size() : sky->reflection.layers[0].mipmaps.size(); + + // Update radiance cubemap + if (sky->reflection.dirty && (sky->processing_layer >= max_processing_layer || update_single_frame)) { + static const Vector3 view_normals[6] = { + Vector3(+1, 0, 0), + Vector3(-1, 0, 0), + Vector3(0, +1, 0), + Vector3(0, -1, 0), + Vector3(0, 0, +1), + Vector3(0, 0, -1) + }; + static const Vector3 view_up[6] = { + Vector3(0, -1, 0), + Vector3(0, -1, 0), + Vector3(0, 0, +1), + Vector3(0, 0, -1), + Vector3(0, -1, 0), + Vector3(0, -1, 0) + }; + + CameraMatrix cm; + cm.set_perspective(90, 1, 0.01, 10.0); + CameraMatrix correction; + correction.set_depth_correction(true); + cm = correction * cm; + + if (shader_data->uses_quarter_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_QUARTER_RES]; + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + RD::DrawListID cubemap_draw_list; + + for (int i = 0; i < 6; i++) { + Transform local_view; + local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES, sky_shader.default_shader_rd); + + cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[2].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + storage->get_effects()->render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[2].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + } + + if (shader_data->uses_half_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_HALF_RES]; + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + RD::DrawListID cubemap_draw_list; + + for (int i = 0; i < 6; i++) { + Transform local_view; + local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP_HALF_RES, sky_shader.default_shader_rd); + + cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[1].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + storage->get_effects()->render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[1].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + } + + RD::DrawListID cubemap_draw_list; + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP]; + + for (int i = 0; i < 6; i++) { + Transform local_view; + local_view.set_look_at(Vector3(0, 0, 0), view_normals[i], view_up[i]); + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP, sky_shader.default_shader_rd); + + cubemap_draw_list = RD::get_singleton()->draw_list_begin(sky->reflection.layers[0].mipmaps[0].framebuffers[i], RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + storage->get_effects()->render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[0].framebuffers[i], sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, cm, local_view.basis, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + + if (sky_mode == RS::SKY_MODE_REALTIME) { + sky->reflection.create_reflection_fast_filter(storage, sky_use_cubemap_array); + if (sky_use_cubemap_array) { + sky->reflection.update_reflection_mipmaps(storage, 0, sky->reflection.layers.size()); + } + } else { + if (update_single_frame) { + for (int i = 1; i < max_processing_layer; i++) { + sky->reflection.create_reflection_importance_sample(storage, sky_use_cubemap_array, 10, i, sky_ggx_samples_quality); + } + if (sky_use_cubemap_array) { + sky->reflection.update_reflection_mipmaps(storage, 0, sky->reflection.layers.size()); + } + } else { + if (sky_use_cubemap_array) { + // Multi-Frame so just update the first array level + sky->reflection.update_reflection_mipmaps(storage, 0, 1); + } + } + sky->processing_layer = 1; + } + + sky->reflection.dirty = false; + + } else { + if (sky_mode == RS::SKY_MODE_INCREMENTAL && sky->processing_layer < max_processing_layer) { + sky->reflection.create_reflection_importance_sample(storage, sky_use_cubemap_array, 10, sky->processing_layer, sky_ggx_samples_quality); + + if (sky_use_cubemap_array) { + sky->reflection.update_reflection_mipmaps(storage, sky->processing_layer, sky->processing_layer + 1); + } + + sky->processing_layer++; + } + } +} + +void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, const CameraMatrix &p_projection, const Transform &p_transform, double p_time) { + ERR_FAIL_COND(!p_env); + + Sky *sky = get_sky(p_env->sky); + ERR_FAIL_COND(!sky); + + SkyMaterialData *material = nullptr; + RID sky_material; + + RS::EnvironmentBG background = p_env->background; + + if (!(background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) || sky) { + ERR_FAIL_COND(!sky); + sky_material = sky_get_material(p_env->sky); + + if (sky_material.is_valid()) { + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + if (!material || !material->shader_data->valid) { + material = nullptr; + } + } + + if (!material) { + sky_material = sky_shader.default_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + } + + if (background == RS::ENV_BG_CLEAR_COLOR || background == RS::ENV_BG_COLOR) { + sky_material = sky_scene_state.fog_material; + material = (SkyMaterialData *)storage->material_get_data(sky_material, RendererStorageRD::SHADER_TYPE_SKY); + } + + ERR_FAIL_COND(!material); + + SkyShaderData *shader_data = material->shader_data; + + ERR_FAIL_COND(!shader_data); + + Basis sky_transform = p_env->sky_orientation; + sky_transform.invert(); + + float multiplier = p_env->bg_energy; + float custom_fov = p_env->sky_custom_fov; + // Camera + CameraMatrix camera; + + if (custom_fov) { + float near_plane = p_projection.get_z_near(); + float far_plane = p_projection.get_z_far(); + float aspect = p_projection.get_aspect(); + + camera.set_perspective(custom_fov, aspect, near_plane, far_plane); + + } else { + camera = p_projection; + } + + sky_transform = p_transform.basis * sky_transform; + + if (shader_data->uses_quarter_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_QUARTER_RES]; + + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_QUARTER_RES, sky_shader.default_shader_rd); + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->quarter_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); + storage->get_effects()->render_sky(draw_list, p_time, sky->quarter_res_framebuffer, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + + if (shader_data->uses_half_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_HALF_RES]; + + RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_HALF_RES, sky_shader.default_shader_rd); + + Vector<Color> clear_colors; + clear_colors.push_back(Color(0.0, 0.0, 0.0)); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(sky->half_res_framebuffer, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, RD::FINAL_ACTION_DISCARD, clear_colors); + storage->get_effects()->render_sky(draw_list, p_time, sky->half_res_framebuffer, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + + PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_BACKGROUND]; + + RID texture_uniform_set; + if (sky) { + texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_BACKGROUND, sky_shader.default_shader_rd); + } else { + texture_uniform_set = sky_scene_state.fog_only_texture_uniform_set; + } + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_fb, RD::INITIAL_ACTION_CONTINUE, p_can_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CONTINUE, p_can_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ); + storage->get_effects()->render_sky(draw_list, p_time, p_fb, sky_scene_state.uniform_set, sky_scene_state.fog_uniform_set, pipeline, material->uniform_set, texture_uniform_set, camera, sky_transform, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); +} + +void RendererSceneSkyRD::invalidate_sky(Sky *p_sky) { + if (!p_sky->dirty) { + p_sky->dirty = true; + p_sky->dirty_list = dirty_sky_list; + dirty_sky_list = p_sky; + } +} + +void RendererSceneSkyRD::update_dirty_skys() { + Sky *sky = dirty_sky_list; + + while (sky) { + bool texture_set_dirty = false; + //update sky configuration if texture is missing + + if (sky->radiance.is_null()) { + int mipmaps = Image::get_image_required_mipmaps(sky->radiance_size, sky->radiance_size, Image::FORMAT_RGBAH) + 1; + + uint32_t w = sky->radiance_size, h = sky->radiance_size; + int layers = roughness_layers; + if (sky->mode == RS::SKY_MODE_REALTIME) { + layers = 8; + if (roughness_layers != 8) { + WARN_PRINT("When using REALTIME skies, roughness_layers should be set to 8 in the project settings for best quality reflections"); + } + } + + if (sky_use_cubemap_array) { + //array (higher quality, 6 times more memory) + RD::TextureFormat tf; + tf.array_layers = layers * 6; + tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.texture_type = RD::TEXTURE_TYPE_CUBE_ARRAY; + tf.mipmaps = mipmaps; + tf.width = w; + tf.height = h; + tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + + sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + sky->reflection.update_reflection_data(sky->radiance_size, mipmaps, true, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers); + + } else { + //regular cubemap, lower quality (aliasing, less memory) + RD::TextureFormat tf; + tf.array_layers = 6; + tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.texture_type = RD::TEXTURE_TYPE_CUBE; + tf.mipmaps = MIN(mipmaps, layers); + tf.width = w; + tf.height = h; + tf.usage_bits = RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT | RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_STORAGE_BIT; + + sky->radiance = RD::get_singleton()->texture_create(tf, RD::TextureView()); + + sky->reflection.update_reflection_data(sky->radiance_size, MIN(mipmaps, layers), false, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers); + } + texture_set_dirty = true; + } + + // Create subpass buffers if they haven't been created already + if (sky->half_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->half_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { + RD::TextureFormat tformat; + tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tformat.width = sky->screen_size.x / 2; + tformat.height = sky->screen_size.y / 2; + tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tformat.texture_type = RD::TEXTURE_TYPE_2D; + + sky->half_res_pass = RD::get_singleton()->texture_create(tformat, RD::TextureView()); + Vector<RID> texs; + texs.push_back(sky->half_res_pass); + sky->half_res_framebuffer = RD::get_singleton()->framebuffer_create(texs); + texture_set_dirty = true; + } + + if (sky->quarter_res_pass.is_null() && !RD::get_singleton()->texture_is_valid(sky->quarter_res_pass) && sky->screen_size.x >= 4 && sky->screen_size.y >= 4) { + RD::TextureFormat tformat; + tformat.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tformat.width = sky->screen_size.x / 4; + tformat.height = sky->screen_size.y / 4; + tformat.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tformat.texture_type = RD::TEXTURE_TYPE_2D; + + sky->quarter_res_pass = RD::get_singleton()->texture_create(tformat, RD::TextureView()); + Vector<RID> texs; + texs.push_back(sky->quarter_res_pass); + sky->quarter_res_framebuffer = RD::get_singleton()->framebuffer_create(texs); + texture_set_dirty = true; + } + + if (texture_set_dirty) { + for (int i = 0; i < SKY_TEXTURE_SET_MAX; i++) { + if (sky->texture_uniform_sets[i].is_valid() && RD::get_singleton()->uniform_set_is_valid(sky->texture_uniform_sets[i])) { + RD::get_singleton()->free(sky->texture_uniform_sets[i]); + sky->texture_uniform_sets[i] = RID(); + } + } + } + + sky->reflection.dirty = true; + sky->processing_layer = 0; + + Sky *next = sky->dirty_list; + sky->dirty_list = nullptr; + sky->dirty = false; + sky = next; + } + + dirty_sky_list = nullptr; +} + +RID RendererSceneSkyRD::sky_get_material(RID p_sky) const { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND_V(!sky, RID()); + + return sky->material; +} + +RID RendererSceneSkyRD::allocate_sky_rid() { + return sky_owner.allocate_rid(); +} + +void RendererSceneSkyRD::initialize_sky_rid(RID p_rid) { + sky_owner.initialize_rid(p_rid, Sky()); +} + +RendererSceneSkyRD::Sky *RendererSceneSkyRD::get_sky(RID p_sky) const { + return sky_owner.getornull(p_sky); +} + +void RendererSceneSkyRD::free_sky(RID p_sky) { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND(!sky); + + sky->free(storage); + sky_owner.free(p_sky); +} + +void RendererSceneSkyRD::sky_set_radiance_size(RID p_sky, int p_radiance_size) { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND(!sky); + + if (sky->set_radiance_size(p_radiance_size)) { + invalidate_sky(sky); + } +} + +void RendererSceneSkyRD::sky_set_mode(RID p_sky, RS::SkyMode p_mode) { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND(!sky); + + if (sky->set_mode(p_mode)) { + invalidate_sky(sky); + } +} + +void RendererSceneSkyRD::sky_set_material(RID p_sky, RID p_material) { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND(!sky); + + if (sky->set_material(p_material)) { + invalidate_sky(sky); + } +} + +Ref<Image> RendererSceneSkyRD::sky_bake_panorama(RID p_sky, float p_energy, bool p_bake_irradiance, const Size2i &p_size) { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND_V(!sky, Ref<Image>()); + + update_dirty_skys(); + + return sky->bake_panorama(storage, p_energy, p_bake_irradiance ? roughness_layers : 0, p_size); +} + +RID RendererSceneSkyRD::sky_get_radiance_texture_rd(RID p_sky) const { + Sky *sky = get_sky(p_sky); + ERR_FAIL_COND_V(!sky, RID()); + + return sky->radiance; +} diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.h b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h new file mode 100644 index 0000000000..73390a586b --- /dev/null +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h @@ -0,0 +1,292 @@ +/*************************************************************************/ +/* renderer_scene_sky_rd.h */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#ifndef RENDERING_SERVER_SCENE_SKY_RD_H +#define RENDERING_SERVER_SCENE_SKY_RD_H + +#include "core/templates/rid_owner.h" +#include "servers/rendering/renderer_compositor.h" +#include "servers/rendering/renderer_rd/renderer_scene_environment_rd.h" +#include "servers/rendering/renderer_rd/renderer_storage_rd.h" +#include "servers/rendering/renderer_rd/shaders/sky.glsl.gen.h" +#include "servers/rendering/renderer_scene_render.h" +#include "servers/rendering/rendering_device.h" + +// Forward declare RendererSceneRenderRD so we can pass it into some of our methods, these classes are pretty tightly bound +class RendererSceneRenderRD; + +class RendererSceneSkyRD { +private: + RendererStorageRD *storage; + +public: + enum SkySet { + SKY_SET_UNIFORMS, + SKY_SET_MATERIAL, + SKY_SET_TEXTURES, + SKY_SET_FOG, + SKY_SET_MAX + }; + + enum SkyTextureSetVersion { + SKY_TEXTURE_SET_BACKGROUND, + SKY_TEXTURE_SET_HALF_RES, + SKY_TEXTURE_SET_QUARTER_RES, + SKY_TEXTURE_SET_CUBEMAP, + SKY_TEXTURE_SET_CUBEMAP_HALF_RES, + SKY_TEXTURE_SET_CUBEMAP_QUARTER_RES, + SKY_TEXTURE_SET_MAX + }; + + enum SkyVersion { + SKY_VERSION_BACKGROUND, + SKY_VERSION_HALF_RES, + SKY_VERSION_QUARTER_RES, + SKY_VERSION_CUBEMAP, + SKY_VERSION_CUBEMAP_HALF_RES, + SKY_VERSION_CUBEMAP_QUARTER_RES, + SKY_VERSION_MAX + }; + + // Skys need less info from Directional Lights than the normal shaders + struct SkyDirectionalLightData { + float direction[3]; + float energy; + float color[3]; + float size; + uint32_t enabled; + uint32_t pad[3]; + }; + + struct SkySceneState { + struct UBO { + uint32_t volumetric_fog_enabled; + float volumetric_fog_inv_length; + float volumetric_fog_detail_spread; + + float fog_aerial_perspective; + + float fog_light_color[3]; + float fog_sun_scatter; + + uint32_t fog_enabled; + float fog_density; + + float z_far; + uint32_t directional_light_count; + }; + + UBO ubo; + + SkyDirectionalLightData *directional_lights; + SkyDirectionalLightData *last_frame_directional_lights; + uint32_t max_directional_lights; + uint32_t last_frame_directional_light_count; + RID directional_light_buffer; + RID uniform_set; + RID uniform_buffer; + RID fog_uniform_set; + RID default_fog_uniform_set; + + RID fog_shader; + RID fog_material; + RID fog_only_texture_uniform_set; + } sky_scene_state; + + struct ReflectionData { + struct Layer { + struct Mipmap { + RID framebuffers[6]; + RID views[6]; + Size2i size; + }; + Vector<Mipmap> mipmaps; //per-face view + Vector<RID> views; // per-cubemap view + }; + + struct DownsampleLayer { + struct Mipmap { + RID view; + Size2i size; + }; + Vector<Mipmap> mipmaps; + }; + + RID radiance_base_cubemap; //cubemap for first layer, first cubemap + RID downsampled_radiance_cubemap; + DownsampleLayer downsampled_layer; + RID coefficient_buffer; + + bool dirty = true; + + Vector<Layer> layers; + + void clear_reflection_data(); + void update_reflection_data(int p_size, int p_mipmaps, bool p_use_array, RID p_base_cube, int p_base_layer, bool p_low_quality, int p_roughness_layers); + void create_reflection_fast_filter(RendererStorageRD *p_storage, bool p_use_arrays); + void create_reflection_importance_sample(RendererStorageRD *p_storage, bool p_use_arrays, int p_cube_side, int p_base_layer, uint32_t p_sky_ggx_samples_quality); + void update_reflection_mipmaps(RendererStorageRD *p_storage, int p_start, int p_end); + }; + + struct SkyShaderData : public RendererStorageRD::ShaderData { + bool valid; + RID version; + + PipelineCacheRD pipelines[SKY_VERSION_MAX]; + Map<StringName, ShaderLanguage::ShaderNode::Uniform> uniforms; + Vector<ShaderCompilerRD::GeneratedCode::Texture> texture_uniforms; + + Vector<uint32_t> ubo_offsets; + uint32_t ubo_size; + + String path; + String code; + Map<StringName, RID> default_texture_params; + + bool uses_time; + bool uses_position; + bool uses_half_res; + bool uses_quarter_res; + bool uses_light; + + virtual void set_code(const String &p_Code); + virtual void set_default_texture_param(const StringName &p_name, RID p_texture); + virtual void get_param_list(List<PropertyInfo> *p_param_list) const; + virtual void get_instance_param_list(List<RendererStorage::InstanceShaderParam> *p_param_list) const; + virtual bool is_param_texture(const StringName &p_param) const; + virtual bool is_animated() const; + virtual bool casts_shadows() const; + virtual Variant get_default_parameter(const StringName &p_parameter) const; + virtual RS::ShaderNativeSourceCode get_native_source_code() const; + SkyShaderData(); + virtual ~SkyShaderData(); + }; + + /* Sky shader */ + + struct SkyShader { + SkyShaderRD shader; + ShaderCompilerRD compiler; + + RID default_shader; + RID default_material; + RID default_shader_rd; + } sky_shader; + + struct SkyMaterialData : public RendererStorageRD::MaterialData { + uint64_t last_frame; + SkyShaderData *shader_data; + RID uniform_buffer; + RID uniform_set; + Vector<RID> texture_cache; + Vector<uint8_t> ubo_data; + bool uniform_set_updated; + + virtual void set_render_priority(int p_priority) {} + virtual void set_next_pass(RID p_pass) {} + virtual void update_parameters(const Map<StringName, Variant> &p_parameters, bool p_uniform_dirty, bool p_textures_dirty); + virtual ~SkyMaterialData(); + }; + + struct Sky { + RID radiance; + RID half_res_pass; + RID half_res_framebuffer; + RID quarter_res_pass; + RID quarter_res_framebuffer; + Size2i screen_size; + + RID texture_uniform_sets[SKY_TEXTURE_SET_MAX]; + RID uniform_set; + + RID material; + RID uniform_buffer; + + int radiance_size = 256; + + RS::SkyMode mode = RS::SKY_MODE_AUTOMATIC; + + ReflectionData reflection; + bool dirty = false; + int processing_layer = 0; + Sky *dirty_list = nullptr; + + //State to track when radiance cubemap needs updating + SkyMaterialData *prev_material; + Vector3 prev_position; + float prev_time; + + void free(RendererStorageRD *p_storage); + + RID get_textures(RendererStorageRD *p_storage, SkyTextureSetVersion p_version, RID p_default_shader_rd); + bool set_radiance_size(int p_radiance_size); + bool set_mode(RS::SkyMode p_mode); + bool set_material(RID p_material); + Ref<Image> bake_panorama(RendererStorageRD *p_storage, float p_energy, int p_roughness_layers, const Size2i &p_size); + }; + + uint32_t sky_ggx_samples_quality; + bool sky_use_cubemap_array; + Sky *dirty_sky_list = nullptr; + mutable RID_Owner<Sky, true> sky_owner; + int roughness_layers; + + RendererStorageRD::ShaderData *_create_sky_shader_func(); + static RendererStorageRD::ShaderData *_create_sky_shader_funcs(); + + RendererStorageRD::MaterialData *_create_sky_material_func(SkyShaderData *p_shader); + static RendererStorageRD::MaterialData *_create_sky_material_funcs(RendererStorageRD::ShaderData *p_shader); + + RendererSceneSkyRD(); + + void init(RendererStorageRD *p_storage); + + void setup(RendererSceneEnvironmentRD *p_env, RID p_render_buffers, const CameraMatrix &p_projection, const Transform &p_transform, const Size2i p_screen_size, RendererSceneRenderRD *p_scene_render); + void update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform &p_transform, double p_time); + void draw(RendererSceneEnvironmentRD *p_env, bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, const CameraMatrix &p_projection, const Transform &p_transform, double p_time); + + void invalidate_sky(Sky *p_sky); + void update_dirty_skys(); + + RID sky_get_material(RID p_sky) const; + + RID allocate_sky_rid(); + void initialize_sky_rid(RID p_rid); + Sky *get_sky(RID p_sky) const; + void free_sky(RID p_sky); + void sky_set_radiance_size(RID p_sky, int p_radiance_size); + void sky_set_mode(RID p_sky, RS::SkyMode p_mode); + void sky_set_material(RID p_sky, RID p_material); + Ref<Image> sky_bake_panorama(RID p_sky, float p_energy, bool p_bake_irradiance, const Size2i &p_size); + + RID sky_get_radiance_texture_rd(RID p_sky) const; +}; + +#endif /* RENDERING_SERVER_SCENE_SKY_RD_H */ diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.cpp b/servers/rendering/renderer_rd/renderer_storage_rd.cpp index bf7237cad0..b984f850a0 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_storage_rd.cpp @@ -36,6 +36,10 @@ #include "renderer_compositor_rd.h" #include "servers/rendering/shader_language.h" +bool RendererStorageRD::can_create_resources_async() const { + return true; +} + Ref<Image> RendererStorageRD::_validate_texture_format(const Ref<Image> &p_image, TextureToRDFormat &r_format) { Ref<Image> image = p_image->duplicate(); @@ -535,9 +539,13 @@ Ref<Image> RendererStorageRD::_validate_texture_format(const Ref<Image> &p_image return image; } -RID RendererStorageRD::texture_2d_create(const Ref<Image> &p_image) { - ERR_FAIL_COND_V(p_image.is_null(), RID()); - ERR_FAIL_COND_V(p_image->is_empty(), RID()); +RID RendererStorageRD::texture_allocate() { + return texture_owner.allocate_rid(); +} + +void RendererStorageRD::texture_2d_initialize(RID p_texture, const Ref<Image> &p_image) { + ERR_FAIL_COND(p_image.is_null()); + ERR_FAIL_COND(p_image->is_empty()); TextureToRDFormat ret_format; Ref<Image> image = _validate_texture_format(p_image, ret_format); @@ -585,13 +593,13 @@ RID RendererStorageRD::texture_2d_create(const Ref<Image> &p_image) { Vector<Vector<uint8_t>> data_slices; data_slices.push_back(data); texture.rd_texture = RD::get_singleton()->texture_create(rd_format, rd_view, data_slices); - ERR_FAIL_COND_V(texture.rd_texture.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture.is_null()); if (texture.rd_format_srgb != RD::DATA_FORMAT_MAX) { rd_view.format_override = texture.rd_format_srgb; texture.rd_texture_srgb = RD::get_singleton()->texture_create_shared(rd_view, texture.rd_texture); if (texture.rd_texture_srgb.is_null()) { RD::get_singleton()->free(texture.rd_texture); - ERR_FAIL_COND_V(texture.rd_texture_srgb.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture_srgb.is_null()); } } @@ -602,14 +610,14 @@ RID RendererStorageRD::texture_2d_create(const Ref<Image> &p_image) { texture.rd_view = rd_view; texture.is_proxy = false; - return texture_owner.make_rid(texture); + texture_owner.initialize_rid(p_texture, texture); } -RID RendererStorageRD::texture_2d_layered_create(const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type) { - ERR_FAIL_COND_V(p_layers.size() == 0, RID()); +void RendererStorageRD::texture_2d_layered_initialize(RID p_texture, const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type) { + ERR_FAIL_COND(p_layers.size() == 0); - ERR_FAIL_COND_V(p_layered_type == RS::TEXTURE_LAYERED_CUBEMAP && p_layers.size() != 6, RID()); - ERR_FAIL_COND_V(p_layered_type == RS::TEXTURE_LAYERED_CUBEMAP_ARRAY && (p_layers.size() < 6 || (p_layers.size() % 6) != 0), RID()); + ERR_FAIL_COND(p_layered_type == RS::TEXTURE_LAYERED_CUBEMAP && p_layers.size() != 6); + ERR_FAIL_COND(p_layered_type == RS::TEXTURE_LAYERED_CUBEMAP_ARRAY && (p_layers.size() < 6 || (p_layers.size() % 6) != 0)); TextureToRDFormat ret_format; Vector<Ref<Image>> images; @@ -620,7 +628,7 @@ RID RendererStorageRD::texture_2d_layered_create(const Vector<Ref<Image>> &p_lay Image::Format valid_format = Image::FORMAT_MAX; for (int i = 0; i < p_layers.size(); i++) { - ERR_FAIL_COND_V(p_layers[i]->is_empty(), RID()); + ERR_FAIL_COND(p_layers[i]->is_empty()); if (i == 0) { valid_width = p_layers[i]->get_width(); @@ -628,10 +636,10 @@ RID RendererStorageRD::texture_2d_layered_create(const Vector<Ref<Image>> &p_lay valid_format = p_layers[i]->get_format(); valid_mipmaps = p_layers[i]->has_mipmaps(); } else { - ERR_FAIL_COND_V(p_layers[i]->get_width() != valid_width, RID()); - ERR_FAIL_COND_V(p_layers[i]->get_height() != valid_height, RID()); - ERR_FAIL_COND_V(p_layers[i]->get_format() != valid_format, RID()); - ERR_FAIL_COND_V(p_layers[i]->has_mipmaps() != valid_mipmaps, RID()); + ERR_FAIL_COND(p_layers[i]->get_width() != valid_width); + ERR_FAIL_COND(p_layers[i]->get_height() != valid_height); + ERR_FAIL_COND(p_layers[i]->get_format() != valid_format); + ERR_FAIL_COND(p_layers[i]->has_mipmaps() != valid_mipmaps); } images.push_back(_validate_texture_format(p_layers[i], ret_format)); @@ -695,13 +703,13 @@ RID RendererStorageRD::texture_2d_layered_create(const Vector<Ref<Image>> &p_lay data_slices.push_back(data); } texture.rd_texture = RD::get_singleton()->texture_create(rd_format, rd_view, data_slices); - ERR_FAIL_COND_V(texture.rd_texture.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture.is_null()); if (texture.rd_format_srgb != RD::DATA_FORMAT_MAX) { rd_view.format_override = texture.rd_format_srgb; texture.rd_texture_srgb = RD::get_singleton()->texture_create_shared(rd_view, texture.rd_texture); if (texture.rd_texture_srgb.is_null()) { RD::get_singleton()->free(texture.rd_texture); - ERR_FAIL_COND_V(texture.rd_texture_srgb.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture_srgb.is_null()); } } @@ -712,14 +720,14 @@ RID RendererStorageRD::texture_2d_layered_create(const Vector<Ref<Image>> &p_lay texture.rd_view = rd_view; texture.is_proxy = false; - return texture_owner.make_rid(texture); + texture_owner.initialize_rid(p_texture, texture); } -RID RendererStorageRD::texture_3d_create(Image::Format p_format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) { - ERR_FAIL_COND_V(p_data.size() == 0, RID()); +void RendererStorageRD::texture_3d_initialize(RID p_texture, Image::Format p_format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) { + ERR_FAIL_COND(p_data.size() == 0); Image::Image3DValidateError verr = Image::validate_3d_image(p_format, p_width, p_height, p_depth, p_mipmaps, p_data); if (verr != Image::VALIDATE_3D_OK) { - ERR_FAIL_V_MSG(RID(), Image::get_3d_image_validation_error_text(verr)); + ERR_FAIL_MSG(Image::get_3d_image_validation_error_text(verr)); } TextureToRDFormat ret_format; @@ -811,13 +819,13 @@ RID RendererStorageRD::texture_3d_create(Image::Format p_format, int p_width, in data_slices.push_back(all_data); //one slice texture.rd_texture = RD::get_singleton()->texture_create(rd_format, rd_view, data_slices); - ERR_FAIL_COND_V(texture.rd_texture.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture.is_null()); if (texture.rd_format_srgb != RD::DATA_FORMAT_MAX) { rd_view.format_override = texture.rd_format_srgb; texture.rd_texture_srgb = RD::get_singleton()->texture_create_shared(rd_view, texture.rd_texture); if (texture.rd_texture_srgb.is_null()) { RD::get_singleton()->free(texture.rd_texture); - ERR_FAIL_COND_V(texture.rd_texture_srgb.is_null(), RID()); + ERR_FAIL_COND(texture.rd_texture_srgb.is_null()); } } @@ -828,12 +836,12 @@ RID RendererStorageRD::texture_3d_create(Image::Format p_format, int p_width, in texture.rd_view = rd_view; texture.is_proxy = false; - return texture_owner.make_rid(texture); + texture_owner.initialize_rid(p_texture, texture); } -RID RendererStorageRD::texture_proxy_create(RID p_base) { +void RendererStorageRD::texture_proxy_initialize(RID p_texture, RID p_base) { Texture *tex = texture_owner.getornull(p_base); - ERR_FAIL_COND_V(!tex, RID()); + ERR_FAIL_COND(!tex); Texture proxy_tex = *tex; proxy_tex.rd_view.format_override = tex->rd_format; @@ -847,11 +855,9 @@ RID RendererStorageRD::texture_proxy_create(RID p_base) { proxy_tex.is_proxy = true; proxy_tex.proxies.clear(); - RID rid = texture_owner.make_rid(proxy_tex); + texture_owner.initialize_rid(p_texture, proxy_tex); - tex->proxies.push_back(rid); - - return rid; + tex->proxies.push_back(p_texture); } void RendererStorageRD::_texture_2d_update(RID p_texture, const Ref<Image> &p_image, int p_layer, bool p_immediate) { @@ -873,7 +879,7 @@ void RendererStorageRD::_texture_2d_update(RID p_texture, const Ref<Image> &p_im TextureToRDFormat f; Ref<Image> validated = _validate_texture_format(p_image, f); - RD::get_singleton()->texture_update(tex->rd_texture, p_layer, validated->get_data(), !p_immediate); + RD::get_singleton()->texture_update(tex->rd_texture, p_layer, validated->get_data()); } void RendererStorageRD::texture_2d_update_immediate(RID p_texture, const Ref<Image> &p_image, int p_layer) { @@ -918,7 +924,7 @@ void RendererStorageRD::texture_3d_update(RID p_texture, const Vector<Ref<Image> } } - RD::get_singleton()->texture_update(tex->rd_texture, 0, all_data, true); + RD::get_singleton()->texture_update(tex->rd_texture, 0, all_data); } void RendererStorageRD::texture_proxy_update(RID p_texture, RID p_proxy_to) { @@ -961,7 +967,7 @@ void RendererStorageRD::texture_proxy_update(RID p_texture, RID p_proxy_to) { } //these two APIs can be used together or in combination with the others. -RID RendererStorageRD::texture_2d_placeholder_create() { +void RendererStorageRD::texture_2d_placeholder_initialize(RID p_texture) { //this could be better optimized to reuse an existing image , done this way //for now to get it working Ref<Image> image; @@ -974,10 +980,10 @@ RID RendererStorageRD::texture_2d_placeholder_create() { } } - return texture_2d_create(image); + texture_2d_initialize(p_texture, image); } -RID RendererStorageRD::texture_2d_layered_placeholder_create(RS::TextureLayeredType p_layered_type) { +void RendererStorageRD::texture_2d_layered_placeholder_initialize(RID p_texture, RS::TextureLayeredType p_layered_type) { //this could be better optimized to reuse an existing image , done this way //for now to get it working Ref<Image> image; @@ -1000,10 +1006,10 @@ RID RendererStorageRD::texture_2d_layered_placeholder_create(RS::TextureLayeredT } } - return texture_2d_layered_create(images, p_layered_type); + texture_2d_layered_initialize(p_texture, images, p_layered_type); } -RID RendererStorageRD::texture_3d_placeholder_create() { +void RendererStorageRD::texture_3d_placeholder_initialize(RID p_texture) { //this could be better optimized to reuse an existing image , done this way //for now to get it working Ref<Image> image; @@ -1022,7 +1028,7 @@ RID RendererStorageRD::texture_3d_placeholder_create() { images.push_back(image); } - return texture_3d_create(Image::FORMAT_RGBA8, 4, 4, 4, false, images); + texture_3d_initialize(p_texture, Image::FORMAT_RGBA8, 4, 4, 4, false, images); } Ref<Image> RendererStorageRD::texture_2d_get(RID p_texture) const { @@ -1223,8 +1229,11 @@ RendererStorageRD::CanvasTexture::~CanvasTexture() { clear_sets(); } -RID RendererStorageRD::canvas_texture_create() { - return canvas_texture_owner.make_rid(memnew(CanvasTexture)); +RID RendererStorageRD::canvas_texture_allocate() { + return canvas_texture_owner.allocate_rid(); +} +void RendererStorageRD::canvas_texture_initialize(RID p_rid) { + canvas_texture_owner.initialize_rid(p_rid, memnew(CanvasTexture)); } void RendererStorageRD::canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) { @@ -1365,12 +1374,15 @@ bool RendererStorageRD::canvas_texture_get_uniform_set(RID p_texture, RS::Canvas /* SHADER API */ -RID RendererStorageRD::shader_create() { +RID RendererStorageRD::shader_allocate() { + return shader_owner.allocate_rid(); +} +void RendererStorageRD::shader_initialize(RID p_rid) { Shader shader; shader.data = nullptr; shader.type = SHADER_TYPE_MAX; - return shader_owner.make_rid(shader); + shader_owner.initialize_rid(p_rid, shader); } void RendererStorageRD::shader_set_code(RID p_shader, const String &p_code) { @@ -1499,9 +1511,21 @@ void RendererStorageRD::shader_set_data_request_function(ShaderType p_shader_typ shader_data_request_func[p_shader_type] = p_function; } +RS::ShaderNativeSourceCode RendererStorageRD::shader_get_native_source_code(RID p_shader) const { + Shader *shader = shader_owner.getornull(p_shader); + ERR_FAIL_COND_V(!shader, RS::ShaderNativeSourceCode()); + if (shader->data) { + return shader->data->get_native_source_code(); + } + return RS::ShaderNativeSourceCode(); +} + /* COMMON MATERIAL API */ -RID RendererStorageRD::material_create() { +RID RendererStorageRD::material_allocate() { + return material_owner.allocate_rid(); +} +void RendererStorageRD::material_initialize(RID p_rid) { Material material; material.data = nullptr; material.shader = nullptr; @@ -1511,12 +1535,8 @@ RID RendererStorageRD::material_create() { material.uniform_dirty = false; material.texture_dirty = false; material.priority = 0; - RID id = material_owner.make_rid(material); - { - Material *material_ptr = material_owner.getornull(id); - material_ptr->self = id; - } - return id; + material.self = p_rid; + material_owner.initialize_rid(p_rid, material); } void RendererStorageRD::_material_queue_update(Material *material, bool p_uniform, bool p_texture) { @@ -2390,8 +2410,11 @@ void RendererStorageRD::_update_queued_materials() { /* MESH API */ -RID RendererStorageRD::mesh_create() { - return mesh_owner.make_rid(Mesh()); +RID RendererStorageRD::mesh_allocate() { + return mesh_owner.allocate_rid(); +} +void RendererStorageRD::mesh_initialize(RID p_rid) { + mesh_owner.initialize_rid(p_rid, Mesh()); } void RendererStorageRD::mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape_count) { @@ -2600,6 +2623,12 @@ void RendererStorageRD::mesh_add_surface(RID p_mesh, const RS::SurfaceData &p_su mesh->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); + for (Set<Mesh *>::Element *E = mesh->shadow_owners.front(); E; E = E->next()) { + Mesh *shadow_owner = E->get(); + shadow_owner->shadow_mesh = RID(); + shadow_owner->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); + } + mesh->material_cache.clear(); } @@ -2815,6 +2844,25 @@ AABB RendererStorageRD::mesh_get_aabb(RID p_mesh, RID p_skeleton) { return aabb; } +void RendererStorageRD::mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) { + Mesh *mesh = mesh_owner.getornull(p_mesh); + ERR_FAIL_COND(!mesh); + + Mesh *shadow_mesh = mesh_owner.getornull(mesh->shadow_mesh); + if (shadow_mesh) { + shadow_mesh->shadow_owners.erase(mesh); + } + mesh->shadow_mesh = p_shadow_mesh; + + shadow_mesh = mesh_owner.getornull(mesh->shadow_mesh); + + if (shadow_mesh) { + shadow_mesh->shadow_owners.insert(mesh); + } + + mesh->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); +} + void RendererStorageRD::mesh_clear(RID p_mesh) { Mesh *mesh = mesh_owner.getornull(p_mesh); ERR_FAIL_COND(!mesh); @@ -2862,6 +2910,12 @@ void RendererStorageRD::mesh_clear(RID p_mesh) { } mesh->has_bone_weights = false; mesh->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); + + for (Set<Mesh *>::Element *E = mesh->shadow_owners.front(); E; E = E->next()) { + Mesh *shadow_owner = E->get(); + shadow_owner->shadow_mesh = RID(); + shadow_owner->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); + } } bool RendererStorageRD::mesh_needs_instance(RID p_mesh, bool p_has_skeleton) { @@ -3004,7 +3058,7 @@ void RendererStorageRD::update_mesh_instances() { MeshInstance *mi = dirty_mesh_instance_weights.first()->self(); if (mi->blend_weights_buffer.is_valid()) { - RD::get_singleton()->buffer_update(mi->blend_weights_buffer, 0, mi->blend_weights.size() * sizeof(float), mi->blend_weights.ptr(), true); + RD::get_singleton()->buffer_update(mi->blend_weights_buffer, 0, mi->blend_weights.size() * sizeof(float), mi->blend_weights.ptr()); } dirty_mesh_instance_weights.remove(&mi->weight_update_list); mi->weights_dirty = false; @@ -3058,7 +3112,7 @@ void RendererStorageRD::update_mesh_instances() { RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(SkeletonShader::PushConstant)); //dispatch without barrier, so all is done at the same time - RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.vertex_count, 1, 1, 64, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.vertex_count, 1, 1); } mi->dirty = false; @@ -3258,11 +3312,14 @@ void RendererStorageRD::_mesh_surface_generate_version_for_input_mask(Mesh::Surf ////////////////// MULTIMESH -RID RendererStorageRD::multimesh_create() { - return multimesh_owner.make_rid(MultiMesh()); +RID RendererStorageRD::multimesh_allocate() { + return multimesh_owner.allocate_rid(); +} +void RendererStorageRD::multimesh_initialize(RID p_rid) { + multimesh_owner.initialize_rid(p_rid, MultiMesh()); } -void RendererStorageRD::multimesh_allocate(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors, bool p_use_custom_data) { +void RendererStorageRD::multimesh_allocate_data(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors, bool p_use_custom_data) { MultiMesh *multimesh = multimesh_owner.getornull(p_multimesh); ERR_FAIL_COND(!multimesh); @@ -3672,7 +3729,7 @@ void RendererStorageRD::multimesh_set_buffer(RID p_multimesh, const Vector<float { const float *r = p_buffer.ptr(); - RD::get_singleton()->buffer_update(multimesh->buffer, 0, p_buffer.size() * sizeof(float), r, false); + RD::get_singleton()->buffer_update(multimesh->buffer, 0, p_buffer.size() * sizeof(float), r); multimesh->buffer_set = true; } @@ -3771,14 +3828,14 @@ void RendererStorageRD::_update_dirty_multimeshes() { if (multimesh->data_cache_used_dirty_regions > 32 || multimesh->data_cache_used_dirty_regions > visible_region_count / 2) { //if there too many dirty regions, or represent the majority of regions, just copy all, else transfer cost piles up too much - RD::get_singleton()->buffer_update(multimesh->buffer, 0, MIN(visible_region_count * region_size, multimesh->instances * multimesh->stride_cache * sizeof(float)), data, false); + RD::get_singleton()->buffer_update(multimesh->buffer, 0, MIN(visible_region_count * region_size, multimesh->instances * multimesh->stride_cache * sizeof(float)), data); } else { //not that many regions? update them all for (uint32_t i = 0; i < visible_region_count; i++) { if (multimesh->data_cache_dirty_regions[i]) { uint64_t offset = i * region_size; uint64_t size = multimesh->stride_cache * multimesh->instances * sizeof(float); - RD::get_singleton()->buffer_update(multimesh->buffer, offset, MIN(region_size, size - offset), &data[i * region_size], false); + RD::get_singleton()->buffer_update(multimesh->buffer, offset, MIN(region_size, size - offset), &data[i * region_size]); } } } @@ -3809,8 +3866,11 @@ void RendererStorageRD::_update_dirty_multimeshes() { /* PARTICLES */ -RID RendererStorageRD::particles_create() { - return particles_owner.make_rid(Particles()); +RID RendererStorageRD::particles_allocate() { + return particles_owner.allocate_rid(); +} +void RendererStorageRD::particles_initialize(RID p_rid) { + particles_owner.initialize_rid(p_rid, Particles()); } void RendererStorageRD::particles_set_emitting(RID p_particles, bool p_emitting) { @@ -4469,7 +4529,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta if (sub_emitter && sub_emitter->emission_storage_buffer.is_valid()) { // print_line("updating subemitter buffer"); int32_t zero[4] = { 0, sub_emitter->amount, 0, 0 }; - RD::get_singleton()->buffer_update(sub_emitter->emission_storage_buffer, 0, sizeof(uint32_t) * 4, zero, true); + RD::get_singleton()->buffer_update(sub_emitter->emission_storage_buffer, 0, sizeof(uint32_t) * 4, zero); push_constant.can_emit = true; if (sub_emitter->emitting) { @@ -4487,13 +4547,13 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta } if (p_particles->emission_buffer && p_particles->emission_buffer->particle_count) { - RD::get_singleton()->buffer_update(p_particles->emission_storage_buffer, 0, sizeof(uint32_t) * 4 + sizeof(ParticleEmissionBuffer::Data) * p_particles->emission_buffer->particle_count, p_particles->emission_buffer, true); + RD::get_singleton()->buffer_update(p_particles->emission_storage_buffer, 0, sizeof(uint32_t) * 4 + sizeof(ParticleEmissionBuffer::Data) * p_particles->emission_buffer->particle_count, p_particles->emission_buffer); p_particles->emission_buffer->particle_count = 0; } p_particles->clear = false; - RD::get_singleton()->buffer_update(p_particles->frame_params_buffer, 0, sizeof(ParticlesFrameParams), &frame_params, true); + RD::get_singleton()->buffer_update(p_particles->frame_params_buffer, 0, sizeof(ParticlesFrameParams), &frame_params); ParticlesMaterialData *m = (ParticlesMaterialData *)material_get_data(p_particles->process_material, SHADER_TYPE_PARTICLES); if (!m) { @@ -4515,7 +4575,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(ParticlesShader::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_particles->amount, 1, 1, 64, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, p_particles->amount, 1, 1); RD::get_singleton()->compute_list_end(); } @@ -4569,7 +4629,7 @@ void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 & RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1, 64, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); RD::get_singleton()->compute_list_end(); @@ -4581,7 +4641,7 @@ void RendererStorageRD::particles_set_view_axis(RID p_particles, const Vector3 & RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_sort_uniform_set, 1); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1, 64, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); RD::get_singleton()->compute_list_end(); } @@ -4632,10 +4692,11 @@ void RendererStorageRD::update_particles() { if (particles->clear && particles->pre_process_time > 0.0) { float frame_time; - if (particles->fixed_fps > 0) + if (particles->fixed_fps > 0) { frame_time = 1.0 / particles->fixed_fps; - else + } else { frame_time = 1.0 / 30.0; + } float todo = particles->pre_process_time; @@ -4671,10 +4732,11 @@ void RendererStorageRD::update_particles() { particles->frame_remainder = todo; } else { - if (zero_time_scale) + if (zero_time_scale) { _particles_process(particles, 0.0); - else + } else { _particles_process(particles, RendererCompositorRD::singleton->get_frame_delta_time()); + } } //copy particles to instance buffer @@ -4688,7 +4750,7 @@ void RendererStorageRD::update_particles() { RD::get_singleton()->compute_list_bind_uniform_set(compute_list, particles->particles_copy_uniform_set, 0); RD::get_singleton()->compute_list_set_push_constant(compute_list, ©_push_constant, sizeof(ParticlesShader::CopyPushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1, 64, 1, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, particles->amount, 1, 1); RD::get_singleton()->compute_list_end(); } @@ -4823,6 +4885,10 @@ Variant RendererStorageRD::ParticlesShaderData::get_default_parameter(const Stri return Variant(); } +RS::ShaderNativeSourceCode RendererStorageRD::ParticlesShaderData::get_native_source_code() const { + return base_singleton->particles_shader.shader.version_get_native_source_code(version); +} + RendererStorageRD::ParticlesShaderData::ParticlesShaderData() { valid = false; } @@ -4940,8 +5006,11 @@ RendererStorageRD::MaterialData *RendererStorageRD::_create_particles_material_f /* PARTICLES COLLISION API */ -RID RendererStorageRD::particles_collision_create() { - return particles_collision_owner.make_rid(ParticlesCollision()); +RID RendererStorageRD::particles_collision_allocate() { + return particles_collision_owner.allocate_rid(); +} +void RendererStorageRD::particles_collision_initialize(RID p_rid) { + particles_collision_owner.initialize_rid(p_rid, ParticlesCollision()); } RID RendererStorageRD::particles_collision_get_heightfield_framebuffer(RID p_particles_collision) const { @@ -5054,6 +5123,7 @@ void RendererStorageRD::particles_collision_height_field_update(RID p_particles_ void RendererStorageRD::particles_collision_set_height_field_resolution(RID p_particles_collision, RS::ParticlesCollisionHeightfieldResolution p_resolution) { ParticlesCollision *particles_collision = particles_collision_owner.getornull(p_particles_collision); ERR_FAIL_COND(!particles_collision); + ERR_FAIL_INDEX(p_resolution, RS::PARTICLES_COLLISION_HEIGHTFIELD_RESOLUTION_MAX); if (particles_collision->heightfield_resolution == p_resolution) { return; @@ -5120,8 +5190,11 @@ void RendererStorageRD::particles_collision_instance_set_active(RID p_collision_ /* SKELETON API */ -RID RendererStorageRD::skeleton_create() { - return skeleton_owner.make_rid(Skeleton()); +RID RendererStorageRD::skeleton_allocate() { + return skeleton_owner.allocate_rid(); +} +void RendererStorageRD::skeleton_initialize(RID p_rid) { + skeleton_owner.initialize_rid(p_rid, Skeleton()); } void RendererStorageRD::_skeleton_make_dirty(Skeleton *skeleton) { @@ -5132,7 +5205,7 @@ void RendererStorageRD::_skeleton_make_dirty(Skeleton *skeleton) { } } -void RendererStorageRD::skeleton_allocate(RID p_skeleton, int p_bones, bool p_2d_skeleton) { +void RendererStorageRD::skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton) { Skeleton *skeleton = skeleton_owner.getornull(p_skeleton); ERR_FAIL_COND(!skeleton); ERR_FAIL_COND(p_bones < 0); @@ -5288,7 +5361,7 @@ void RendererStorageRD::_update_dirty_skeletons() { Skeleton *skeleton = skeleton_dirty_list; if (skeleton->size) { - RD::get_singleton()->buffer_update(skeleton->buffer, 0, skeleton->data.size() * sizeof(float), skeleton->data.ptr(), false); + RD::get_singleton()->buffer_update(skeleton->buffer, 0, skeleton->data.size() * sizeof(float), skeleton->data.ptr()); } skeleton_dirty_list = skeleton->dirty_list; @@ -5306,7 +5379,7 @@ void RendererStorageRD::_update_dirty_skeletons() { /* LIGHT */ -RID RendererStorageRD::light_create(RS::LightType p_type) { +void RendererStorageRD::_light_initialize(RID p_light, RS::LightType p_type) { Light light; light.type = p_type; @@ -5327,10 +5400,31 @@ RID RendererStorageRD::light_create(RS::LightType p_type) { light.param[RS::LIGHT_PARAM_SHADOW_BIAS] = 0.02; light.param[RS::LIGHT_PARAM_SHADOW_BLUR] = 0; light.param[RS::LIGHT_PARAM_SHADOW_PANCAKE_SIZE] = 20.0; - light.param[RS::LIGHT_PARAM_SHADOW_VOLUMETRIC_FOG_FADE] = 1.0; + light.param[RS::LIGHT_PARAM_SHADOW_VOLUMETRIC_FOG_FADE] = 0.1; light.param[RS::LIGHT_PARAM_TRANSMITTANCE_BIAS] = 0.05; - return light_owner.make_rid(light); + light_owner.initialize_rid(p_light, light); +} + +RID RendererStorageRD::directional_light_allocate() { + return light_owner.allocate_rid(); +} +void RendererStorageRD::directional_light_initialize(RID p_light) { + _light_initialize(p_light, RS::LIGHT_DIRECTIONAL); +} + +RID RendererStorageRD::omni_light_allocate() { + return light_owner.allocate_rid(); +} +void RendererStorageRD::omni_light_initialize(RID p_light) { + _light_initialize(p_light, RS::LIGHT_OMNI); +} + +RID RendererStorageRD::spot_light_allocate() { + return light_owner.allocate_rid(); +} +void RendererStorageRD::spot_light_initialize(RID p_light) { + _light_initialize(p_light, RS::LIGHT_SPOT); } void RendererStorageRD::light_set_color(RID p_light, const Color &p_color) { @@ -5568,8 +5662,11 @@ AABB RendererStorageRD::light_get_aabb(RID p_light) const { /* REFLECTION PROBE */ -RID RendererStorageRD::reflection_probe_create() { - return reflection_probe_owner.make_rid(ReflectionProbe()); +RID RendererStorageRD::reflection_probe_allocate() { + return reflection_probe_owner.allocate_rid(); +} +void RendererStorageRD::reflection_probe_initialize(RID p_reflection_probe) { + reflection_probe_owner.initialize_rid(p_reflection_probe, ReflectionProbe()); } void RendererStorageRD::reflection_probe_set_update_mode(RID p_probe, RS::ReflectionProbeUpdateMode p_mode) { @@ -5791,8 +5888,11 @@ float RendererStorageRD::reflection_probe_get_ambient_color_energy(RID p_probe) return reflection_probe->ambient_color_energy; } -RID RendererStorageRD::decal_create() { - return decal_owner.make_rid(Decal()); +RID RendererStorageRD::decal_allocate() { + return decal_owner.allocate_rid(); +} +void RendererStorageRD::decal_initialize(RID p_decal) { + decal_owner.initialize_rid(p_decal, Decal()); } void RendererStorageRD::decal_set_extents(RID p_decal, const Vector3 &p_extents) { @@ -5879,11 +5979,14 @@ AABB RendererStorageRD::decal_get_aabb(RID p_decal) const { return AABB(-decal->extents, decal->extents * 2.0); } -RID RendererStorageRD::gi_probe_create() { - return gi_probe_owner.make_rid(GIProbe()); +RID RendererStorageRD::gi_probe_allocate() { + return gi_probe_owner.allocate_rid(); +} +void RendererStorageRD::gi_probe_initialize(RID p_gi_probe) { + gi_probe_owner.initialize_rid(p_gi_probe, GIProbe()); } -void RendererStorageRD::gi_probe_allocate(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) { +void RendererStorageRD::gi_probe_allocate_data(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) { GIProbe *gi_probe = gi_probe_owner.getornull(p_gi_probe); ERR_FAIL_COND(!gi_probe); @@ -6232,8 +6335,12 @@ RID RendererStorageRD::gi_probe_get_sdf_texture(RID p_gi_probe) { /* LIGHTMAP API */ -RID RendererStorageRD::lightmap_create() { - return lightmap_owner.make_rid(Lightmap()); +RID RendererStorageRD::lightmap_allocate() { + return lightmap_owner.allocate_rid(); +} + +void RendererStorageRD::lightmap_initialize(RID p_lightmap) { + lightmap_owner.initialize_rid(p_lightmap, Lightmap()); } void RendererStorageRD::lightmap_set_textures(RID p_lightmap, RID p_light, bool p_uses_spherical_haromics) { @@ -6436,7 +6543,8 @@ void RendererStorageRD::_clear_render_target(RenderTarget *rt) { void RendererStorageRD::_update_render_target(RenderTarget *rt) { if (rt->texture.is_null()) { //create a placeholder until updated - rt->texture = texture_2d_placeholder_create(); + rt->texture = texture_allocate(); + texture_2d_placeholder_initialize(rt->texture); Texture *tex = texture_owner.getornull(rt->texture); tex->is_render_target = true; } @@ -6936,7 +7044,7 @@ void RendererStorageRD::render_target_sdf_process(RID p_render_target) { RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rt->sdf_buffer_process_uniform_sets[1], 0); //fill [0] RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(RenderTargetSDF::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1); /* Process */ @@ -6952,7 +7060,7 @@ void RendererStorageRD::render_target_sdf_process(RID p_render_target) { RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rt->sdf_buffer_process_uniform_sets[swap ? 1 : 0], 0); push_constant.stride = stride; RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(RenderTargetSDF::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1); stride /= 2; swap = !swap; RD::get_singleton()->compute_list_add_barrier(compute_list); @@ -6963,7 +7071,7 @@ void RendererStorageRD::render_target_sdf_process(RID p_render_target) { RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, rt_sdf.pipelines[shrink ? RenderTargetSDF::SHADER_STORE_SHRINK : RenderTargetSDF::SHADER_STORE]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, rt->sdf_buffer_process_uniform_sets[swap ? 1 : 0], 0); RD::get_singleton()->compute_list_set_push_constant(compute_list, &push_constant, sizeof(RenderTargetSDF::PushConstant)); - RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1, 8, 8, 1); + RD::get_singleton()->compute_list_dispatch_threads(compute_list, push_constant.size[0], push_constant.size[1], 1); RD::get_singleton()->compute_list_end(); } @@ -7327,6 +7435,7 @@ void RendererStorageRD::_update_decal_atlas() { tformat.shareable_formats.push_back(RD::DATA_FORMAT_R8G8B8A8_SRGB); decal_atlas.texture = RD::get_singleton()->texture_create(tformat, RD::TextureView()); + RD::get_singleton()->texture_clear(decal_atlas.texture, Color(0, 0, 0, 0), 0, decal_atlas.mipmaps, 0, 1); { //create the framebuffer @@ -7381,7 +7490,7 @@ void RendererStorageRD::_update_decal_atlas() { prev_texture = mm.texture; } } else { - RD::get_singleton()->texture_clear(mm.texture, clear_color, 0, 1, 0, 1, false); + RD::get_singleton()->texture_clear(mm.texture, clear_color, 0, 1, 0, 1); } } } @@ -8147,29 +8256,38 @@ bool RendererStorageRD::free(RID p_rid) { material_owner.free(p_rid); } else if (mesh_owner.owns(p_rid)) { mesh_clear(p_rid); + mesh_set_shadow_mesh(p_rid, RID()); Mesh *mesh = mesh_owner.getornull(p_rid); mesh->dependency.deleted_notify(p_rid); if (mesh->instances.size()) { ERR_PRINT("deleting mesh with active instances"); } + if (mesh->shadow_owners.size()) { + for (Set<Mesh *>::Element *E = mesh->shadow_owners.front(); E; E = E->next()) { + Mesh *shadow_owner = E->get(); + shadow_owner->shadow_mesh = RID(); + shadow_owner->dependency.changed_notify(DEPENDENCY_CHANGED_MESH); + } + } mesh_owner.free(p_rid); } else if (mesh_instance_owner.owns(p_rid)) { MeshInstance *mi = mesh_instance_owner.getornull(p_rid); _mesh_instance_clear(mi); mi->mesh->instances.erase(mi->I); mi->I = nullptr; + mesh_instance_owner.free(p_rid); memdelete(mi); } else if (multimesh_owner.owns(p_rid)) { _update_dirty_multimeshes(); - multimesh_allocate(p_rid, 0, RS::MULTIMESH_TRANSFORM_2D); + multimesh_allocate_data(p_rid, 0, RS::MULTIMESH_TRANSFORM_2D); MultiMesh *multimesh = multimesh_owner.getornull(p_rid); multimesh->dependency.deleted_notify(p_rid); multimesh_owner.free(p_rid); } else if (skeleton_owner.owns(p_rid)) { _update_dirty_skeletons(); - skeleton_allocate(p_rid, 0); + skeleton_allocate_data(p_rid, 0); Skeleton *skeleton = skeleton_owner.getornull(p_rid); skeleton->dependency.deleted_notify(p_rid); skeleton_owner.free(p_rid); @@ -8187,7 +8305,7 @@ bool RendererStorageRD::free(RID p_rid) { decal->dependency.deleted_notify(p_rid); decal_owner.free(p_rid); } else if (gi_probe_owner.owns(p_rid)) { - gi_probe_allocate(p_rid, Transform(), AABB(), Vector3i(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<int>()); //deallocate + gi_probe_allocate_data(p_rid, Transform(), AABB(), Vector3i(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<uint8_t>(), Vector<int>()); //deallocate GIProbe *gi_probe = gi_probe_owner.getornull(p_rid); gi_probe->dependency.deleted_notify(p_rid); gi_probe_owner.free(p_rid); @@ -8243,11 +8361,11 @@ EffectsRD *RendererStorageRD::get_effects() { } void RendererStorageRD::capture_timestamps_begin() { - RD::get_singleton()->capture_timestamp("Frame Begin", false); + RD::get_singleton()->capture_timestamp("Frame Begin"); } void RendererStorageRD::capture_timestamp(const String &p_name) { - RD::get_singleton()->capture_timestamp(p_name, true); + RD::get_singleton()->capture_timestamp(p_name); } uint32_t RendererStorageRD::get_captured_timestamps_count() const { @@ -8281,7 +8399,7 @@ RendererStorageRD::RendererStorageRD() { static_assert(sizeof(GlobalVariables::Value) == 16); - global_variables.buffer_size = GLOBAL_GET("rendering/high_end/global_shader_variables_buffer_size"); + global_variables.buffer_size = GLOBAL_GET("rendering/limits/global_shader_variables/buffer_size"); global_variables.buffer_size = MAX(4096, global_variables.buffer_size); global_variables.buffer_values = memnew_arr(GlobalVariables::Value, global_variables.buffer_size); zeromem(global_variables.buffer_values, sizeof(GlobalVariables::Value) * global_variables.buffer_size); @@ -8549,14 +8667,14 @@ RendererStorageRD::RendererStorageRD() { sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.use_anisotropy = true; - sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/quality/texture_filters/anisotropic_filtering_level")); + sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/textures/default_filters/anisotropic_filtering_level")); } break; case RS::CANVAS_ITEM_TEXTURE_FILTER_LINEAR_WITH_MIPMAPS_ANISOTROPIC: { sampler_state.mag_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.min_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.mip_filter = RD::SAMPLER_FILTER_LINEAR; sampler_state.use_anisotropy = true; - sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/quality/texture_filters/anisotropic_filtering_level")); + sampler_state.anisotropy_max = 1 << int(GLOBAL_GET("rendering/textures/default_filters/anisotropic_filtering_level")); } break; default: { @@ -8726,7 +8844,7 @@ RendererStorageRD::RendererStorageRD() { } } - lightmap_probe_capture_update_speed = GLOBAL_GET("rendering/lightmapper/probe_capture_update_speed"); + lightmap_probe_capture_update_speed = GLOBAL_GET("rendering/lightmapping/probe_capture/update_speed"); /* Particles */ @@ -8767,7 +8885,7 @@ RendererStorageRD::RendererStorageRD() { actions.renames["RESTART_VELOCITY"] = "restart_velocity"; actions.renames["RESTART_COLOR"] = "restart_color"; actions.renames["RESTART_CUSTOM"] = "restart_custom"; - actions.renames["emit_particle"] = "emit_particle"; + actions.renames["emit_subparticle"] = "emit_subparticle"; actions.renames["COLLIDED"] = "collided"; actions.renames["COLLISION_NORMAL"] = "collision_normal"; actions.renames["COLLISION_DEPTH"] = "collision_depth"; @@ -8793,9 +8911,11 @@ RendererStorageRD::RendererStorageRD() { { // default material and shader for particles shader - particles_shader.default_shader = shader_create(); + particles_shader.default_shader = shader_allocate(); + shader_initialize(particles_shader.default_shader); shader_set_code(particles_shader.default_shader, "shader_type particles; void compute() { COLOR = vec4(1.0); } \n"); - particles_shader.default_material = material_create(); + particles_shader.default_material = material_allocate(); + material_initialize(particles_shader.default_material); material_set_shader(particles_shader.default_material, particles_shader.default_shader); ParticlesMaterialData *md = (ParticlesMaterialData *)material_get_data(particles_shader.default_material, RendererStorageRD::SHADER_TYPE_PARTICLES); diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.h b/servers/rendering/renderer_rd/renderer_storage_rd.h index bd27936e38..6405bb75b0 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.h +++ b/servers/rendering/renderer_rd/renderer_storage_rd.h @@ -95,6 +95,21 @@ public: p_array[11] = 0; } + static _FORCE_INLINE_ void store_transform_transposed_3x4(const Transform &p_mtx, float *p_array) { + p_array[0] = p_mtx.basis.elements[0][0]; + p_array[1] = p_mtx.basis.elements[0][1]; + p_array[2] = p_mtx.basis.elements[0][2]; + p_array[3] = p_mtx.origin.x; + p_array[4] = p_mtx.basis.elements[1][0]; + p_array[5] = p_mtx.basis.elements[1][1]; + p_array[6] = p_mtx.basis.elements[1][2]; + p_array[7] = p_mtx.origin.y; + p_array[8] = p_mtx.basis.elements[2][0]; + p_array[9] = p_mtx.basis.elements[2][1]; + p_array[10] = p_mtx.basis.elements[2][2]; + p_array[11] = p_mtx.origin.z; + } + static _FORCE_INLINE_ void store_camera(const CameraMatrix &p_mtx, float *p_array) { for (int i = 0; i < 4; i++) { for (int j = 0; j < 4; j++) { @@ -127,6 +142,8 @@ public: virtual bool is_animated() const = 0; virtual bool casts_shadows() const = 0; virtual Variant get_default_parameter(const StringName &p_parameter) const = 0; + virtual RS::ShaderNativeSourceCode get_native_source_code() const { return RS::ShaderNativeSourceCode(); } + virtual ~ShaderData() {} }; @@ -204,7 +221,7 @@ private: ~CanvasTexture(); }; - RID_PtrOwner<CanvasTexture> canvas_texture_owner; + RID_PtrOwner<CanvasTexture, true> canvas_texture_owner; /* TEXTURE API */ struct Texture { @@ -350,7 +367,7 @@ private: }; ShaderDataRequestFunction shader_data_request_func[SHADER_TYPE_MAX]; - mutable RID_Owner<Shader> shader_owner; + mutable RID_Owner<Shader, true> shader_owner; /* Material */ @@ -372,7 +389,7 @@ private: }; MaterialDataRequestFunction material_data_request_func[SHADER_TYPE_MAX]; - mutable RID_Owner<Material> material_owner; + mutable RID_Owner<Material, true> material_owner; Material *material_update_list; void _material_queue_update(Material *material, bool p_uniform, bool p_texture); @@ -461,10 +478,13 @@ private: List<MeshInstance *> instances; + RID shadow_mesh; + Set<Mesh *> shadow_owners; + Dependency dependency; }; - mutable RID_Owner<Mesh> mesh_owner; + mutable RID_Owner<Mesh, true> mesh_owner; struct MeshInstance { Mesh *mesh; @@ -567,7 +587,7 @@ private: Dependency dependency; }; - mutable RID_Owner<MultiMesh> multimesh_owner; + mutable RID_Owner<MultiMesh, true> multimesh_owner; MultiMesh *multimesh_dirty_list = nullptr; @@ -671,21 +691,21 @@ private: }; struct Particles { - bool inactive; - float inactive_time; - bool emitting; - bool one_shot; - int amount; - float lifetime; - float pre_process_time; - float explosiveness; - float randomness; - bool restart_request; - AABB custom_aabb; - bool use_local_coords; + bool inactive = true; + float inactive_time = 0.0; + bool emitting = false; + bool one_shot = false; + int amount = 0; + float lifetime = 1.0; + float pre_process_time = 0.0; + float explosiveness = 0.0; + float randomness = 0.0; + bool restart_request = false; + AABB custom_aabb = AABB(Vector3(-4, -4, -4), Vector3(8, 8, 8)); + bool use_local_coords = true; RID process_material; - RS::ParticlesDrawOrder draw_order; + RS::ParticlesDrawOrder draw_order = RS::PARTICLES_DRAW_ORDER_INDEX; Vector<RID> draw_passes; @@ -710,21 +730,21 @@ private: RID sub_emitter; - float phase; - float prev_phase; - uint64_t prev_ticks; - uint32_t random_seed; + float phase = 0.0; + float prev_phase = 0.0; + uint64_t prev_ticks = 0; + uint32_t random_seed = 0; - uint32_t cycle_number; + uint32_t cycle_number = 0; - float speed_scale; + float speed_scale = 1.0; - int fixed_fps; - bool fractional_delta; - float frame_remainder; - float collision_base_size; + int fixed_fps = 0; + bool fractional_delta = false; + float frame_remainder = 0; + float collision_base_size = 0.01; - bool clear; + bool clear = true; bool force_sub_emit = false; @@ -737,31 +757,6 @@ private: Set<RID> collisions; - Particles() : - inactive(true), - inactive_time(0.0), - emitting(false), - one_shot(false), - amount(0), - lifetime(1.0), - pre_process_time(0.0), - explosiveness(0.0), - randomness(0.0), - restart_request(false), - custom_aabb(AABB(Vector3(-4, -4, -4), Vector3(8, 8, 8))), - use_local_coords(true), - draw_order(RS::PARTICLES_DRAW_ORDER_INDEX), - prev_ticks(0), - random_seed(0), - cycle_number(0), - speed_scale(1.0), - fixed_fps(0), - fractional_delta(false), - frame_remainder(0), - collision_base_size(0.01), - clear(true) { - } - Dependency dependency; ParticlesFrameParams frame_params; @@ -840,6 +835,8 @@ private: virtual bool is_animated() const; virtual bool casts_shadows() const; virtual Variant get_default_parameter(const StringName &p_parameter) const; + virtual RS::ShaderNativeSourceCode get_native_source_code() const; + ParticlesShaderData(); virtual ~ParticlesShaderData(); }; @@ -871,7 +868,7 @@ private: void update_particles(); - mutable RID_Owner<Particles> particles_owner; + mutable RID_Owner<Particles, true> particles_owner; /* Particles Collision */ @@ -893,7 +890,7 @@ private: Dependency dependency; }; - mutable RID_Owner<ParticlesCollision> particles_collision_owner; + mutable RID_Owner<ParticlesCollision, true> particles_collision_owner; struct ParticlesCollisionInstance { RID collision; @@ -923,7 +920,7 @@ private: Dependency dependency; }; - mutable RID_Owner<Skeleton> skeleton_owner; + mutable RID_Owner<Skeleton, true> skeleton_owner; _FORCE_INLINE_ void _skeleton_make_dirty(Skeleton *skeleton); @@ -955,7 +952,7 @@ private: Dependency dependency; }; - mutable RID_Owner<Light> light_owner; + mutable RID_Owner<Light, true> light_owner; /* REFLECTION PROBE */ @@ -978,7 +975,7 @@ private: Dependency dependency; }; - mutable RID_Owner<ReflectionProbe> reflection_probe_owner; + mutable RID_Owner<ReflectionProbe, true> reflection_probe_owner; /* DECAL */ @@ -999,7 +996,7 @@ private: Dependency dependency; }; - mutable RID_Owner<Decal> decal_owner; + mutable RID_Owner<Decal, true> decal_owner; /* GI PROBE */ @@ -1042,7 +1039,7 @@ private: RID giprobe_sdf_shader_version_shader; RID giprobe_sdf_shader_pipeline; - mutable RID_Owner<GIProbe> gi_probe_owner; + mutable RID_Owner<GIProbe, true> gi_probe_owner; /* REFLECTION PROBE */ @@ -1073,7 +1070,7 @@ private: uint64_t lightmap_array_version = 0; - mutable RID_Owner<Lightmap> lightmap_owner; + mutable RID_Owner<Lightmap, true> lightmap_owner; float lightmap_probe_capture_update_speed = 4; @@ -1227,12 +1224,16 @@ private: EffectsRD effects; public: + virtual bool can_create_resources_async() const; + /* TEXTURE API */ - virtual RID texture_2d_create(const Ref<Image> &p_image); - virtual RID texture_2d_layered_create(const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type); - virtual RID texture_3d_create(Image::Format p_format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data); //all slices, then all the mipmaps, must be coherent - virtual RID texture_proxy_create(RID p_base); + virtual RID texture_allocate(); + + virtual void texture_2d_initialize(RID p_texture, const Ref<Image> &p_image); + virtual void texture_2d_layered_initialize(RID p_texture, const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type); + virtual void texture_3d_initialize(RID p_texture, Image::Format p_format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data); //all slices, then all the mipmaps, must be coherent + virtual void texture_proxy_initialize(RID p_texture, RID p_base); virtual void _texture_2d_update(RID p_texture, const Ref<Image> &p_image, int p_layer, bool p_immediate); @@ -1242,9 +1243,9 @@ public: virtual void texture_proxy_update(RID p_texture, RID p_proxy_to); //these two APIs can be used together or in combination with the others. - virtual RID texture_2d_placeholder_create(); - virtual RID texture_2d_layered_placeholder_create(RenderingServer::TextureLayeredType p_layered_type); - virtual RID texture_3d_placeholder_create(); + virtual void texture_2d_placeholder_initialize(RID p_texture); + virtual void texture_2d_layered_placeholder_initialize(RID p_texture, RenderingServer::TextureLayeredType p_layered_type); + virtual void texture_3d_placeholder_initialize(RID p_texture); virtual Ref<Image> texture_2d_get(RID p_texture) const; virtual Ref<Image> texture_2d_layer_get(RID p_texture, int p_layer) const; @@ -1316,7 +1317,8 @@ public: /* CANVAS TEXTURE API */ - virtual RID canvas_texture_create(); + RID canvas_texture_allocate(); + void canvas_texture_initialize(RID p_canvas_texture); virtual void canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture); virtual void canvas_texture_set_shading_parameters(RID p_canvas_texture, const Color &p_specular_color, float p_shininess); @@ -1328,7 +1330,8 @@ public: /* SHADER API */ - RID shader_create(); + RID shader_allocate(); + void shader_initialize(RID p_shader); void shader_set_code(RID p_shader, const String &p_code); String shader_get_code(RID p_shader) const; @@ -1339,9 +1342,12 @@ public: Variant shader_get_param_default(RID p_shader, const StringName &p_param) const; void shader_set_data_request_function(ShaderType p_shader_type, ShaderDataRequestFunction p_function); + virtual RS::ShaderNativeSourceCode shader_get_native_source_code(RID p_shader) const; + /* COMMON MATERIAL API */ - RID material_create(); + RID material_allocate(); + void material_initialize(RID p_material); void material_set_shader(RID p_material, RID p_shader); @@ -1377,7 +1383,8 @@ public: /* MESH API */ - virtual RID mesh_create(); + RID mesh_allocate(); + void mesh_initialize(RID p_mesh); virtual void mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape_count); @@ -1402,6 +1409,7 @@ public: virtual AABB mesh_get_custom_aabb(RID p_mesh) const; virtual AABB mesh_get_aabb(RID p_mesh, RID p_skeleton = RID()); + virtual void mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh); virtual void mesh_clear(RID p_mesh); @@ -1440,6 +1448,13 @@ public: return mesh->surfaces[p_surface_index]; } + _FORCE_INLINE_ RID mesh_get_shadow_mesh(RID p_mesh) { + Mesh *mesh = mesh_owner.getornull(p_mesh); + ERR_FAIL_COND_V(!mesh, RID()); + + return mesh->shadow_mesh; + } + _FORCE_INLINE_ RS::PrimitiveType mesh_surface_get_primitive(void *p_surface) { Mesh::Surface *surface = reinterpret_cast<Mesh::Surface *>(p_surface); return surface->primitive; @@ -1450,13 +1465,7 @@ public: return s->lod_count > 0; } - _FORCE_INLINE_ RID mesh_surface_get_index_array(void *p_surface) const { - Mesh::Surface *s = reinterpret_cast<Mesh::Surface *>(p_surface); - - return s->index_array; - } - - _FORCE_INLINE_ RID mesh_surface_get_index_array_with_lod(void *p_surface, float p_model_scale, float p_distance_threshold, float p_lod_threshold) const { + _FORCE_INLINE_ uint32_t mesh_surface_get_lod(void *p_surface, float p_model_scale, float p_distance_threshold, float p_lod_threshold) const { Mesh::Surface *s = reinterpret_cast<Mesh::Surface *>(p_surface); int32_t current_lod = -1; @@ -1468,9 +1477,19 @@ public: current_lod = i; } if (current_lod == -1) { + return 0; + } else { + return current_lod + 1; + } + } + + _FORCE_INLINE_ RID mesh_surface_get_index_array(void *p_surface, uint32_t p_lod) const { + Mesh::Surface *s = reinterpret_cast<Mesh::Surface *>(p_surface); + + if (p_lod == 0) { return s->index_array; } else { - return s->lods[current_lod].index_array; + return s->lods[p_lod - 1].index_array; } } @@ -1586,9 +1605,10 @@ public: /* MULTIMESH API */ - RID multimesh_create(); + RID multimesh_allocate(); + void multimesh_initialize(RID p_multimesh); - void multimesh_allocate(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false); + void multimesh_allocate_data(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false); int multimesh_get_instance_count(RID p_multimesh) const; void multimesh_set_mesh(RID p_multimesh, RID p_mesh); @@ -1652,24 +1672,28 @@ public: /* IMMEDIATE API */ - RID immediate_create() { return RID(); } - void immediate_begin(RID p_immediate, RS::PrimitiveType p_rimitive, RID p_texture = RID()) {} - void immediate_vertex(RID p_immediate, const Vector3 &p_vertex) {} - void immediate_normal(RID p_immediate, const Vector3 &p_normal) {} - void immediate_tangent(RID p_immediate, const Plane &p_tangent) {} - void immediate_color(RID p_immediate, const Color &p_color) {} - void immediate_uv(RID p_immediate, const Vector2 &tex_uv) {} - void immediate_uv2(RID p_immediate, const Vector2 &tex_uv) {} - void immediate_end(RID p_immediate) {} - void immediate_clear(RID p_immediate) {} - void immediate_set_material(RID p_immediate, RID p_material) {} - RID immediate_get_material(RID p_immediate) const { return RID(); } - AABB immediate_get_aabb(RID p_immediate) const { return AABB(); } + RID immediate_allocate() { return RID(); } + void immediate_initialize(RID p_immediate) {} + + virtual void immediate_begin(RID p_immediate, RS::PrimitiveType p_rimitive, RID p_texture = RID()) {} + virtual void immediate_vertex(RID p_immediate, const Vector3 &p_vertex) {} + virtual void immediate_normal(RID p_immediate, const Vector3 &p_normal) {} + virtual void immediate_tangent(RID p_immediate, const Plane &p_tangent) {} + virtual void immediate_color(RID p_immediate, const Color &p_color) {} + virtual void immediate_uv(RID p_immediate, const Vector2 &tex_uv) {} + virtual void immediate_uv2(RID p_immediate, const Vector2 &tex_uv) {} + virtual void immediate_end(RID p_immediate) {} + virtual void immediate_clear(RID p_immediate) {} + virtual void immediate_set_material(RID p_immediate, RID p_material) {} + virtual RID immediate_get_material(RID p_immediate) const { return RID(); } + virtual AABB immediate_get_aabb(RID p_immediate) const { return AABB(); } /* SKELETON API */ - RID skeleton_create(); - void skeleton_allocate(RID p_skeleton, int p_bones, bool p_2d_skeleton = false); + RID skeleton_allocate(); + void skeleton_initialize(RID p_skeleton); + + void skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton = false); void skeleton_set_base_transform_2d(RID p_skeleton, const Transform2D &p_base_transform); void skeleton_set_world_transform(RID p_skeleton, bool p_enable, const Transform &p_world_transform); int skeleton_get_bone_count(RID p_skeleton) const; @@ -1703,11 +1727,16 @@ public: } /* Light API */ - RID light_create(RS::LightType p_type); + void _light_initialize(RID p_rid, RS::LightType p_type); - RID directional_light_create() { return light_create(RS::LIGHT_DIRECTIONAL); } - RID omni_light_create() { return light_create(RS::LIGHT_OMNI); } - RID spot_light_create() { return light_create(RS::LIGHT_SPOT); } + RID directional_light_allocate(); + void directional_light_initialize(RID p_light); + + RID omni_light_allocate(); + void omni_light_initialize(RID p_light); + + RID spot_light_allocate(); + void spot_light_initialize(RID p_light); void light_set_color(RID p_light, const Color &p_color); void light_set_param(RID p_light, RS::LightParam p_param, float p_value); @@ -1810,7 +1839,8 @@ public: /* PROBE API */ - RID reflection_probe_create(); + RID reflection_probe_allocate(); + void reflection_probe_initialize(RID p_reflection_probe); void reflection_probe_set_update_mode(RID p_probe, RS::ReflectionProbeUpdateMode p_mode); void reflection_probe_set_intensity(RID p_probe, float p_intensity); @@ -1850,7 +1880,9 @@ public: /* DECAL API */ - virtual RID decal_create(); + RID decal_allocate(); + void decal_initialize(RID p_decal); + virtual void decal_set_extents(RID p_decal, const Vector3 &p_extents); virtual void decal_set_texture(RID p_decal, RS::DecalTexture p_type, RID p_texture); virtual void decal_set_emission_energy(RID p_decal, float p_energy); @@ -1925,9 +1957,10 @@ public: /* GI PROBE API */ - RID gi_probe_create(); + RID gi_probe_allocate(); + void gi_probe_initialize(RID p_gi_probe); - void gi_probe_allocate(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts); + void gi_probe_allocate_data(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts); AABB gi_probe_get_bounds(RID p_gi_probe) const; Vector3i gi_probe_get_octree_size(RID p_gi_probe) const; @@ -1978,7 +2011,8 @@ public: /* LIGHTMAP CAPTURE */ - virtual RID lightmap_create(); + RID lightmap_allocate(); + void lightmap_initialize(RID p_lightmap); virtual void lightmap_set_textures(RID p_lightmap, RID p_light, bool p_uses_spherical_haromics); virtual void lightmap_set_probe_bounds(RID p_lightmap, const AABB &p_bounds); @@ -2027,7 +2061,8 @@ public: /* PARTICLES */ - RID particles_create(); + RID particles_allocate(); + void particles_initialize(RID p_particles_collision); void particles_set_emitting(RID p_particles, bool p_emitting); void particles_set_amount(RID p_particles, int p_amount); @@ -2105,7 +2140,9 @@ public: /* PARTICLES COLLISION */ - virtual RID particles_collision_create(); + RID particles_collision_allocate(); + void particles_collision_initialize(RID p_particles_collision); + virtual void particles_collision_set_collision_type(RID p_particles_collision, RS::ParticlesCollisionType p_type); virtual void particles_collision_set_cull_mask(RID p_particles_collision, uint32_t p_cull_mask); virtual void particles_collision_set_sphere_radius(RID p_particles_collision, float p_radius); //for spheres @@ -2201,7 +2238,7 @@ public: void render_info_end_capture() {} int get_captured_render_info(RS::RenderInfo p_info) { return 0; } - int get_render_info(RS::RenderInfo p_info) { return 0; } + uint64_t get_render_info(RS::RenderInfo p_info) { return 0; } String get_video_adapter_name() const { return String(); } String get_video_adapter_vendor() const { return String(); } diff --git a/servers/rendering/renderer_rd/shader_compiler_rd.cpp b/servers/rendering/renderer_rd/shader_compiler_rd.cpp index e77141b26c..8135d388e1 100644 --- a/servers/rendering/renderer_rd/shader_compiler_rd.cpp +++ b/servers/rendering/renderer_rd/shader_compiler_rd.cpp @@ -687,7 +687,15 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge uint32_t index = p_default_actions.base_varying_index; + List<Pair<StringName, SL::ShaderNode::Varying>> var_frag_to_light; + for (Map<StringName, SL::ShaderNode::Varying>::Element *E = pnode->varyings.front(); E; E = E->next()) { + if (E->get().stage == SL::ShaderNode::Varying::STAGE_FRAGMENT_TO_LIGHT || E->get().stage == SL::ShaderNode::Varying::STAGE_FRAGMENT) { + var_frag_to_light.push_back(Pair<StringName, SL::ShaderNode::Varying>(E->key(), E->get())); + fragment_varyings.insert(E->key()); + continue; + } + String vcode; String interp_mode = _interpstr(E->get().interpolation); vcode += _prestr(E->get().precision); @@ -705,6 +713,21 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge index++; } + if (var_frag_to_light.size() > 0) { + String gcode = "\n\nstruct {\n"; + for (List<Pair<StringName, SL::ShaderNode::Varying>>::Element *E = var_frag_to_light.front(); E; E = E->next()) { + gcode += "\t" + _prestr(E->get().second.precision) + _typestr(E->get().second.type) + " " + _mkid(E->get().first); + if (E->get().second.array_size > 0) { + gcode += "["; + gcode += itos(E->get().second.array_size); + gcode += "]"; + } + gcode += ";\n"; + } + gcode += "} frag_to_light;\n"; + r_gen_code.fragment_global += gcode; + } + for (int i = 0; i < pnode->vconstants.size(); i++) { const SL::ShaderNode::Constant &cnode = pnode->vconstants[i]; String gcode; @@ -833,6 +856,19 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge } break; case SL::Node::TYPE_VARIABLE: { SL::VariableNode *vnode = (SL::VariableNode *)p_node; + bool use_fragment_varying = false; + + if (current_func_name != vertex_name) { + if (p_assigning) { + if (shader->varyings.has(vnode->name)) { + use_fragment_varying = true; + } + } else { + if (fragment_varyings.has(vnode->name)) { + use_fragment_varying = true; + } + } + } if (p_assigning && p_actions.write_flag_pointers.has(vnode->name)) { *p_actions.write_flag_pointers[vnode->name] = true; @@ -877,7 +913,10 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge } } else { - code = _mkid(vnode->name); //its something else (local var most likely) use as is + if (use_fragment_varying) { + code = "frag_to_light."; + } + code += _mkid(vnode->name); //its something else (local var most likely) use as is } } @@ -962,6 +1001,23 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge } break; case SL::Node::TYPE_ARRAY: { SL::ArrayNode *anode = (SL::ArrayNode *)p_node; + bool use_fragment_varying = false; + + if (current_func_name != vertex_name) { + if (anode->assign_expression != nullptr) { + use_fragment_varying = true; + } else { + if (p_assigning) { + if (shader->varyings.has(anode->name)) { + use_fragment_varying = true; + } + } else { + if (fragment_varyings.has(anode->name)) { + use_fragment_varying = true; + } + } + } + } if (p_assigning && p_actions.write_flag_pointers.has(anode->name)) { *p_actions.write_flag_pointers[anode->name] = true; @@ -984,7 +1040,10 @@ String ShaderCompilerRD::_dump_node_code(const SL::Node *p_node, int p_level, Ge if (p_default_actions.renames.has(anode->name)) { code = p_default_actions.renames[anode->name]; } else { - code = _mkid(anode->name); + if (use_fragment_varying) { + code = "frag_to_light."; + } + code += _mkid(anode->name); } if (anode->call_expression != nullptr) { @@ -1277,6 +1336,7 @@ Error ShaderCompilerRD::compile(RS::ShaderMode p_mode, const String &p_code, Ide used_name_defines.clear(); used_rmode_defines.clear(); used_flag_pointers.clear(); + fragment_varyings.clear(); shader = parser.get_shader(); function = nullptr; @@ -1408,7 +1468,7 @@ ShaderCompilerRD::ShaderCompilerRD() { actions[RS::SHADER_SPATIAL].render_mode_defines["cull_front"] = "#define DO_SIDE_CHECK\n"; actions[RS::SHADER_SPATIAL].render_mode_defines["cull_disabled"] = "#define DO_SIDE_CHECK\n"; - bool force_lambert = GLOBAL_GET("rendering/quality/shading/force_lambert_over_burley"); + bool force_lambert = GLOBAL_GET("rendering/shading/overrides/force_lambert_over_burley"); if (!force_lambert) { actions[RS::SHADER_SPATIAL].render_mode_defines["diffuse_burley"] = "#define DIFFUSE_BURLEY\n"; @@ -1418,7 +1478,7 @@ ShaderCompilerRD::ShaderCompilerRD() { actions[RS::SHADER_SPATIAL].render_mode_defines["diffuse_lambert_wrap"] = "#define DIFFUSE_LAMBERT_WRAP\n"; actions[RS::SHADER_SPATIAL].render_mode_defines["diffuse_toon"] = "#define DIFFUSE_TOON\n"; - bool force_blinn = GLOBAL_GET("rendering/quality/shading/force_blinn_over_ggx"); + bool force_blinn = GLOBAL_GET("rendering/shading/overrides/force_blinn_over_ggx"); if (!force_blinn) { actions[RS::SHADER_SPATIAL].render_mode_defines["specular_schlick_ggx"] = "#define SPECULAR_SCHLICK_GGX\n"; diff --git a/servers/rendering/renderer_rd/shader_compiler_rd.h b/servers/rendering/renderer_rd/shader_compiler_rd.h index d127d8e01c..6575829e73 100644 --- a/servers/rendering/renderer_rd/shader_compiler_rd.h +++ b/servers/rendering/renderer_rd/shader_compiler_rd.h @@ -114,6 +114,7 @@ private: Set<StringName> used_flag_pointers; Set<StringName> used_rmode_defines; Set<StringName> internal_functions; + Set<StringName> fragment_varyings; DefaultIdentifierActions actions; diff --git a/servers/rendering/renderer_rd/shader_rd.cpp b/servers/rendering/renderer_rd/shader_rd.cpp index e955cead05..e4a39ff813 100644 --- a/servers/rendering/renderer_rd/shader_rd.cpp +++ b/servers/rendering/renderer_rd/shader_rd.cpp @@ -301,6 +301,7 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { builder.append(compute_codev.get_data()); // version info (if exists) builder.append("\n"); //make sure defines begin at newline + builder.append(base_compute_defines.get_data()); builder.append(general_defines.get_data()); builder.append(variant_defines[p_variant].get_data()); @@ -351,6 +352,127 @@ void ShaderRD::_compile_variant(uint32_t p_variant, Version *p_version) { } } +RS::ShaderNativeSourceCode ShaderRD::version_get_native_source_code(RID p_version) { + Version *version = version_owner.getornull(p_version); + RS::ShaderNativeSourceCode source_code; + ERR_FAIL_COND_V(!version, source_code); + + source_code.versions.resize(variant_defines.size()); + + for (int i = 0; i < source_code.versions.size(); i++) { + if (!is_compute) { + //vertex stage + + StringBuilder builder; + + builder.append(vertex_codev.get_data()); // version info (if exists) + builder.append("\n"); //make sure defines begin at newline + builder.append(general_defines.get_data()); + builder.append(variant_defines[i].get_data()); + + for (int j = 0; j < version->custom_defines.size(); j++) { + builder.append(version->custom_defines[j].get_data()); + } + + builder.append(vertex_code0.get_data()); //first part of vertex + + builder.append(version->uniforms.get_data()); //uniforms (same for vertex and fragment) + + builder.append(vertex_code1.get_data()); //second part of vertex + + builder.append(version->vertex_globals.get_data()); // vertex globals + + builder.append(vertex_code2.get_data()); //third part of vertex + + builder.append(version->vertex_code.get_data()); // code + + builder.append(vertex_code3.get_data()); //fourth of vertex + + RS::ShaderNativeSourceCode::Version::Stage stage; + stage.name = "vertex"; + stage.code = builder.as_string(); + + source_code.versions.write[i].stages.push_back(stage); + } + + if (!is_compute) { + //fragment stage + + StringBuilder builder; + + builder.append(fragment_codev.get_data()); // version info (if exists) + builder.append("\n"); //make sure defines begin at newline + builder.append(general_defines.get_data()); + builder.append(variant_defines[i].get_data()); + for (int j = 0; j < version->custom_defines.size(); j++) { + builder.append(version->custom_defines[j].get_data()); + } + + builder.append(fragment_code0.get_data()); //first part of fragment + + builder.append(version->uniforms.get_data()); //uniforms (same for fragment and fragment) + + builder.append(fragment_code1.get_data()); //first part of fragment + + builder.append(version->fragment_globals.get_data()); // fragment globals + + builder.append(fragment_code2.get_data()); //third part of fragment + + builder.append(version->fragment_light.get_data()); // fragment light + + builder.append(fragment_code3.get_data()); //fourth part of fragment + + builder.append(version->fragment_code.get_data()); // fragment code + + builder.append(fragment_code4.get_data()); //fourth part of fragment + + RS::ShaderNativeSourceCode::Version::Stage stage; + stage.name = "fragment"; + stage.code = builder.as_string(); + + source_code.versions.write[i].stages.push_back(stage); + } + + if (is_compute) { + //compute stage + + StringBuilder builder; + + builder.append(compute_codev.get_data()); // version info (if exists) + builder.append("\n"); //make sure defines begin at newline + builder.append(base_compute_defines.get_data()); + builder.append(general_defines.get_data()); + builder.append(variant_defines[i].get_data()); + + for (int j = 0; j < version->custom_defines.size(); j++) { + builder.append(version->custom_defines[j].get_data()); + } + + builder.append(compute_code0.get_data()); //first part of compute + + builder.append(version->uniforms.get_data()); //uniforms (same for compute and fragment) + + builder.append(compute_code1.get_data()); //second part of compute + + builder.append(version->compute_globals.get_data()); // compute globals + + builder.append(compute_code2.get_data()); //third part of compute + + builder.append(version->compute_code.get_data()); // code + + builder.append(compute_code3.get_data()); //fourth of compute + + RS::ShaderNativeSourceCode::Version::Stage stage; + stage.name = "compute"; + stage.code = builder.as_string(); + + source_code.versions.write[i].stages.push_back(stage); + } + } + + return source_code; +} + void ShaderRD::_compile_version(Version *p_version) { _clear_version(p_version); @@ -475,6 +597,22 @@ bool ShaderRD::is_variant_enabled(int p_variant) const { return variants_enabled[p_variant]; } +ShaderRD::ShaderRD() { + // Do not feel forced to use this, in most cases it makes little to no difference. + bool use_32_threads = false; + if (RD::get_singleton()->get_device_vendor_name() == "NVIDIA") { + use_32_threads = true; + } + String base_compute_define_text; + if (use_32_threads) { + base_compute_define_text = "\n#define NATIVE_LOCAL_GROUP_SIZE 32\n#define NATIVE_LOCAL_SIZE_2D_X 8\n#define NATIVE_LOCAL_SIZE_2D_Y 4\n"; + } else { + base_compute_define_text = "\n#define NATIVE_LOCAL_GROUP_SIZE 64\n#define NATIVE_LOCAL_SIZE_2D_X 8\n#define NATIVE_LOCAL_SIZE_2D_Y 8\n"; + } + + base_compute_defines = base_compute_define_text.ascii(); +} + void ShaderRD::initialize(const Vector<String> &p_variant_defines, const String &p_general_defines) { ERR_FAIL_COND(variant_defines.size()); ERR_FAIL_COND(p_variant_defines.size() == 0); diff --git a/servers/rendering/renderer_rd/shader_rd.h b/servers/rendering/renderer_rd/shader_rd.h index a80d08050a..e0f4dcf2d0 100644 --- a/servers/rendering/renderer_rd/shader_rd.h +++ b/servers/rendering/renderer_rd/shader_rd.h @@ -36,6 +36,7 @@ #include "core/templates/map.h" #include "core/templates/rid_owner.h" #include "core/variant/variant.h" +#include "servers/rendering_server.h" #include <stdio.h> /** @@ -98,8 +99,10 @@ class ShaderRD { const char *name; + CharString base_compute_defines; + protected: - ShaderRD() {} + ShaderRD(); void setup(const char *p_vertex_code, const char *p_fragment_code, const char *p_compute_code, const char *p_name); public: @@ -133,6 +136,8 @@ public: void set_variant_enabled(int p_variant, bool p_enabled); bool is_variant_enabled(int p_variant) const; + RS::ShaderNativeSourceCode version_get_native_source_code(RID p_version); + void initialize(const Vector<String> &p_variant_defines, const String &p_general_defines = ""); virtual ~ShaderRD(); }; diff --git a/servers/rendering/renderer_rd/shaders/SCsub b/servers/rendering/renderer_rd/shaders/SCsub index deaa9668df..fc513d3fb9 100644 --- a/servers/rendering/renderer_rd/shaders/SCsub +++ b/servers/rendering/renderer_rd/shaders/SCsub @@ -3,44 +3,15 @@ Import("env") if "RD_GLSL" in env["BUILDERS"]: - env.RD_GLSL("canvas.glsl") - env.RD_GLSL("canvas_occlusion.glsl") - env.RD_GLSL("canvas_sdf.glsl") - env.RD_GLSL("copy.glsl") - env.RD_GLSL("copy_to_fb.glsl") - env.RD_GLSL("cubemap_roughness.glsl") - env.RD_GLSL("cubemap_downsampler.glsl") - env.RD_GLSL("cubemap_filter.glsl") - env.RD_GLSL("scene_forward.glsl") - env.RD_GLSL("sky.glsl") - env.RD_GLSL("tonemap.glsl") - env.RD_GLSL("cube_to_dp.glsl") - env.RD_GLSL("giprobe.glsl") - env.RD_GLSL("giprobe_debug.glsl") - env.RD_GLSL("giprobe_sdf.glsl") - env.RD_GLSL("luminance_reduce.glsl") - env.RD_GLSL("bokeh_dof.glsl") - env.RD_GLSL("ssao.glsl") - env.RD_GLSL("ssao_downsample.glsl") - env.RD_GLSL("ssao_importance_map.glsl") - env.RD_GLSL("ssao_blur.glsl") - env.RD_GLSL("ssao_interleave.glsl") - env.RD_GLSL("roughness_limiter.glsl") - env.RD_GLSL("screen_space_reflection.glsl") - env.RD_GLSL("screen_space_reflection_filter.glsl") - env.RD_GLSL("screen_space_reflection_scale.glsl") - env.RD_GLSL("subsurface_scattering.glsl") - env.RD_GLSL("specular_merge.glsl") - env.RD_GLSL("gi.glsl") - env.RD_GLSL("resolve.glsl") - env.RD_GLSL("sdfgi_preprocess.glsl") - env.RD_GLSL("sdfgi_integrate.glsl") - env.RD_GLSL("sdfgi_direct_light.glsl") - env.RD_GLSL("sdfgi_debug.glsl") - env.RD_GLSL("sdfgi_debug_probes.glsl") - env.RD_GLSL("volumetric_fog.glsl") - env.RD_GLSL("shadow_reduce.glsl") - env.RD_GLSL("particles.glsl") - env.RD_GLSL("particles_copy.glsl") - env.RD_GLSL("sort.glsl") - env.RD_GLSL("skeleton.glsl") + # find all include files + gl_include_files = [str(f) for f in Glob("*_inc.glsl")] + + # find all shader code(all glsl files excluding our include files) + glsl_files = [str(f) for f in Glob("*.glsl") if str(f) not in gl_include_files] + + # make sure we recompile shaders if include files change + env.Depends([f + ".gen.h" for f in glsl_files], gl_include_files) + + # compile shaders + for glsl_file in glsl_files: + env.RD_GLSL(glsl_file) diff --git a/servers/rendering/renderer_rd/shaders/canvas.glsl b/servers/rendering/renderer_rd/shaders/canvas.glsl index 9c4e95a7c2..3b39edc70e 100644 --- a/servers/rendering/renderer_rd/shaders/canvas.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas.glsl @@ -396,7 +396,7 @@ vec4 light_shadow_compute(uint light_base, vec4 light_color, vec4 shadow_uv vec4 shadow_color = unpackUnorm4x8(light_array.data[light_base].shadow_color); #ifdef LIGHT_SHADER_CODE_USED - shadow_color *= shadow_modulate; + shadow_color.rgb *= shadow_modulate; #endif shadow_color.a *= light_color.a; //respect light alpha @@ -546,7 +546,7 @@ FRAGMENT_SHADER_CODE #ifdef LIGHT_SHADER_CODE_USED vec4 shadow_modulate = vec4(1.0); - light_color = light_compute(light_vertex, direction, normal, light_color, light_color.a, specular_shininess, shadow_modulate, screen_uv, color, uv, true); + light_color = light_compute(light_vertex, vec3(direction, light_array.data[light_base].height), normal, light_color, light_color.a, specular_shininess, shadow_modulate, screen_uv, uv, color, true); #else if (normal_used) { @@ -563,7 +563,7 @@ FRAGMENT_SHADER_CODE light_color = light_shadow_compute(light_base, light_color, shadow_uv #ifdef LIGHT_SHADER_CODE_USED , - shadow_modulate + shadow_modulate.rgb #endif ); } @@ -605,7 +605,7 @@ FRAGMENT_SHADER_CODE vec3 light_position = vec3(light_array.data[light_base].position, light_array.data[light_base].height); light_color.rgb *= light_base_color.rgb; - light_color = light_compute(light_vertex, light_position, normal, light_color, light_base_color.a, specular_shininess, shadow_modulate, screen_uv, color, uv, false); + light_color = light_compute(light_vertex, light_position, normal, light_color, light_base_color.a, specular_shininess, shadow_modulate, screen_uv, uv, color, false); #else light_color.rgb *= light_base_color.rgb * light_base_color.a; @@ -659,7 +659,7 @@ FRAGMENT_SHADER_CODE light_color = light_shadow_compute(light_base, light_color, shadow_uv #ifdef LIGHT_SHADER_CODE_USED , - shadow_modulate + shadow_modulate.rgb #endif ); } diff --git a/servers/rendering/renderer_rd/shaders/cluster_data_inc.glsl b/servers/rendering/renderer_rd/shaders/cluster_data_inc.glsl index e723468dd8..3a4bf4da07 100644 --- a/servers/rendering/renderer_rd/shaders/cluster_data_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/cluster_data_inc.glsl @@ -6,12 +6,18 @@ struct LightData { //this structure needs to be as packed as possible vec3 position; float inv_radius; + vec3 direction; float size; - uint attenuation_energy; //attenuation - uint color_specular; //rgb color, a specular (8 bit unorm) - uint cone_attenuation_angle; // attenuation and angle, (16bit float) - uint shadow_color_enabled; //shadow rgb color, a>0.5 enabled (8bit unorm) + + vec3 color; + float attenuation; + + float cone_attenuation; + float cone_angle; + float specular_amount; + bool shadow_enabled; + vec4 atlas_rect; // rect in the shadow atlas mat4 shadow_matrix; float shadow_bias; @@ -34,9 +40,13 @@ struct ReflectionData { float index; vec3 box_offset; uint mask; - vec4 params; // intensity, 0, interior , boxproject vec3 ambient; // ambient color + float intensity; + bool exterior; + bool box_project; uint ambient_mode; + uint pad; + //0-8 is intensity,8-9 is ambient, mode mat4 local_matrix; // up to here for spot and omni, rest is for directional // notes: for ambientblend, use distance to edge to blend between already existing global environment }; diff --git a/servers/rendering/renderer_rd/shaders/cluster_debug.glsl b/servers/rendering/renderer_rd/shaders/cluster_debug.glsl new file mode 100644 index 0000000000..70a875192c --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cluster_debug.glsl @@ -0,0 +1,115 @@ +#[compute] + +#version 450 + +VERSION_DEFINES + +layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; + +const vec3 usage_gradient[33] = vec3[]( // 1 (none) + 32 + vec3(0.14, 0.17, 0.23), + vec3(0.24, 0.44, 0.83), + vec3(0.23, 0.57, 0.84), + vec3(0.22, 0.71, 0.84), + vec3(0.22, 0.85, 0.83), + vec3(0.21, 0.85, 0.72), + vec3(0.21, 0.85, 0.57), + vec3(0.20, 0.85, 0.42), + vec3(0.20, 0.85, 0.27), + vec3(0.27, 0.86, 0.19), + vec3(0.51, 0.85, 0.19), + vec3(0.57, 0.86, 0.19), + vec3(0.62, 0.85, 0.19), + vec3(0.67, 0.86, 0.20), + vec3(0.73, 0.85, 0.20), + vec3(0.78, 0.85, 0.20), + vec3(0.83, 0.85, 0.20), + vec3(0.85, 0.82, 0.20), + vec3(0.85, 0.76, 0.20), + vec3(0.85, 0.81, 0.20), + vec3(0.85, 0.65, 0.20), + vec3(0.84, 0.60, 0.21), + vec3(0.84, 0.56, 0.21), + vec3(0.84, 0.51, 0.21), + vec3(0.84, 0.46, 0.21), + vec3(0.84, 0.41, 0.21), + vec3(0.84, 0.36, 0.21), + vec3(0.84, 0.31, 0.21), + vec3(0.84, 0.27, 0.21), + vec3(0.83, 0.22, 0.22), + vec3(0.83, 0.22, 0.27), + vec3(0.83, 0.22, 0.32), + vec3(1.00, 0.63, 0.70)); +layout(push_constant, binding = 0, std430) uniform Params { + uvec2 screen_size; + uvec2 cluster_screen_size; + + uint cluster_shift; + uint cluster_type; + float z_near; + float z_far; + + bool orthogonal; + uint max_cluster_element_count_div_32; + uint pad1; + uint pad2; +} +params; + +layout(set = 0, binding = 1, std430) buffer restrict readonly ClusterData { + uint data[]; +} +cluster_data; + +layout(rgba16f, set = 0, binding = 2) uniform restrict writeonly image2D screen_buffer; +layout(set = 0, binding = 3) uniform texture2D depth_buffer; +layout(set = 0, binding = 4) uniform sampler depth_buffer_sampler; + +void main() { + uvec2 screen_pos = gl_GlobalInvocationID.xy; + if (any(greaterThanEqual(screen_pos, params.screen_size))) { + return; + } + + uvec2 cluster_pos = screen_pos >> params.cluster_shift; + + uint offset = cluster_pos.y * params.cluster_screen_size.x + cluster_pos.x; + offset += params.cluster_screen_size.x * params.cluster_screen_size.y * params.cluster_type; + offset *= (params.max_cluster_element_count_div_32 + 32); + + //depth buffers generally can't be accessed via image API + float depth = texelFetch(sampler2D(depth_buffer, depth_buffer_sampler), ivec2(screen_pos), 0).r * 2.0 - 1.0; + + if (params.orthogonal) { + depth = ((depth + (params.z_far + params.z_near) / (params.z_far - params.z_near)) * (params.z_far - params.z_near)) / 2.0; + } else { + depth = 2.0 * params.z_near * params.z_far / (params.z_far + params.z_near - depth * (params.z_far - params.z_near)); + } + depth /= params.z_far; + + uint slice = uint(clamp(floor(depth * 32.0), 0.0, 31.0)); + uint slice_minmax = cluster_data.data[offset + params.max_cluster_element_count_div_32 + slice]; + uint item_min = slice_minmax & 0xFFFF; + uint item_max = slice_minmax >> 16; + + uint item_count = 0; + for (uint i = 0; i < params.max_cluster_element_count_div_32; i++) { + uint slice_bits = cluster_data.data[offset + i]; + while (slice_bits != 0) { + uint bit = findLSB(slice_bits); + uint item = i * 32 + bit; + if ((item >= item_min && item < item_max)) { + item_count++; + } + slice_bits &= ~(1 << bit); + } + } + + item_count = min(item_count, 32); + + vec3 color = usage_gradient[item_count]; + + color = mix(color * 1.2, color * 0.3, float(slice) / 31.0); + + imageStore(screen_buffer, ivec2(screen_pos), vec4(color, 1.0)); +} diff --git a/servers/rendering/renderer_rd/shaders/cluster_render.glsl b/servers/rendering/renderer_rd/shaders/cluster_render.glsl new file mode 100644 index 0000000000..ca92d2104e --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cluster_render.glsl @@ -0,0 +1,168 @@ +#[vertex] + +#version 450 + +VERSION_DEFINES + +layout(location = 0) in vec3 vertex_attrib; + +layout(location = 0) out float depth_interp; +layout(location = 1) out flat uint element_index; + +layout(push_constant, binding = 0, std430) uniform Params { + uint base_index; + uint pad0; + uint pad1; + uint pad2; +} +params; + +layout(set = 0, binding = 1, std140) uniform State { + mat4 projection; + + float inv_z_far; + uint screen_to_clusters_shift; // shift to obtain coordinates in block indices + uint cluster_screen_width; // + uint cluster_data_size; // how much data for a single cluster takes + + uint cluster_depth_offset; + uint pad0; + uint pad1; + uint pad2; +} +state; + +struct RenderElement { + uint type; //0-4 + bool touches_near; + bool touches_far; + uint original_index; + mat3x4 transform_inv; + vec3 scale; + uint pad; +}; + +layout(set = 0, binding = 2, std430) buffer restrict readonly RenderElements { + RenderElement data[]; +} +render_elements; + +void main() { + element_index = params.base_index + gl_InstanceIndex; + + vec3 vertex = vertex_attrib; + vertex *= render_elements.data[element_index].scale; + + vertex = vec4(vertex, 1.0) * render_elements.data[element_index].transform_inv; + depth_interp = -vertex.z; + + gl_Position = state.projection * vec4(vertex, 1.0); +} + +#[fragment] + +#version 450 + +VERSION_DEFINES + +#if defined(has_GL_KHR_shader_subgroup_ballot) && defined(has_GL_KHR_shader_subgroup_arithmetic) && defined(has_GL_KHR_shader_subgroup_vote) + +#extension GL_KHR_shader_subgroup_ballot : enable +#extension GL_KHR_shader_subgroup_arithmetic : enable +#extension GL_KHR_shader_subgroup_vote : enable + +#define USE_SUBGROUPS +#endif + +layout(location = 0) in float depth_interp; +layout(location = 1) in flat uint element_index; + +layout(set = 0, binding = 1, std140) uniform State { + mat4 projection; + float inv_z_far; + uint screen_to_clusters_shift; // shift to obtain coordinates in block indices + uint cluster_screen_width; // + uint cluster_data_size; // how much data for a single cluster takes + uint cluster_depth_offset; + uint pad0; + uint pad1; + uint pad2; +} +state; + +//cluster data is layout linearly, each cell contains the follow information: +// - list of bits for every element to mark as used, so (max_elem_count/32)*4 uints +// - a uint for each element to mark the depth bits used when rendering (0-31) + +layout(set = 0, binding = 3, std430) buffer restrict ClusterRender { + uint data[]; +} +cluster_render; + +void main() { + //convert from screen to cluster + uvec2 cluster = uvec2(gl_FragCoord.xy) >> state.screen_to_clusters_shift; + + //get linear cluster offset from screen poss + uint cluster_offset = cluster.x + state.cluster_screen_width * cluster.y; + //multiply by data size to position at the beginning of the element list for this cluster + cluster_offset *= state.cluster_data_size; + + //find the current element in the list and plot the bit to mark it as used + uint usage_write_offset = cluster_offset + (element_index >> 5); + uint usage_write_bit = 1 << (element_index & 0x1F); + +#ifdef USE_SUBGROUPS + + uint cluster_thread_group_index; + + if (!gl_HelperInvocation) { + //http://advances.realtimerendering.com/s2017/2017_Sig_Improved_Culling_final.pdf + + uvec4 mask; + + while (true) { + // find the cluster offset of the first active thread + // threads that did break; go inactive and no longer count + uint first = subgroupBroadcastFirst(cluster_offset); + // update the mask for thread that match this cluster + mask = subgroupBallot(first == cluster_offset); + if (first == cluster_offset) { + // This thread belongs to the group of threads that match this offset, + // so exit the loop. + break; + } + } + + cluster_thread_group_index = subgroupBallotExclusiveBitCount(mask); + + if (cluster_thread_group_index == 0) { + atomicOr(cluster_render.data[usage_write_offset], usage_write_bit); + } + } +#else + if (!gl_HelperInvocation) { + atomicOr(cluster_render.data[usage_write_offset], usage_write_bit); + } +#endif + //find the current element in the depth usage list and mark the current depth as used + float unit_depth = depth_interp * state.inv_z_far; + + uint z_bit = clamp(uint(floor(unit_depth * 32.0)), 0, 31); + + uint z_write_offset = cluster_offset + state.cluster_depth_offset + element_index; + uint z_write_bit = 1 << z_bit; + +#ifdef USE_SUBGROUPS + if (!gl_HelperInvocation) { + z_write_bit = subgroupOr(z_write_bit); //merge all Zs + if (cluster_thread_group_index == 0) { + atomicOr(cluster_render.data[z_write_offset], z_write_bit); + } + } +#else + if (!gl_HelperInvocation) { + atomicOr(cluster_render.data[z_write_offset], z_write_bit); + } +#endif +} diff --git a/servers/rendering/renderer_rd/shaders/cluster_store.glsl b/servers/rendering/renderer_rd/shaders/cluster_store.glsl new file mode 100644 index 0000000000..5be0893c4f --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cluster_store.glsl @@ -0,0 +1,119 @@ +#[compute] + +#version 450 + +VERSION_DEFINES + +layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; + +layout(push_constant, binding = 0, std430) uniform Params { + uint cluster_render_data_size; // how much data for a single cluster takes + uint max_render_element_count_div_32; //divided by 32 + uvec2 cluster_screen_size; + uint render_element_count_div_32; //divided by 32 + + uint max_cluster_element_count_div_32; //divided by 32 + uint pad1; + uint pad2; +} +params; + +layout(set = 0, binding = 1, std430) buffer restrict readonly ClusterRender { + uint data[]; +} +cluster_render; + +layout(set = 0, binding = 2, std430) buffer restrict ClusterStore { + uint data[]; +} +cluster_store; + +struct RenderElement { + uint type; //0-4 + bool touches_near; + bool touches_far; + uint original_index; + mat3x4 transform_inv; + vec3 scale; + uint pad; +}; + +layout(set = 0, binding = 3, std430) buffer restrict readonly RenderElements { + RenderElement data[]; +} +render_elements; + +void main() { + uvec2 pos = gl_GlobalInvocationID.xy; + if (any(greaterThanEqual(pos, params.cluster_screen_size))) { + return; + } + + //counter for each type of render_element + + //base offset for this cluster + uint base_offset = (pos.x + params.cluster_screen_size.x * pos.y); + uint src_offset = base_offset * params.cluster_render_data_size; + + uint render_element_offset = 0; + + //check all render_elements and see which one was written to + while (render_element_offset < params.render_element_count_div_32) { + uint bits = cluster_render.data[src_offset + render_element_offset]; + while (bits != 0) { + //if bits exist, check the render_element + uint index_bit = findLSB(bits); + uint index = render_element_offset * 32 + index_bit; + uint type = render_elements.data[index].type; + + uint z_range_offset = src_offset + params.max_render_element_count_div_32 + index; + uint z_range = cluster_render.data[z_range_offset]; + + //if object was written, z was written, but check just in case + if (z_range != 0) { //should always be > 0 + + uint from_z = findLSB(z_range); + uint to_z = findMSB(z_range) + 1; + + if (render_elements.data[index].touches_near) { + from_z = 0; + } + + if (render_elements.data[index].touches_far) { + to_z = 32; + } + + // find cluster offset in the buffer used for indexing in the renderer + uint dst_offset = (base_offset + type * (params.cluster_screen_size.x * params.cluster_screen_size.y)) * (params.max_cluster_element_count_div_32 + 32); + + uint orig_index = render_elements.data[index].original_index; + //store this index in the Z slices by setting the relevant bit + for (uint i = from_z; i < to_z; i++) { + uint slice_ofs = dst_offset + params.max_cluster_element_count_div_32 + i; + + uint minmax = cluster_store.data[slice_ofs]; + + if (minmax == 0) { + minmax = 0xFFFF; //min 0, max 0xFFFF + } + + uint elem_min = min(orig_index, minmax & 0xFFFF); + uint elem_max = max(orig_index + 1, minmax >> 16); //always store plus one, so zero means range is empty when not written to + + minmax = elem_min | (elem_max << 16); + cluster_store.data[slice_ofs] = minmax; + } + + uint store_word = orig_index >> 5; + uint store_bit = orig_index & 0x1F; + + //store the actual render_element index at the end, so the rendering code can reference it + cluster_store.data[dst_offset + store_word] |= 1 << store_bit; + } + + bits &= ~(1 << index_bit); //clear the bit to continue iterating + } + + render_element_offset++; + } +} diff --git a/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl b/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl index 54d67db6c6..c3ac0bee57 100644 --- a/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl +++ b/servers/rendering/renderer_rd/shaders/cube_to_dp.glsl @@ -1,33 +1,48 @@ -#[compute] +#[vertex] #version 450 VERSION_DEFINES -layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; +layout(push_constant, binding = 1, std430) uniform Params { + float z_far; + float z_near; + bool z_flip; + uint pad; + vec4 screen_rect; +} +params; + +layout(location = 0) out vec2 uv_interp; + +void main() { + vec2 base_arr[4] = vec2[](vec2(0.0, 0.0), vec2(0.0, 1.0), vec2(1.0, 1.0), vec2(1.0, 0.0)); + uv_interp = base_arr[gl_VertexIndex]; + vec2 screen_pos = uv_interp * params.screen_rect.zw + params.screen_rect.xy; + gl_Position = vec4(screen_pos * 2.0 - 1.0, 0.0, 1.0); +} + +#[fragment] + +#version 450 + +VERSION_DEFINES + +layout(location = 0) in vec2 uv_interp; layout(set = 0, binding = 0) uniform samplerCube source_cube; layout(push_constant, binding = 1, std430) uniform Params { - ivec2 screen_size; - ivec2 offset; - float bias; float z_far; float z_near; bool z_flip; + uint pad; + vec4 screen_rect; } params; -layout(r32f, set = 1, binding = 0) uniform restrict writeonly image2D depth_buffer; - void main() { - ivec2 pos = ivec2(gl_GlobalInvocationID.xy); - if (any(greaterThan(pos, params.screen_size))) { //too large, do nothing - return; - } - - vec2 pixel_size = 1.0 / vec2(params.screen_size); - vec2 uv = (vec2(pos) + 0.5) * pixel_size; + vec2 uv = uv_interp; vec3 normal = vec3(uv * 2.0 - 1.0, 0.0); @@ -65,5 +80,5 @@ void main() { float linear_depth = 2.0 * params.z_near * params.z_far / (params.z_far + params.z_near - depth * (params.z_far - params.z_near)); depth = (linear_depth * depth_fix) / params.z_far; - imageStore(depth_buffer, pos + params.offset, vec4(depth)); + gl_FragDepth = depth; } diff --git a/servers/rendering/renderer_rd/shaders/gi.glsl b/servers/rendering/renderer_rd/shaders/gi.glsl index 8011dadc72..92a5682572 100644 --- a/servers/rendering/renderer_rd/shaders/gi.glsl +++ b/servers/rendering/renderer_rd/shaders/gi.glsl @@ -97,13 +97,12 @@ layout(push_constant, binding = 0, std430) uniform Params { vec4 proj_info; + vec3 ao_color; uint max_giprobes; + bool high_quality_vct; - bool use_sdfgi; bool orthogonal; - - vec3 ao_color; - uint pad; + uint pad[2]; mat3x4 cam_rotation; } @@ -331,7 +330,7 @@ void sdfgi_process(vec3 vertex, vec3 normal, vec3 reflection, float roughness, o } ambient_light.rgb = diffuse; -#if 1 + if (roughness < 0.2) { vec3 pos_to_uvw = 1.0 / sdfgi.grid_size; vec4 light_accum = vec4(0.0); @@ -363,59 +362,63 @@ void sdfgi_process(vec3 vertex, vec3 normal, vec3 reflection, float roughness, o //ray_pos += ray_dir * (bias / sdfgi.cascades[cascade].to_cell); //bias to avoid self occlusion ray_pos += (ray_dir * 1.0 / max(abs_ray_dir.x, max(abs_ray_dir.y, abs_ray_dir.z)) + cam_normal * 1.4) * bias / sdfgi.cascades[cascade].to_cell; } - float softness = 0.2 + min(1.0, roughness * 5.0) * 4.0; //approximation to roughness so it does not seem like a hard fade - while (length(ray_pos) < max_distance) { - for (uint i = 0; i < sdfgi.max_cascades; i++) { - if (i >= cascade && length(ray_pos) < radius_sizes[i]) { - cascade = max(i, cascade); //never go down - - vec3 pos = ray_pos - sdfgi.cascades[i].position; - pos *= sdfgi.cascades[i].to_cell * pos_to_uvw; - - float distance = texture(sampler3D(sdf_cascades[i], linear_sampler), pos).r * 255.0 - 1.1; - - vec4 hit_light = vec4(0.0); - if (distance < softness) { - hit_light.rgb = texture(sampler3D(light_cascades[i], linear_sampler), pos).rgb; - hit_light.rgb *= 0.5; //approximation given value read is actually meant for anisotropy - hit_light.a = clamp(1.0 - (distance / softness), 0.0, 1.0); - hit_light.rgb *= hit_light.a; - } + uint i = 0; + bool found = false; + while (true) { + if (length(ray_pos) >= max_distance || light_accum.a > 0.99) { + break; + } + if (!found && i >= cascade && length(ray_pos) < radius_sizes[i]) { + uint next_i = min(i + 1, sdfgi.max_cascades - 1); + cascade = max(i, cascade); //never go down - distance /= sdfgi.cascades[i].to_cell; + vec3 pos = ray_pos - sdfgi.cascades[i].position; + pos *= sdfgi.cascades[i].to_cell * pos_to_uvw; - if (i < (sdfgi.max_cascades - 1)) { - pos = ray_pos - sdfgi.cascades[i + 1].position; - pos *= sdfgi.cascades[i + 1].to_cell * pos_to_uvw; + float fdistance = textureLod(sampler3D(sdf_cascades[i], linear_sampler), pos, 0.0).r * 255.0 - 1.1; - float distance2 = texture(sampler3D(sdf_cascades[i + 1], linear_sampler), pos).r * 255.0 - 1.1; + vec4 hit_light = vec4(0.0); + if (fdistance < softness) { + hit_light.rgb = textureLod(sampler3D(light_cascades[i], linear_sampler), pos, 0.0).rgb; + hit_light.rgb *= 0.5; //approximation given value read is actually meant for anisotropy + hit_light.a = clamp(1.0 - (fdistance / softness), 0.0, 1.0); + hit_light.rgb *= hit_light.a; + } - vec4 hit_light2 = vec4(0.0); - if (distance2 < softness) { - hit_light2.rgb = texture(sampler3D(light_cascades[i + 1], linear_sampler), pos).rgb; - hit_light2.rgb *= 0.5; //approximation given value read is actually meant for anisotropy - hit_light2.a = clamp(1.0 - (distance2 / softness), 0.0, 1.0); - hit_light2.rgb *= hit_light2.a; - } + fdistance /= sdfgi.cascades[i].to_cell; - float prev_radius = i == 0 ? 0.0 : radius_sizes[i - 1]; - float blend = clamp((length(ray_pos) - prev_radius) / (radius_sizes[i] - prev_radius), 0.0, 1.0); + if (i < (sdfgi.max_cascades - 1)) { + pos = ray_pos - sdfgi.cascades[next_i].position; + pos *= sdfgi.cascades[next_i].to_cell * pos_to_uvw; - distance2 /= sdfgi.cascades[i + 1].to_cell; + float fdistance2 = textureLod(sampler3D(sdf_cascades[next_i], linear_sampler), pos, 0.0).r * 255.0 - 1.1; - hit_light = mix(hit_light, hit_light2, blend); - distance = mix(distance, distance2, blend); + vec4 hit_light2 = vec4(0.0); + if (fdistance2 < softness) { + hit_light2.rgb = textureLod(sampler3D(light_cascades[next_i], linear_sampler), pos, 0.0).rgb; + hit_light2.rgb *= 0.5; //approximation given value read is actually meant for anisotropy + hit_light2.a = clamp(1.0 - (fdistance2 / softness), 0.0, 1.0); + hit_light2.rgb *= hit_light2.a; } - light_accum += hit_light; - ray_pos += ray_dir * distance; - break; + float prev_radius = i == 0 ? 0.0 : radius_sizes[max(0, i - 1)]; + float blend = clamp((length(ray_pos) - prev_radius) / (radius_sizes[i] - prev_radius), 0.0, 1.0); + + fdistance2 /= sdfgi.cascades[next_i].to_cell; + + hit_light = mix(hit_light, hit_light2, blend); + fdistance = mix(fdistance, fdistance2, blend); } - } - if (light_accum.a > 0.99) { - break; + light_accum += hit_light; + ray_pos += ray_dir * fdistance; + found = true; + } + i++; + if (i == sdfgi.max_cascades) { + i = 0; + found = false; } } @@ -434,8 +437,6 @@ void sdfgi_process(vec3 vertex, vec3 normal, vec3 reflection, float roughness, o } } -#endif - reflection_light.rgb = specular; ambient_light.rgb *= sdfgi.energy; @@ -597,35 +598,24 @@ vec4 fetch_normal_and_roughness(ivec2 pos) { return normal_roughness; } -void main() { - // Pixel being shaded - ivec2 pos = ivec2(gl_GlobalInvocationID.xy); - if (any(greaterThanEqual(pos, params.screen_size))) { //too large, do nothing - return; - } - - vec3 vertex = reconstruct_position(pos); - vertex.y = -vertex.y; - +void process_gi(ivec2 pos, vec3 vertex, inout vec4 ambient_light, inout vec4 reflection_light) { vec4 normal_roughness = fetch_normal_and_roughness(pos); - vec3 normal = normal_roughness.xyz; - vec4 ambient_light = vec4(0.0), reflection_light = vec4(0.0); + vec3 normal = normal_roughness.xyz; if (normal.length() > 0.5) { //valid normal, can do GI float roughness = normal_roughness.w; - vertex = mat3(params.cam_rotation) * vertex; normal = normalize(mat3(params.cam_rotation) * normal); - vec3 reflection = normalize(reflect(normalize(vertex), normal)); - if (params.use_sdfgi) { - sdfgi_process(vertex, normal, reflection, roughness, ambient_light, reflection_light); - } +#ifdef USE_SDFGI + sdfgi_process(vertex, normal, reflection, roughness, ambient_light, reflection_light); +#endif - if (params.max_giprobes > 0) { +#ifdef USE_GIPROBES + { uvec2 giprobe_tex = texelFetch(usampler2D(giprobe_buffer, linear_sampler), pos, 0).rg; roughness *= roughness; //find arbitrary tangent and bitangent, then build a matrix @@ -648,16 +638,40 @@ void main() { spec_accum /= blend_accum; } - if (params.use_sdfgi) { - reflection_light = blend_color(spec_accum, reflection_light); - ambient_light = blend_color(amb_accum, ambient_light); - } else { - reflection_light = spec_accum; - ambient_light = amb_accum; - } +#ifdef USE_SDFGI + reflection_light = blend_color(spec_accum, reflection_light); + ambient_light = blend_color(amb_accum, ambient_light); +#else + reflection_light = spec_accum; + ambient_light = amb_accum; +#endif } +#endif + } +} + +void main() { + ivec2 pos = ivec2(gl_GlobalInvocationID.xy); + +#ifdef MODE_HALF_RES + pos <<= 1; +#endif + if (any(greaterThanEqual(pos, params.screen_size))) { //too large, do nothing + return; } + vec4 ambient_light = vec4(0.0); + vec4 reflection_light = vec4(0.0); + + vec3 vertex = reconstruct_position(pos); + vertex.y = -vertex.y; + + process_gi(pos, vertex, ambient_light, reflection_light); + +#ifdef MODE_HALF_RES + pos >>= 1; +#endif + imageStore(ambient_buffer, pos, ambient_light); imageStore(reflection_buffer, pos, reflection_light); } diff --git a/servers/rendering/renderer_rd/shaders/giprobe.glsl b/servers/rendering/renderer_rd/shaders/giprobe.glsl index 4f4753d147..b931461b31 100644 --- a/servers/rendering/renderer_rd/shaders/giprobe.glsl +++ b/servers/rendering/renderer_rd/shaders/giprobe.glsl @@ -51,10 +51,10 @@ struct Light { float attenuation; vec3 color; - float spot_angle_radians; + float cos_spot_angle; vec3 position; - float spot_attenuation; + float inv_spot_attenuation; vec3 direction; bool has_shadow; @@ -233,13 +233,15 @@ bool compute_light_vector(uint light, vec3 pos, out float attenuation, out vec3 if (lights.data[light].type == LIGHT_TYPE_SPOT) { vec3 rel = normalize(pos - light_pos); - float angle = acos(dot(rel, lights.data[light].direction)); - if (angle > lights.data[light].spot_angle_radians) { + float cos_spot_angle = lights.data[light].cos_spot_angle; + float cos_angle = dot(rel, lights.data[light].direction); + if (cos_angle < cos_spot_angle) { return false; } - float d = clamp(angle / lights.data[light].spot_angle_radians, 0, 1); - attenuation *= pow(1.0 - d, lights.data[light].spot_attenuation); + float scos = max(cos_angle, cos_spot_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - cos_spot_angle)); + attenuation *= 1.0 - pow(spot_rim, lights.data[light].inv_spot_attenuation); } } diff --git a/servers/rendering/renderer_rd/shaders/giprobe_write.glsl b/servers/rendering/renderer_rd/shaders/giprobe_write.glsl index 9c794f1bcc..56b3b7ccb4 100644 --- a/servers/rendering/renderer_rd/shaders/giprobe_write.glsl +++ b/servers/rendering/renderer_rd/shaders/giprobe_write.glsl @@ -43,10 +43,10 @@ struct Light { float attenuation; vec3 color; - float spot_angle_radians; + float cos_spot_angle; vec3 position; - float spot_attenuation; + float inv_spot_attenuation; vec3 direction; bool has_shadow; @@ -146,13 +146,15 @@ bool compute_light_vector(uint light, uint cell, vec3 pos, out float attenuation if (lights.data[light].type == LIGHT_TYPE_SPOT) { vec3 rel = normalize(pos - light_pos); - float angle = acos(dot(rel, lights.data[light].direction)); - if (angle > lights.data[light].spot_angle_radians) { + float cos_spot_angle = lights.data[light].cos_spot_angle; + float cos_angle = dot(rel, lights.data[light].direction); + if (cos_angle < cos_spot_angle) { return false; } - float d = clamp(angle / lights.data[light].spot_angle_radians, 0, 1); - attenuation *= pow(1.0 - d, lights.data[light].spot_attenuation); + float scos = max(cos_angle, cos_spot_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - cos_spot_angle)); + attenuation *= 1.0 - pow(spot_rim, lights.data[light].inv_spot_attenuation); } } diff --git a/servers/rendering/renderer_rd/shaders/particles.glsl b/servers/rendering/renderer_rd/shaders/particles.glsl index 926c7ef9fc..cb6d8dc7f6 100644 --- a/servers/rendering/renderer_rd/shaders/particles.glsl +++ b/servers/rendering/renderer_rd/shaders/particles.glsl @@ -173,7 +173,7 @@ uint hash(uint x) { return x; } -bool emit_particle(mat4 p_xform, vec3 p_velocity, vec4 p_color, vec4 p_custom, uint p_flags) { +bool emit_subparticle(mat4 p_xform, vec3 p_velocity, vec4 p_color, vec4 p_custom, uint p_flags) { if (!params.can_emit) { return false; } diff --git a/servers/rendering/renderer_rd/shaders/resolve.glsl b/servers/rendering/renderer_rd/shaders/resolve.glsl index 9429a66dc9..e83c4ca93b 100644 --- a/servers/rendering/renderer_rd/shaders/resolve.glsl +++ b/servers/rendering/renderer_rd/shaders/resolve.glsl @@ -58,6 +58,116 @@ void main() { #else +#if 1 + + vec4 group1; + vec4 group2; + vec4 group3; + vec4 group4; + int best_index = 0; + + //2X + group1.x = texelFetch(source_depth, pos, 0).r; + group1.y = texelFetch(source_depth, pos, 1).r; + + //4X + if (params.sample_count >= 4) { + group1.z = texelFetch(source_depth, pos, 2).r; + group1.w = texelFetch(source_depth, pos, 3).r; + } + //8X + if (params.sample_count >= 8) { + group2.x = texelFetch(source_depth, pos, 4).r; + group2.y = texelFetch(source_depth, pos, 5).r; + group2.z = texelFetch(source_depth, pos, 6).r; + group2.w = texelFetch(source_depth, pos, 7).r; + } + //16X + if (params.sample_count >= 16) { + group3.x = texelFetch(source_depth, pos, 8).r; + group3.y = texelFetch(source_depth, pos, 9).r; + group3.z = texelFetch(source_depth, pos, 10).r; + group3.w = texelFetch(source_depth, pos, 11).r; + + group4.x = texelFetch(source_depth, pos, 12).r; + group4.y = texelFetch(source_depth, pos, 13).r; + group4.z = texelFetch(source_depth, pos, 14).r; + group4.w = texelFetch(source_depth, pos, 15).r; + } + + if (params.sample_count == 2) { + best_index = (pos.x & 1) ^ ((pos.y >> 1) & 1); //not much can be done here + } else if (params.sample_count == 4) { + vec4 freq = vec4(equal(group1, vec4(group1.x))); + freq += vec4(equal(group1, vec4(group1.y))); + freq += vec4(equal(group1, vec4(group1.z))); + freq += vec4(equal(group1, vec4(group1.w))); + + float min_f = freq.x; + best_index = 0; + if (freq.y < min_f) { + best_index = 1; + min_f = freq.y; + } + if (freq.z < min_f) { + best_index = 2; + min_f = freq.z; + } + if (freq.w < min_f) { + best_index = 3; + } + } else if (params.sample_count == 8) { + vec4 freq0 = vec4(equal(group1, vec4(group1.x))); + vec4 freq1 = vec4(equal(group2, vec4(group1.x))); + freq0 += vec4(equal(group1, vec4(group1.y))); + freq1 += vec4(equal(group2, vec4(group1.y))); + freq0 += vec4(equal(group1, vec4(group1.z))); + freq1 += vec4(equal(group2, vec4(group1.z))); + freq0 += vec4(equal(group1, vec4(group1.w))); + freq1 += vec4(equal(group2, vec4(group1.w))); + freq0 += vec4(equal(group1, vec4(group2.x))); + freq1 += vec4(equal(group2, vec4(group2.x))); + freq0 += vec4(equal(group1, vec4(group2.y))); + freq1 += vec4(equal(group2, vec4(group2.y))); + freq0 += vec4(equal(group1, vec4(group2.z))); + freq1 += vec4(equal(group2, vec4(group2.z))); + freq0 += vec4(equal(group1, vec4(group2.w))); + freq1 += vec4(equal(group2, vec4(group2.w))); + + float min_f0 = freq0.x; + int best_index0 = 0; + if (freq0.y < min_f0) { + best_index0 = 1; + min_f0 = freq0.y; + } + if (freq0.z < min_f0) { + best_index0 = 2; + min_f0 = freq0.z; + } + if (freq0.w < min_f0) { + best_index0 = 3; + min_f0 = freq0.w; + } + + float min_f1 = freq1.x; + int best_index1 = 4; + if (freq1.y < min_f1) { + best_index1 = 5; + min_f1 = freq1.y; + } + if (freq1.z < min_f1) { + best_index1 = 6; + min_f1 = freq1.z; + } + if (freq1.w < min_f1) { + best_index1 = 7; + min_f1 = freq1.w; + } + + best_index = mix(best_index0, best_index1, min_f0 < min_f1); + } + +#else float depths[16]; int depth_indices[16]; int depth_amount[16]; @@ -91,7 +201,7 @@ void main() { depth_least = depth_amount[j]; } } - +#endif best_depth = texelFetch(source_depth, pos, best_index).r; best_normal_roughness = texelFetch(source_normal_roughness, pos, best_index); #ifdef GIPROBE_RESOLVE diff --git a/servers/rendering/renderer_rd/shaders/scene_forward.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl index 0518976322..7b86dac143 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl @@ -4,7 +4,7 @@ VERSION_DEFINES -#include "scene_forward_inc.glsl" +#include "scene_forward_clustered_inc.glsl" /* INPUT ATTRIBS */ @@ -89,12 +89,6 @@ MATERIAL_UNIFORMS } material; #endif -/* clang-format off */ - -VERTEX_SHADER_GLOBALS - -/* clang-format on */ - invariant gl_Position; #ifdef MODE_DUAL_PARABOLOID @@ -103,28 +97,43 @@ layout(location = 8) out float dp_clip; #endif +layout(location = 9) out flat uint instance_index; + +/* clang-format off */ + +VERTEX_SHADER_GLOBALS + +/* clang-format on */ + void main() { vec4 instance_custom = vec4(0.0); #if defined(COLOR_USED) color_interp = color_attrib; #endif - mat4 world_matrix = draw_call.transform; + instance_index = draw_call.instance_index; + + bool is_multimesh = bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH); + if (!is_multimesh) { + instance_index += gl_InstanceIndex; + } + + mat4 world_matrix = instances.data[instance_index].transform; mat3 world_normal_matrix; - if (bool(draw_call.flags & INSTANCE_FLAGS_NON_UNIFORM_SCALE)) { + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_NON_UNIFORM_SCALE)) { world_normal_matrix = inverse(mat3(world_matrix)); } else { world_normal_matrix = mat3(world_matrix); } - if (bool(draw_call.flags & INSTANCE_FLAGS_MULTIMESH)) { + if (is_multimesh) { //multimesh, instances are for it - uint offset = (draw_call.flags >> INSTANCE_FLAGS_MULTIMESH_STRIDE_SHIFT) & INSTANCE_FLAGS_MULTIMESH_STRIDE_MASK; + uint offset = (instances.data[instance_index].flags >> INSTANCE_FLAGS_MULTIMESH_STRIDE_SHIFT) & INSTANCE_FLAGS_MULTIMESH_STRIDE_MASK; offset *= gl_InstanceIndex; mat4 matrix; - if (bool(draw_call.flags & INSTANCE_FLAGS_MULTIMESH_FORMAT_2D)) { + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_FORMAT_2D)) { matrix = mat4(transforms.data[offset + 0], transforms.data[offset + 1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0)); offset += 2; } else { @@ -132,14 +141,14 @@ void main() { offset += 3; } - if (bool(draw_call.flags & INSTANCE_FLAGS_MULTIMESH_HAS_COLOR)) { + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_HAS_COLOR)) { #ifdef COLOR_USED color_interp *= transforms.data[offset]; #endif offset += 1; } - if (bool(draw_call.flags & INSTANCE_FLAGS_MULTIMESH_HAS_CUSTOM_DATA)) { + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH_HAS_CUSTOM_DATA)) { instance_custom = transforms.data[offset]; } @@ -161,7 +170,7 @@ void main() { #endif #if 0 - if (bool(draw_call.flags & INSTANCE_FLAGS_SKELETON)) { + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_SKELETON)) { //multimesh, instances are for it uvec2 bones_01 = uvec2(bone_attrib.x & 0xFFFF, bone_attrib.x >> 16) * 3; @@ -194,7 +203,7 @@ void main() { uv2_interp = uv2_attrib; #endif -#ifdef USE_OVERRIDE_POSITION +#ifdef OVERRIDE_POSITION vec4 position; #endif @@ -289,7 +298,7 @@ VERTEX_SHADER_CODE #endif //MODE_RENDER_DEPTH -#ifdef USE_OVERRIDE_POSITION +#ifdef OVERRIDE_POSITION gl_Position = position; #else gl_Position = projection_matrix * vec4(vertex_interp, 1.0); @@ -304,7 +313,8 @@ VERTEX_SHADER_CODE #endif #ifdef MODE_RENDER_MATERIAL if (scene_data.material_uv2_mode) { - gl_Position.xy = (uv2_attrib.xy + draw_call.lightmap_uv_scale.xy) * 2.0 - 1.0; + vec2 uv_offset = unpackHalf2x16(draw_call.uv_offset); + gl_Position.xy = (uv2_attrib.xy + uv_offset) * 2.0 - 1.0; gl_Position.z = 0.00001; gl_Position.w = 1.0; } @@ -317,7 +327,7 @@ VERTEX_SHADER_CODE VERSION_DEFINES -#include "scene_forward_inc.glsl" +#include "scene_forward_clustered_inc.glsl" /* Varyings */ @@ -350,9 +360,11 @@ layout(location = 8) in float dp_clip; #endif +layout(location = 9) in flat uint instance_index; + //defines to keep compatibility with vertex -#define world_matrix draw_call.transform +#define world_matrix instances.data[instance_index].transform #define projection_matrix scene_data.projection_matrix #if defined(ENABLE_SSS) && defined(ENABLE_TRANSMITTANCE) @@ -541,7 +553,7 @@ vec3 F0(float metallic, float specular, vec3 albedo) { return mix(vec3(dielectric), albedo, vec3(metallic)); } -void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float attenuation, vec3 shadow_attenuation, vec3 diffuse_color, float roughness, float metallic, float specular, float specular_blob_intensity, +void light_compute(vec3 N, vec3 L, vec3 V, vec3 light_color, float attenuation, vec3 f0, uint orms, float specular_amount, #ifdef LIGHT_BACKLIGHT_USED vec3 backlight, #endif @@ -553,7 +565,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float transmittance_z, #endif #ifdef LIGHT_RIM_USED - float rim, float rim_tint, + float rim, float rim_tint, vec3 rim_color, #endif #ifdef LIGHT_CLEARCOAT_USED float clearcoat, float clearcoat_gloss, @@ -561,6 +573,9 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #ifdef LIGHT_ANISOTROPY_USED vec3 B, vec3 T, float anisotropy, #endif +#ifdef USE_SOFT_SHADOWS + float A, +#endif #ifdef USE_SHADOW_TO_OPACITY inout float alpha, #endif @@ -570,7 +585,6 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte // light is written by the light shader vec3 normal = N; - vec3 albedo = diffuse_color; vec3 light = L; vec3 view = V; @@ -581,7 +595,12 @@ LIGHT_SHADER_CODE /* clang-format on */ #else + +#ifdef USE_SOFT_SHADOWS float NdotL = min(A + dot(N, L), 1.0); +#else + float NdotL = dot(N, L); +#endif float cNdotL = max(NdotL, 0.0); // clamped NdotL float NdotV = dot(N, V); float cNdotV = max(NdotV, 0.0); @@ -591,14 +610,25 @@ LIGHT_SHADER_CODE #endif #if defined(SPECULAR_BLINN) || defined(SPECULAR_SCHLICK_GGX) || defined(LIGHT_CLEARCOAT_USED) +#ifdef USE_SOFT_SHADOWS float cNdotH = clamp(A + dot(N, H), 0.0, 1.0); +#else + float cNdotH = clamp(dot(N, H), 0.0, 1.0); +#endif #endif #if defined(DIFFUSE_BURLEY) || defined(SPECULAR_SCHLICK_GGX) || defined(LIGHT_CLEARCOAT_USED) +#ifdef USE_SOFT_SHADOWS float cLdotH = clamp(A + dot(L, H), 0.0, 1.0); +#else + float cLdotH = clamp(dot(L, H), 0.0, 1.0); +#endif #endif + float metallic = unpackUnorm4x8(orms).z; if (metallic < 1.0) { + float roughness = unpackUnorm4x8(orms).y; + #if defined(DIFFUSE_OREN_NAYAR) vec3 diffuse_brdf_NL; #else @@ -608,23 +638,6 @@ LIGHT_SHADER_CODE #if defined(DIFFUSE_LAMBERT_WRAP) // energy conserving lambert wrap shader diffuse_brdf_NL = max(0.0, (NdotL + roughness) / ((1.0 + roughness) * (1.0 + roughness))); - -#elif defined(DIFFUSE_OREN_NAYAR) - - { - // see http://mimosa-pudica.net/improved-oren-nayar.html - float LdotV = dot(L, V); - - float s = LdotV - NdotL * NdotV; - float t = mix(1.0, max(NdotL, NdotV), step(0.0, s)); - - float sigma2 = roughness * roughness; // TODO: this needs checking - vec3 A = 1.0 + sigma2 * (-0.5 / (sigma2 + 0.33) + 0.17 * diffuse_color / (sigma2 + 0.13)); - float B = 0.45 * sigma2 / (sigma2 + 0.09); - - diffuse_brdf_NL = cNdotL * (A + vec3(B) * s / t) * (1.0 / M_PI); - } - #elif defined(DIFFUSE_TOON) diffuse_brdf_NL = smoothstep(-roughness, max(roughness, 0.01), NdotL); @@ -652,15 +665,15 @@ LIGHT_SHADER_CODE diffuse_brdf_NL = cNdotL * (1.0 / M_PI); #endif - diffuse_light += light_color * diffuse_color * shadow_attenuation * diffuse_brdf_NL * attenuation; + diffuse_light += light_color * diffuse_brdf_NL * attenuation; #if defined(LIGHT_BACKLIGHT_USED) - diffuse_light += light_color * diffuse_color * (vec3(1.0 / M_PI) - diffuse_brdf_NL) * backlight * attenuation; + diffuse_light += light_color * (vec3(1.0 / M_PI) - diffuse_brdf_NL) * backlight * attenuation; #endif #if defined(LIGHT_RIM_USED) float rim_light = pow(max(0.0, 1.0 - cNdotV), max(0.0, (1.0 - roughness) * 16.0)); - diffuse_light += rim_light * rim * mix(vec3(1.0), diffuse_color, rim_tint) * light_color; + diffuse_light += rim_light * rim * mix(vec3(1.0), rim_color, rim_tint) * light_color; #endif #ifdef LIGHT_TRANSMITTANCE_USED @@ -678,7 +691,7 @@ LIGHT_SHADER_CODE vec3(0.358, 0.004, 0.0) * exp(dd / 1.99) + vec3(0.078, 0.0, 0.0) * exp(dd / 7.41); - diffuse_light += profile * transmittance_color.a * diffuse_color * light_color * clamp(transmittance_boost - NdotL, 0.0, 1.0) * (1.0 / M_PI) * attenuation; + diffuse_light += profile * transmittance_color.a * light_color * clamp(transmittance_boost - NdotL, 0.0, 1.0) * (1.0 / M_PI); } #else @@ -688,7 +701,7 @@ LIGHT_SHADER_CODE fade = pow(max(0.0, 1.0 - fade), transmittance_curve); fade *= clamp(transmittance_boost - NdotL, 0.0, 1.0); - diffuse_light += diffuse_color * transmittance_color.rgb * light_color * (1.0 / M_PI) * transmittance_color.a * fade * attenuation; + diffuse_light += transmittance_color.rgb * light_color * (1.0 / M_PI) * transmittance_color.a * fade; } #endif //SSS_MODE_SKIN @@ -696,6 +709,7 @@ LIGHT_SHADER_CODE #endif //LIGHT_TRANSMITTANCE_USED } + float roughness = unpackUnorm4x8(orms).y; if (roughness > 0.0) { // FIXME: roughness == 0 should not disable specular light entirely // D @@ -708,7 +722,7 @@ LIGHT_SHADER_CODE blinn *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); float intensity = blinn; - specular_light += light_color * shadow_attenuation * intensity * specular_blob_intensity * attenuation; + specular_light += light_color * intensity * attenuation * specular_amount; #elif defined(SPECULAR_PHONG) @@ -719,7 +733,7 @@ LIGHT_SHADER_CODE phong *= (shininess + 8.0) * (1.0 / (8.0 * M_PI)); float intensity = (phong) / max(4.0 * cNdotV * cNdotL, 0.75); - specular_light += light_color * shadow_attenuation * intensity * specular_blob_intensity * attenuation; + specular_light += light_color * intensity * attenuation * specular_amount; #elif defined(SPECULAR_TOON) @@ -728,7 +742,7 @@ LIGHT_SHADER_CODE float mid = 1.0 - roughness; mid *= mid; float intensity = smoothstep(mid - roughness * 0.5, mid + roughness * 0.5, RdotV) * mid; - diffuse_light += light_color * shadow_attenuation * intensity * specular_blob_intensity * attenuation; // write to diffuse_light, as in toon shading you generally want no reflection + diffuse_light += light_color * intensity * attenuation * specular_amount; // write to diffuse_light, as in toon shading you generally want no reflection #elif defined(SPECULAR_DISABLED) // none.. @@ -753,13 +767,12 @@ LIGHT_SHADER_CODE float G = G_GGX_2cos(cNdotL, alpha_ggx) * G_GGX_2cos(cNdotV, alpha_ggx); #endif // F - vec3 f0 = F0(metallic, specular, diffuse_color); float cLdotH5 = SchlickFresnel(cLdotH); vec3 F = mix(vec3(cLdotH5), vec3(1.0), f0); vec3 specular_brdf_NL = cNdotL * D * F * G; - specular_light += specular_brdf_NL * light_color * shadow_attenuation * specular_blob_intensity * attenuation; + specular_light += specular_brdf_NL * light_color * attenuation * specular_amount; #endif #if defined(LIGHT_CLEARCOAT_USED) @@ -773,12 +786,12 @@ LIGHT_SHADER_CODE float clearcoat_specular_brdf_NL = 0.25 * clearcoat * Gr * Fr * Dr * cNdotL; - specular_light += clearcoat_specular_brdf_NL * light_color * shadow_attenuation * specular_blob_intensity * attenuation; + specular_light += clearcoat_specular_brdf_NL * light_color * attenuation * specular_amount; #endif } #ifdef USE_SHADOW_TO_OPACITY - alpha = min(alpha, clamp(1.0 - length(shadow_attenuation * attenuation), 0.0, 1.0)); + alpha = min(alpha, clamp(1.0 - attenuation), 0.0, 1.0)); #endif #endif //defined(USE_LIGHT_SHADER_CODE) @@ -786,13 +799,11 @@ LIGHT_SHADER_CODE #ifndef USE_NO_SHADOWS -// Produces cheap white noise, optimized for window-space -// Comes from: https://www.shadertoy.com/view/4djSRW -// Copyright: Dave Hoskins, MIT License +// Interleaved Gradient Noise +// http://www.iryoku.com/next-generation-post-processing-in-call-of-duty-advanced-warfare float quick_hash(vec2 pos) { - vec3 p3 = fract(vec3(pos.xyx) * .1031); - p3 += dot(p3, p3.yzx + 33.33); - return fract((p3.x + p3.y) * p3.z); + const vec3 magic = vec3(0.06711056f, 0.00583715f, 52.9829189f); + return fract(magic.z * fract(dot(pos, magic.xy))); } float sample_directional_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec4 coord) { @@ -900,68 +911,30 @@ float get_omni_attenuation(float distance, float inv_range, float decay) { return nd * pow(max(distance, 0.0001), -decay); } -void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 albedo, float roughness, float metallic, float specular, float p_blob_intensity, -#ifdef LIGHT_BACKLIGHT_USED - vec3 backlight, -#endif -#ifdef LIGHT_TRANSMITTANCE_USED - vec4 transmittance_color, - float transmittance_depth, - float transmittance_curve, - float transmittance_boost, -#endif -#ifdef LIGHT_RIM_USED - float rim, float rim_tint, -#endif -#ifdef LIGHT_CLEARCOAT_USED - float clearcoat, float clearcoat_gloss, -#endif -#ifdef LIGHT_ANISOTROPY_USED - vec3 binormal, vec3 tangent, float anisotropy, -#endif -#ifdef USE_SHADOW_TO_OPACITY - inout float alpha, -#endif - inout vec3 diffuse_light, inout vec3 specular_light) { - vec3 light_rel_vec = lights.data[idx].position - vertex; - float light_length = length(light_rel_vec); - vec2 attenuation_energy = unpackHalf2x16(lights.data[idx].attenuation_energy); - float omni_attenuation = get_omni_attenuation(light_length, lights.data[idx].inv_radius, attenuation_energy.x); - float light_attenuation = omni_attenuation; - vec3 shadow_attenuation = vec3(1.0); - vec4 color_specular = unpackUnorm4x8(lights.data[idx].color_specular); - color_specular.rgb *= attenuation_energy.y; - float size_A = 0.0; - - if (lights.data[idx].size > 0.0) { - float t = lights.data[idx].size / max(0.001, light_length); - size_A = max(0.0, 1.0 - 1 / sqrt(1 + t * t)); - } - -#ifdef LIGHT_TRANSMITTANCE_USED - float transmittance_z = transmittance_depth; //no transmittance by default -#endif - +float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { #ifndef USE_NO_SHADOWS - vec4 shadow_color_enabled = unpackUnorm4x8(lights.data[idx].shadow_color_enabled); - if (shadow_color_enabled.w > 0.5) { + if (omni_lights.data[idx].shadow_enabled) { // there is a shadowmap + vec3 light_rel_vec = omni_lights.data[idx].position - vertex; + float light_length = length(light_rel_vec); + vec4 v = vec4(vertex, 1.0); - vec4 splane = (lights.data[idx].shadow_matrix * v); + vec4 splane = (omni_lights.data[idx].shadow_matrix * v); float shadow_len = length(splane.xyz); //need to remember shadow len from here { - vec3 nofs = normal_interp * lights.data[idx].shadow_normal_bias / lights.data[idx].inv_radius; + vec3 nofs = normal_interp * omni_lights.data[idx].shadow_normal_bias / omni_lights.data[idx].inv_radius; nofs *= (1.0 - max(0.0, dot(normalize(light_rel_vec), normalize(normal_interp)))); v.xyz += nofs; - splane = (lights.data[idx].shadow_matrix * v); + splane = (omni_lights.data[idx].shadow_matrix * v); } float shadow; - if (lights.data[idx].soft_shadow_size > 0.0) { +#ifdef USE_SOFT_SHADOWS + if (omni_lights.data[idx].soft_shadow_size > 0.0) { //soft shadow //find blocker @@ -981,10 +954,10 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v vec3 v0 = abs(normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); vec3 tangent = normalize(cross(v0, normal)); vec3 bitangent = normalize(cross(tangent, normal)); - float z_norm = shadow_len * lights.data[idx].inv_radius; + float z_norm = shadow_len * omni_lights.data[idx].inv_radius; - tangent *= lights.data[idx].soft_shadow_size * lights.data[idx].soft_shadow_scale; - bitangent *= lights.data[idx].soft_shadow_size * lights.data[idx].soft_shadow_scale; + tangent *= omni_lights.data[idx].soft_shadow_size * omni_lights.data[idx].soft_shadow_scale; + bitangent *= omni_lights.data[idx].soft_shadow_size * omni_lights.data[idx].soft_shadow_scale; for (uint i = 0; i < scene_data.penumbra_shadow_samples; i++) { vec2 disk = disk_rotation * scene_data.penumbra_shadow_kernel[i].xy; @@ -992,7 +965,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v vec3 pos = splane.xyz + tangent * disk.x + bitangent * disk.y; pos = normalize(pos); - vec4 uv_rect = lights.data[idx].atlas_rect; + vec4 uv_rect = omni_lights.data[idx].atlas_rect; if (pos.z >= 0.0) { pos.z += 1.0; @@ -1020,7 +993,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v tangent *= penumbra; bitangent *= penumbra; - z_norm -= lights.data[idx].inv_radius * lights.data[idx].shadow_bias; + z_norm -= omni_lights.data[idx].inv_radius * omni_lights.data[idx].shadow_bias; shadow = 0.0; for (uint i = 0; i < scene_data.penumbra_shadow_samples; i++) { @@ -1028,7 +1001,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v vec3 pos = splane.xyz + tangent * disk.x + bitangent * disk.y; pos = normalize(pos); - vec4 uv_rect = lights.data[idx].atlas_rect; + vec4 uv_rect = omni_lights.data[idx].atlas_rect; if (pos.z >= 0.0) { pos.z += 1.0; @@ -1051,8 +1024,9 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v shadow = 1.0; } } else { +#endif splane.xyz = normalize(splane.xyz); - vec4 clamp_rect = lights.data[idx].atlas_rect; + vec4 clamp_rect = omni_lights.data[idx].atlas_rect; if (splane.z >= 0.0) { splane.z += 1.0; @@ -1066,101 +1040,149 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v splane.xy /= splane.z; splane.xy = splane.xy * 0.5 + 0.5; - splane.z = (shadow_len - lights.data[idx].shadow_bias) * lights.data[idx].inv_radius; + splane.z = (shadow_len - omni_lights.data[idx].shadow_bias) * omni_lights.data[idx].inv_radius; splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; splane.w = 1.0; //needed? i think it should be 1 already - shadow = sample_pcf_shadow(shadow_atlas, lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, splane); + shadow = sample_pcf_shadow(shadow_atlas, omni_lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, splane); +#ifdef USE_SOFT_SHADOWS } +#endif + + return shadow; + } +#endif + return 1.0; +} + +void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, +#ifdef LIGHT_BACKLIGHT_USED + vec3 backlight, +#endif #ifdef LIGHT_TRANSMITTANCE_USED - { - vec4 clamp_rect = lights.data[idx].atlas_rect; + vec4 transmittance_color, + float transmittance_depth, + float transmittance_curve, + float transmittance_boost, +#endif +#ifdef LIGHT_RIM_USED + float rim, float rim_tint, vec3 rim_color, +#endif +#ifdef LIGHT_CLEARCOAT_USED + float clearcoat, float clearcoat_gloss, +#endif +#ifdef LIGHT_ANISOTROPY_USED + vec3 binormal, vec3 tangent, float anisotropy, +#endif +#ifdef USE_SHADOW_TO_OPACITY + inout float alpha, +#endif + inout vec3 diffuse_light, inout vec3 specular_light) { + vec3 light_rel_vec = omni_lights.data[idx].position - vertex; + float light_length = length(light_rel_vec); + float omni_attenuation = get_omni_attenuation(light_length, omni_lights.data[idx].inv_radius, omni_lights.data[idx].attenuation); + float light_attenuation = omni_attenuation; + vec3 color = omni_lights.data[idx].color; - //redo shadowmapping, but shrink the model a bit to avoid arctifacts - splane = (lights.data[idx].shadow_matrix * vec4(vertex - normalize(normal_interp) * lights.data[idx].transmittance_bias, 1.0)); +#ifdef USE_SOFT_SHADOWS + float size_A = 0.0; - shadow_len = length(splane.xyz); - splane = normalize(splane.xyz); + if (omni_lights.data[idx].size > 0.0) { + float t = omni_lights.data[idx].size / max(0.001, light_length); + size_A = max(0.0, 1.0 - 1 / sqrt(1 + t * t)); + } +#endif - if (splane.z >= 0.0) { - splane.z += 1.0; +#ifdef LIGHT_TRANSMITTANCE_USED + float transmittance_z = transmittance_depth; //no transmittance by default + transmittance_color.a *= light_attenuation; + { + vec4 clamp_rect = omni_lights.data[idx].atlas_rect; - } else { - splane.z = 1.0 - splane.z; - } + //redo shadowmapping, but shrink the model a bit to avoid arctifacts + vec4 splane = (omni_lights.data[idx].shadow_matrix * vec4(vertex - normalize(normal_interp) * omni_lights.data[idx].transmittance_bias, 1.0)); - splane.xy /= splane.z; - splane.xy = splane.xy * 0.5 + 0.5; - splane.z = shadow_len * lights.data[idx].inv_radius; - splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; - splane.w = 1.0; //needed? i think it should be 1 already + shadow_len = length(splane.xyz); + splane = normalize(splane.xyz); + + if (splane.z >= 0.0) { + splane.z += 1.0; - float shadow_z = textureLod(sampler2D(shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), splane.xy, 0.0).r; - transmittance_z = (splane.z - shadow_z) / lights.data[idx].inv_radius; + } else { + splane.z = 1.0 - splane.z; } -#endif - vec3 no_shadow = vec3(1.0); + splane.xy /= splane.z; + splane.xy = splane.xy * 0.5 + 0.5; + splane.z = shadow_len * omni_lights.data[idx].inv_radius; + splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; + splane.w = 1.0; //needed? i think it should be 1 already - if (lights.data[idx].projector_rect != vec4(0.0)) { - vec3 local_v = (lights.data[idx].shadow_matrix * vec4(vertex, 1.0)).xyz; - local_v = normalize(local_v); + float shadow_z = textureLod(sampler2D(shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), splane.xy, 0.0).r; + transmittance_z = (splane.z - shadow_z) / omni_lights.data[idx].inv_radius; + } +#endif - vec4 atlas_rect = lights.data[idx].projector_rect; +#if 0 - if (local_v.z >= 0.0) { - local_v.z += 1.0; - atlas_rect.y += atlas_rect.w; + if (omni_lights.data[idx].projector_rect != vec4(0.0)) { + vec3 local_v = (omni_lights.data[idx].shadow_matrix * vec4(vertex, 1.0)).xyz; + local_v = normalize(local_v); - } else { - local_v.z = 1.0 - local_v.z; - } + vec4 atlas_rect = omni_lights.data[idx].projector_rect; - local_v.xy /= local_v.z; - local_v.xy = local_v.xy * 0.5 + 0.5; - vec2 proj_uv = local_v.xy * atlas_rect.zw; + if (local_v.z >= 0.0) { + local_v.z += 1.0; + atlas_rect.y += atlas_rect.w; - vec2 proj_uv_ddx; - vec2 proj_uv_ddy; - { - vec3 local_v_ddx = (lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddx, 1.0)).xyz; - local_v_ddx = normalize(local_v_ddx); + } else { + local_v.z = 1.0 - local_v.z; + } - if (local_v_ddx.z >= 0.0) { - local_v_ddx.z += 1.0; - } else { - local_v_ddx.z = 1.0 - local_v_ddx.z; - } + local_v.xy /= local_v.z; + local_v.xy = local_v.xy * 0.5 + 0.5; + vec2 proj_uv = local_v.xy * atlas_rect.zw; - local_v_ddx.xy /= local_v_ddx.z; - local_v_ddx.xy = local_v_ddx.xy * 0.5 + 0.5; + vec2 proj_uv_ddx; + vec2 proj_uv_ddy; + { + vec3 local_v_ddx = (omni_lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddx, 1.0)).xyz; + local_v_ddx = normalize(local_v_ddx); - proj_uv_ddx = local_v_ddx.xy * atlas_rect.zw - proj_uv; + if (local_v_ddx.z >= 0.0) { + local_v_ddx.z += 1.0; + } else { + local_v_ddx.z = 1.0 - local_v_ddx.z; + } - vec3 local_v_ddy = (lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddy, 1.0)).xyz; - local_v_ddy = normalize(local_v_ddy); + local_v_ddx.xy /= local_v_ddx.z; + local_v_ddx.xy = local_v_ddx.xy * 0.5 + 0.5; - if (local_v_ddy.z >= 0.0) { - local_v_ddy.z += 1.0; - } else { - local_v_ddy.z = 1.0 - local_v_ddy.z; - } + proj_uv_ddx = local_v_ddx.xy * atlas_rect.zw - proj_uv; - local_v_ddy.xy /= local_v_ddy.z; - local_v_ddy.xy = local_v_ddy.xy * 0.5 + 0.5; + vec3 local_v_ddy = (omni_lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddy, 1.0)).xyz; + local_v_ddy = normalize(local_v_ddy); - proj_uv_ddy = local_v_ddy.xy * atlas_rect.zw - proj_uv; + if (local_v_ddy.z >= 0.0) { + local_v_ddy.z += 1.0; + } else { + local_v_ddy.z = 1.0 - local_v_ddy.z; } - vec4 proj = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), proj_uv + atlas_rect.xy, proj_uv_ddx, proj_uv_ddy); - no_shadow = mix(no_shadow, proj.rgb, proj.a); + local_v_ddy.xy /= local_v_ddy.z; + local_v_ddy.xy = local_v_ddy.xy * 0.5 + 0.5; + + proj_uv_ddy = local_v_ddy.xy * atlas_rect.zw - proj_uv; } - shadow_attenuation = mix(shadow_color_enabled.rgb, no_shadow, shadow); + vec4 proj = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), proj_uv + atlas_rect.xy, proj_uv_ddx, proj_uv_ddy); + no_shadow = mix(no_shadow, proj.rgb, proj.a); } -#endif //USE_NO_SHADOWS +#endif + + light_attenuation *= shadow; - light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color_specular.rgb, light_attenuation, shadow_attenuation, albedo, roughness, metallic, specular, color_specular.a * p_blob_intensity, + light_compute(normal, normalize(light_rel_vec), eye_vec, color, light_attenuation, f0, orms, omni_lights.data[idx].specular_amount, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1172,7 +1194,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim * omni_attenuation, rim_tint, + rim * omni_attenuation, rim_tint, rim_color, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1180,6 +1202,9 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED binormal, tangent, anisotropy, #endif +#ifdef USE_SOFT_SHADOWS + size_A, +#endif #ifdef USE_SHADOW_TO_OPACITY alpha, #endif @@ -1187,88 +1212,39 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v specular_light); } -void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 albedo, float roughness, float metallic, float specular, float p_blob_intensity, -#ifdef LIGHT_BACKLIGHT_USED - vec3 backlight, -#endif -#ifdef LIGHT_TRANSMITTANCE_USED - vec4 transmittance_color, - float transmittance_depth, - float transmittance_curve, - float transmittance_boost, -#endif -#ifdef LIGHT_RIM_USED - float rim, float rim_tint, -#endif -#ifdef LIGHT_CLEARCOAT_USED - float clearcoat, float clearcoat_gloss, -#endif -#ifdef LIGHT_ANISOTROPY_USED - vec3 binormal, vec3 tangent, float anisotropy, -#endif -#ifdef USE_SHADOW_TO_OPACITY - inout float alpha, -#endif - inout vec3 diffuse_light, - inout vec3 specular_light) { - vec3 light_rel_vec = lights.data[idx].position - vertex; - float light_length = length(light_rel_vec); - vec2 attenuation_energy = unpackHalf2x16(lights.data[idx].attenuation_energy); - float spot_attenuation = get_omni_attenuation(light_length, lights.data[idx].inv_radius, attenuation_energy.x); - vec3 spot_dir = lights.data[idx].direction; - vec2 spot_att_angle = unpackHalf2x16(lights.data[idx].cone_attenuation_angle); - float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_att_angle.y); - float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_att_angle.y)); - spot_attenuation *= 1.0 - pow(spot_rim, spot_att_angle.x); - float light_attenuation = spot_attenuation; - vec3 shadow_attenuation = vec3(1.0); - vec4 color_specular = unpackUnorm4x8(lights.data[idx].color_specular); - color_specular.rgb *= attenuation_energy.y; - - float size_A = 0.0; - - if (lights.data[idx].size > 0.0) { - float t = lights.data[idx].size / max(0.001, light_length); - size_A = max(0.0, 1.0 - 1 / sqrt(1 + t * t)); - } -/* - if (lights.data[idx].atlas_rect!=vec4(0.0)) { - //use projector texture - } - */ -#ifdef LIGHT_TRANSMITTANCE_USED - float transmittance_z = transmittance_depth; -#endif - +float light_process_spot_shadow(uint idx, vec3 vertex, vec3 normal) { #ifndef USE_NO_SHADOWS - vec4 shadow_color_enabled = unpackUnorm4x8(lights.data[idx].shadow_color_enabled); - if (shadow_color_enabled.w > 0.5) { + if (spot_lights.data[idx].shadow_enabled) { + vec3 light_rel_vec = spot_lights.data[idx].position - vertex; + float light_length = length(light_rel_vec); + vec3 spot_dir = spot_lights.data[idx].direction; //there is a shadowmap vec4 v = vec4(vertex, 1.0); - v.xyz -= spot_dir * lights.data[idx].shadow_bias; + v.xyz -= spot_dir * spot_lights.data[idx].shadow_bias; - float z_norm = dot(spot_dir, -light_rel_vec) * lights.data[idx].inv_radius; + float z_norm = dot(spot_dir, -light_rel_vec) * spot_lights.data[idx].inv_radius; float depth_bias_scale = 1.0 / (max(0.0001, z_norm)); //the closer to the light origin, the more you have to offset to reach 1px in the map - vec3 normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(spot_dir, -normalize(normal_interp)))) * lights.data[idx].shadow_normal_bias * depth_bias_scale; + vec3 normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(spot_dir, -normalize(normal_interp)))) * spot_lights.data[idx].shadow_normal_bias * depth_bias_scale; normal_bias -= spot_dir * dot(spot_dir, normal_bias); //only XY, no Z v.xyz += normal_bias; //adjust with bias - z_norm = dot(spot_dir, v.xyz - lights.data[idx].position) * lights.data[idx].inv_radius; + z_norm = dot(spot_dir, v.xyz - spot_lights.data[idx].position) * spot_lights.data[idx].inv_radius; float shadow; - vec4 splane = (lights.data[idx].shadow_matrix * v); + vec4 splane = (spot_lights.data[idx].shadow_matrix * v); splane /= splane.w; - if (lights.data[idx].soft_shadow_size > 0.0) { +#ifdef USE_SOFT_SHADOWS + if (spot_lights.data[idx].soft_shadow_size > 0.0) { //soft shadow //find blocker - vec2 shadow_uv = splane.xy * lights.data[idx].atlas_rect.zw + lights.data[idx].atlas_rect.xy; + vec2 shadow_uv = splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy; float blocker_count = 0.0; float blocker_average = 0.0; @@ -1281,11 +1257,11 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v disk_rotation = mat2(vec2(cr, -sr), vec2(sr, cr)); } - float uv_size = lights.data[idx].soft_shadow_size * z_norm * lights.data[idx].soft_shadow_scale; - vec2 clamp_max = lights.data[idx].atlas_rect.xy + lights.data[idx].atlas_rect.zw; + float uv_size = spot_lights.data[idx].soft_shadow_size * z_norm * spot_lights.data[idx].soft_shadow_scale; + vec2 clamp_max = spot_lights.data[idx].atlas_rect.xy + spot_lights.data[idx].atlas_rect.zw; for (uint i = 0; i < scene_data.penumbra_shadow_samples; i++) { vec2 suv = shadow_uv + (disk_rotation * scene_data.penumbra_shadow_kernel[i].xy) * uv_size; - suv = clamp(suv, lights.data[idx].atlas_rect.xy, clamp_max); + suv = clamp(suv, spot_lights.data[idx].atlas_rect.xy, clamp_max); float d = textureLod(sampler2D(shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), suv, 0.0).r; if (d < z_norm) { blocker_average += d; @@ -1302,7 +1278,7 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v shadow = 0.0; for (uint i = 0; i < scene_data.penumbra_shadow_samples; i++) { vec2 suv = shadow_uv + (disk_rotation * scene_data.penumbra_shadow_kernel[i].xy) * uv_size; - suv = clamp(suv, lights.data[idx].atlas_rect.xy, clamp_max); + suv = clamp(suv, spot_lights.data[idx].atlas_rect.xy, clamp_max); shadow += textureProj(sampler2DShadow(shadow_atlas, shadow_sampler), vec4(suv, z_norm, 1.0)); } @@ -1314,54 +1290,93 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v } } else { +#endif //hard shadow - vec4 shadow_uv = vec4(splane.xy * lights.data[idx].atlas_rect.zw + lights.data[idx].atlas_rect.xy, z_norm, 1.0); + vec4 shadow_uv = vec4(splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy, splane.z, 1.0); - shadow = sample_pcf_shadow(shadow_atlas, lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, shadow_uv); + shadow = sample_pcf_shadow(shadow_atlas, spot_lights.data[idx].soft_shadow_scale * scene_data.shadow_atlas_pixel_size, shadow_uv); +#ifdef USE_SOFT_SHADOWS } +#endif - vec3 no_shadow = vec3(1.0); + return shadow; + } - if (lights.data[idx].projector_rect != vec4(0.0)) { - splane = (lights.data[idx].shadow_matrix * vec4(vertex, 1.0)); - splane /= splane.w; +#endif //USE_NO_SHADOWS - vec2 proj_uv = splane.xy * lights.data[idx].projector_rect.zw; + return 1.0; +} - //ensure we have proper mipmaps - vec4 splane_ddx = (lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddx, 1.0)); - splane_ddx /= splane_ddx.w; - vec2 proj_uv_ddx = splane_ddx.xy * lights.data[idx].projector_rect.zw - proj_uv; +void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, +#ifdef LIGHT_BACKLIGHT_USED + vec3 backlight, +#endif +#ifdef LIGHT_TRANSMITTANCE_USED + vec4 transmittance_color, + float transmittance_depth, + float transmittance_curve, + float transmittance_boost, +#endif +#ifdef LIGHT_RIM_USED + float rim, float rim_tint, vec3 rim_color, +#endif +#ifdef LIGHT_CLEARCOAT_USED + float clearcoat, float clearcoat_gloss, +#endif +#ifdef LIGHT_ANISOTROPY_USED + vec3 binormal, vec3 tangent, float anisotropy, +#endif +#ifdef USE_SHADOW_TO_OPACITY + inout float alpha, +#endif + inout vec3 diffuse_light, + inout vec3 specular_light) { + vec3 light_rel_vec = spot_lights.data[idx].position - vertex; + float light_length = length(light_rel_vec); + float spot_attenuation = get_omni_attenuation(light_length, spot_lights.data[idx].inv_radius, spot_lights.data[idx].attenuation); + vec3 spot_dir = spot_lights.data[idx].direction; + float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[idx].cone_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[idx].cone_angle)); + spot_attenuation *= 1.0 - pow(spot_rim, spot_lights.data[idx].cone_attenuation); + float light_attenuation = spot_attenuation; + vec3 color = spot_lights.data[idx].color; + float specular_amount = spot_lights.data[idx].specular_amount; - vec4 splane_ddy = (lights.data[idx].shadow_matrix * vec4(vertex + vertex_ddy, 1.0)); - splane_ddy /= splane_ddy.w; - vec2 proj_uv_ddy = splane_ddy.xy * lights.data[idx].projector_rect.zw - proj_uv; +#ifdef USE_SOFT_SHADOWS + float size_A = 0.0; - vec4 proj = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), proj_uv + lights.data[idx].projector_rect.xy, proj_uv_ddx, proj_uv_ddy); - no_shadow = mix(no_shadow, proj.rgb, proj.a); - } + if (spot_lights.data[idx].size > 0.0) { + float t = spot_lights.data[idx].size / max(0.001, light_length); + size_A = max(0.0, 1.0 - 1 / sqrt(1 + t * t)); + } +#endif - shadow_attenuation = mix(shadow_color_enabled.rgb, no_shadow, shadow); + /* + if (spot_lights.data[idx].atlas_rect!=vec4(0.0)) { + //use projector texture + } + */ #ifdef LIGHT_TRANSMITTANCE_USED - { - splane = (lights.data[idx].shadow_matrix * vec4(vertex - normalize(normal_interp) * lights.data[idx].transmittance_bias, 1.0)); - splane /= splane.w; - splane.xy = splane.xy * lights.data[idx].atlas_rect.zw + lights.data[idx].atlas_rect.xy; - - float shadow_z = textureLod(sampler2D(shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), splane.xy, 0.0).r; - //reconstruct depth - shadow_z /= lights.data[idx].inv_radius; - //distance to light plane - float z = dot(spot_dir, -light_rel_vec); - transmittance_z = z - shadow_z; - } -#endif //LIGHT_TRANSMITTANCE_USED + float transmittance_z = transmittance_depth; + transmittance_color.a *= light_attenuation; + { + splane = (spot_lights.data[idx].shadow_matrix * vec4(vertex - normalize(normal_interp) * spot_lights.data[idx].transmittance_bias, 1.0)); + splane /= splane.w; + splane.xy = splane.xy * spot_lights.data[idx].atlas_rect.zw + spot_lights.data[idx].atlas_rect.xy; + + float shadow_z = textureLod(sampler2D(shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), splane.xy, 0.0).r; + //reconstruct depth + shadow_z /= spot_lights.data[idx].inv_radius; + //distance to light plane + float z = dot(spot_dir, -light_rel_vec); + transmittance_z = z - shadow_z; } +#endif //LIGHT_TRANSMITTANCE_USED -#endif //USE_NO_SHADOWS + light_attenuation *= shadow; - light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color_specular.rgb, light_attenuation, shadow_attenuation, albedo, roughness, metallic, specular, color_specular.a * p_blob_intensity, + light_compute(normal, normalize(light_rel_vec), eye_vec, color, light_attenuation, f0, orms, spot_lights.data[idx].specular_amount, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1373,7 +1388,7 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim * spot_attenuation, rim_tint, + rim * spot_attenuation, rim_tint, rim_color, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1381,6 +1396,9 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED binormal, tangent, anisotropy, #endif +#ifdef USE_SOFT_SHADOW + size_A, +#endif #ifdef USE_SHADOW_TO_OPACITY alpha, #endif @@ -1404,11 +1422,11 @@ void reflection_process(uint ref_index, vec3 vertex, vec3 normal, float roughnes blend *= blend; blend = max(0.0, 1.0 - blend); - if (reflections.data[ref_index].params.x > 0.0) { // compute reflection + if (reflections.data[ref_index].intensity > 0.0) { // compute reflection vec3 local_ref_vec = (reflections.data[ref_index].local_matrix * vec4(ref_vec, 0.0)).xyz; - if (reflections.data[ref_index].params.w > 0.5) { //box project + if (reflections.data[ref_index].box_project) { //box project vec3 nrdir = normalize(local_ref_vec); vec3 rbmax = (box_extents - local_pos) / nrdir; @@ -1425,11 +1443,11 @@ void reflection_process(uint ref_index, vec3 vertex, vec3 normal, float roughnes reflection.rgb = textureLod(samplerCubeArray(reflection_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(local_ref_vec, reflections.data[ref_index].index), roughness * MAX_ROUGHNESS_LOD).rgb; - if (reflections.data[ref_index].params.z < 0.5) { + if (reflections.data[ref_index].exterior) { reflection.rgb = mix(specular_light, reflection.rgb, blend); } - reflection.rgb *= reflections.data[ref_index].params.x; + reflection.rgb *= reflections.data[ref_index].intensity; //intensity reflection.a = blend; reflection.rgb *= reflection.a; @@ -1448,7 +1466,7 @@ void reflection_process(uint ref_index, vec3 vertex, vec3 normal, float roughnes ambient_out.rgb = textureLod(samplerCubeArray(reflection_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), vec4(local_amb_vec, reflections.data[ref_index].index), MAX_ROUGHNESS_LOD).rgb; ambient_out.a = blend; - if (reflections.data[ref_index].params.z < 0.5) { //interior + if (reflections.data[ref_index].exterior) { ambient_out.rgb = mix(ambient_light, ambient_out.rgb, blend); } @@ -1459,7 +1477,7 @@ void reflection_process(uint ref_index, vec3 vertex, vec3 normal, float roughnes vec4 ambient_out; ambient_out.a = blend; ambient_out.rgb = reflections.data[ref_index].ambient; - if (reflections.data[ref_index].params.z < 0.5) { + if (reflections.data[ref_index].exterior) { ambient_out.rgb = mix(ambient_light, ambient_out.rgb, blend); } ambient_out.rgb *= ambient_out.a; @@ -1762,7 +1780,7 @@ vec4 fog_process(vec3 vertex) { } } - float fog_amount = 1.0 - exp(vertex.z * scene_data.fog_density); + float fog_amount = 1.0 - exp(min(0.0, vertex.z * scene_data.fog_density)); if (abs(scene_data.fog_height_density) > 0.001) { float y = (scene_data.camera_matrix * vec4(vertex, 1.0)).y; @@ -1777,7 +1795,43 @@ vec4 fog_process(vec3 vertex) { return vec4(fog_color, fog_amount); } +void cluster_get_item_range(uint p_offset, out uint item_min, out uint item_max, out uint item_from, out uint item_to) { + uint item_min_max = cluster_buffer.data[p_offset]; + item_min = item_min_max & 0xFFFF; + item_max = item_min_max >> 16; + ; + + item_from = item_min >> 5; + item_to = (item_max == 0) ? 0 : ((item_max - 1) >> 5) + 1; //side effect of how it is stored, as item_max 0 means no elements +} + +uint cluster_get_range_clip_mask(uint i, uint z_min, uint z_max) { + int local_min = clamp(int(z_min) - int(i) * 32, 0, 31); + int mask_width = min(int(z_max) - int(z_min), 32 - local_min); + return bitfieldInsert(uint(0), uint(0xFFFFFFFF), local_min, mask_width); +} + +float blur_shadow(float shadow) { + return shadow; +#if 0 + //disabling for now, will investigate later + float interp_shadow = shadow; + if (gl_HelperInvocation) { + interp_shadow = -4.0; // technically anything below -4 will do but just to make sure + } + + uvec2 fc2 = uvec2(gl_FragCoord.xy); + interp_shadow -= dFdx(interp_shadow) * (float(fc2.x & 1) - 0.5); + interp_shadow -= dFdy(interp_shadow) * (float(fc2.y & 1) - 0.5); + + if (interp_shadow >= 0.0) { + shadow = interp_shadow; + } + return shadow; #endif +} + +#endif //!MODE_RENDER DEPTH void main() { #ifdef MODE_DUAL_PARABOLOID @@ -1805,9 +1859,7 @@ void main() { float clearcoat_gloss = 0.0; float anisotropy = 0.0; vec2 anisotropy_flow = vec2(1.0, 0.0); -#if defined(CUSTOM_FOG_USED) - vec4 custom_fog = vec4(0.0); -#endif + vec4 fog = vec4(0.0); #if defined(CUSTOM_RADIANCE_USED) vec4 custom_radiance = vec4(0.0); #endif @@ -1815,10 +1867,8 @@ void main() { vec4 custom_irradiance = vec4(0.0); #endif -#if defined(AO_USED) float ao = 1.0; float ao_light_affect = 0.0; -#endif float alpha = 1.0; @@ -1956,77 +2006,147 @@ FRAGMENT_SHADER_CODE discard; } #endif + + /////////////////////// FOG ////////////////////// +#ifndef MODE_RENDER_DEPTH + +#ifndef CUSTOM_FOG_USED + // fog must be processed as early as possible and then packed. + // to maximize VGPR usage + // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. + + if (scene_data.fog_enabled) { + fog = fog_process(vertex); + } + +#ifndef LOW_END_MODE + if (scene_data.volumetric_fog_enabled) { + vec4 volumetric_fog = volumetric_fog_process(screen_uv, -vertex.z); + if (scene_data.fog_enabled) { + //must use the full blending equation here to blend fogs + vec4 res; + float sa = 1.0 - volumetric_fog.a; + res.a = fog.a * sa + volumetric_fog.a; + if (res.a == 0.0) { + res.rgb = vec3(0.0); + } else { + res.rgb = (fog.rgb * fog.a * sa + volumetric_fog.rgb * volumetric_fog.a) / res.a; + } + fog = res; + } else { + fog = volumetric_fog; + } + } +#endif //!LOW_END_MODE +#endif //!CUSTOM_FOG_USED + + uint fog_rg = packHalf2x16(fog.rg); + uint fog_ba = packHalf2x16(fog.ba); + +#endif //!MODE_RENDER_DEPTH + /////////////////////// DECALS //////////////////////////////// #ifndef MODE_RENDER_DEPTH - uvec4 cluster_cell = texture(usampler3D(cluster_texture, material_samplers[SAMPLER_NEAREST_CLAMP]), vec3(screen_uv, (abs(vertex.z) - scene_data.z_near) / (scene_data.z_far - scene_data.z_near))); + uvec2 cluster_pos = uvec2(gl_FragCoord.xy) >> scene_data.cluster_shift; + uint cluster_offset = (scene_data.cluster_width * cluster_pos.y + cluster_pos.x) * (scene_data.max_cluster_element_count_div_32 + 32); + + uint cluster_z = uint(clamp((-vertex.z / scene_data.z_far) * 32.0, 0.0, 31.0)); + //used for interpolating anything cluster related vec3 vertex_ddx = dFdx(vertex); vec3 vertex_ddy = dFdy(vertex); { // process decals - uint decal_count = cluster_cell.w >> CLUSTER_COUNTER_SHIFT; - uint decal_pointer = cluster_cell.w & CLUSTER_POINTER_MASK; + uint cluster_decal_offset = cluster_offset + scene_data.cluster_type_size * 2; - //do outside for performance and avoiding arctifacts + uint item_min; + uint item_max; + uint item_from; + uint item_to; - for (uint i = 0; i < decal_count; i++) { - uint decal_index = cluster_data.indices[decal_pointer + i]; - if (!bool(decals.data[decal_index].mask & draw_call.layer_mask)) { - continue; //not masked - } + cluster_get_item_range(cluster_decal_offset + scene_data.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); - vec3 uv_local = (decals.data[decal_index].xform * vec4(vertex, 1.0)).xyz; - if (any(lessThan(uv_local, vec3(0.0, -1.0, 0.0))) || any(greaterThan(uv_local, vec3(1.0)))) { - continue; //out of decal - } +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif - //we need ddx/ddy for mipmaps, so simulate them - vec2 ddx = (decals.data[decal_index].xform * vec4(vertex_ddx, 0.0)).xz; - vec2 ddy = (decals.data[decal_index].xform * vec4(vertex_ddy, 0.0)).xz; + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_decal_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif - float fade = pow(1.0 - (uv_local.y > 0.0 ? uv_local.y : -uv_local.y), uv_local.y > 0.0 ? decals.data[decal_index].upper_fade : decals.data[decal_index].lower_fade); + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + uint decal_index = 32 * i + bit; - if (decals.data[decal_index].normal_fade > 0.0) { - fade *= smoothstep(decals.data[decal_index].normal_fade, 1.0, dot(normal_interp, decals.data[decal_index].normal) * 0.5 + 0.5); - } + if (!bool(decals.data[decal_index].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } - if (decals.data[decal_index].albedo_rect != vec4(0.0)) { - //has albedo - vec4 decal_albedo = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].albedo_rect.zw + decals.data[decal_index].albedo_rect.xy, ddx * decals.data[decal_index].albedo_rect.zw, ddy * decals.data[decal_index].albedo_rect.zw); - decal_albedo *= decals.data[decal_index].modulate; - decal_albedo.a *= fade; - albedo = mix(albedo, decal_albedo.rgb, decal_albedo.a * decals.data[decal_index].albedo_mix); - - if (decals.data[decal_index].normal_rect != vec4(0.0)) { - vec3 decal_normal = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].normal_rect.zw + decals.data[decal_index].normal_rect.xy, ddx * decals.data[decal_index].normal_rect.zw, ddy * decals.data[decal_index].normal_rect.zw).xyz; - decal_normal.xy = decal_normal.xy * vec2(2.0, -2.0) - vec2(1.0, -1.0); //users prefer flipped y normal maps in most authoring software - decal_normal.z = sqrt(max(0.0, 1.0 - dot(decal_normal.xy, decal_normal.xy))); - //convert to view space, use xzy because y is up - decal_normal = (decals.data[decal_index].normal_xform * decal_normal.xzy).xyz; - - normal = normalize(mix(normal, decal_normal, decal_albedo.a)); + vec3 uv_local = (decals.data[decal_index].xform * vec4(vertex, 1.0)).xyz; + if (any(lessThan(uv_local, vec3(0.0, -1.0, 0.0))) || any(greaterThan(uv_local, vec3(1.0)))) { + continue; //out of decal } - if (decals.data[decal_index].orm_rect != vec4(0.0)) { - vec3 decal_orm = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].orm_rect.zw + decals.data[decal_index].orm_rect.xy, ddx * decals.data[decal_index].orm_rect.zw, ddy * decals.data[decal_index].orm_rect.zw).xyz; -#if defined(AO_USED) - ao = mix(ao, decal_orm.r, decal_albedo.a); -#endif - roughness = mix(roughness, decal_orm.g, decal_albedo.a); - metallic = mix(metallic, decal_orm.b, decal_albedo.a); + //we need ddx/ddy for mipmaps, so simulate them + vec2 ddx = (decals.data[decal_index].xform * vec4(vertex_ddx, 0.0)).xz; + vec2 ddy = (decals.data[decal_index].xform * vec4(vertex_ddy, 0.0)).xz; + + float fade = pow(1.0 - (uv_local.y > 0.0 ? uv_local.y : -uv_local.y), uv_local.y > 0.0 ? decals.data[decal_index].upper_fade : decals.data[decal_index].lower_fade); + + if (decals.data[decal_index].normal_fade > 0.0) { + fade *= smoothstep(decals.data[decal_index].normal_fade, 1.0, dot(normal_interp, decals.data[decal_index].normal) * 0.5 + 0.5); + } + + if (decals.data[decal_index].albedo_rect != vec4(0.0)) { + //has albedo + vec4 decal_albedo = textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].albedo_rect.zw + decals.data[decal_index].albedo_rect.xy, ddx * decals.data[decal_index].albedo_rect.zw, ddy * decals.data[decal_index].albedo_rect.zw); + decal_albedo *= decals.data[decal_index].modulate; + decal_albedo.a *= fade; + albedo = mix(albedo, decal_albedo.rgb, decal_albedo.a * decals.data[decal_index].albedo_mix); + + if (decals.data[decal_index].normal_rect != vec4(0.0)) { + vec3 decal_normal = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].normal_rect.zw + decals.data[decal_index].normal_rect.xy, ddx * decals.data[decal_index].normal_rect.zw, ddy * decals.data[decal_index].normal_rect.zw).xyz; + decal_normal.xy = decal_normal.xy * vec2(2.0, -2.0) - vec2(1.0, -1.0); //users prefer flipped y normal maps in most authoring software + decal_normal.z = sqrt(max(0.0, 1.0 - dot(decal_normal.xy, decal_normal.xy))); + //convert to view space, use xzy because y is up + decal_normal = (decals.data[decal_index].normal_xform * decal_normal.xzy).xyz; + + normal = normalize(mix(normal, decal_normal, decal_albedo.a)); + } + + if (decals.data[decal_index].orm_rect != vec4(0.0)) { + vec3 decal_orm = textureGrad(sampler2D(decal_atlas, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].orm_rect.zw + decals.data[decal_index].orm_rect.xy, ddx * decals.data[decal_index].orm_rect.zw, ddy * decals.data[decal_index].orm_rect.zw).xyz; + ao = mix(ao, decal_orm.r, decal_albedo.a); + roughness = mix(roughness, decal_orm.g, decal_albedo.a); + metallic = mix(metallic, decal_orm.b, decal_albedo.a); + } } - } - if (decals.data[decal_index].emission_rect != vec4(0.0)) { - //emission is additive, so its independent from albedo - emission += textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].emission_rect.zw + decals.data[decal_index].emission_rect.xy, ddx * decals.data[decal_index].emission_rect.zw, ddy * decals.data[decal_index].emission_rect.zw).xyz * decals.data[decal_index].emission_energy * fade; + if (decals.data[decal_index].emission_rect != vec4(0.0)) { + //emission is additive, so its independent from albedo + emission += textureGrad(sampler2D(decal_atlas_srgb, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), uv_local.xz * decals.data[decal_index].emission_rect.zw + decals.data[decal_index].emission_rect.xy, ddx * decals.data[decal_index].emission_rect.zw, ddy * decals.data[decal_index].emission_rect.zw).xyz * decals.data[decal_index].emission_energy * fade; + } } } } + //pack albedo until needed again, saves 2 VGPRs in the meantime + #endif //not render depth /////////////////////// LIGHTING ////////////////////////////// @@ -2094,19 +2214,14 @@ FRAGMENT_SHADER_CODE //radiance - float specular_blob_intensity = 1.0; - -#if defined(SPECULAR_TOON) - specular_blob_intensity *= specular * 2.0; -#endif - +/// GI /// #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) #ifdef USE_LIGHTMAP //lightmap - if (bool(draw_call.flags & INSTANCE_FLAGS_USE_LIGHTMAP_CAPTURE)) { //has lightmap capture - uint index = draw_call.gi_offset; + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_LIGHTMAP_CAPTURE)) { //has lightmap capture + uint index = instances.data[instance_index].gi_offset; vec3 wnormal = mat3(scene_data.camera_matrix) * normal; const float c1 = 0.429043; @@ -2125,12 +2240,12 @@ FRAGMENT_SHADER_CODE 2.0 * c2 * lightmap_captures.data[index].sh[1].rgb * wnormal.y + 2.0 * c2 * lightmap_captures.data[index].sh[2].rgb * wnormal.z); - } else if (bool(draw_call.flags & INSTANCE_FLAGS_USE_LIGHTMAP)) { // has actual lightmap - bool uses_sh = bool(draw_call.flags & INSTANCE_FLAGS_USE_SH_LIGHTMAP); - uint ofs = draw_call.gi_offset & 0xFFFF; + } else if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_LIGHTMAP)) { // has actual lightmap + bool uses_sh = bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_SH_LIGHTMAP); + uint ofs = instances.data[instance_index].gi_offset & 0xFFFF; vec3 uvw; - uvw.xy = uv2 * draw_call.lightmap_uv_scale.zw + draw_call.lightmap_uv_scale.xy; - uvw.z = float((draw_call.gi_offset >> 16) & 0xFFFF); + uvw.xy = uv2 * instances.data[instance_index].lightmap_uv_scale.zw + instances.data[instance_index].lightmap_uv_scale.xy; + uvw.z = float((instances.data[instance_index].gi_offset >> 16) & 0xFFFF); if (uses_sh) { uvw.z *= 4.0; //SH textures use 4 times more data @@ -2139,7 +2254,7 @@ FRAGMENT_SHADER_CODE vec3 lm_light_l1_0 = textureLod(sampler2DArray(lightmap_textures[ofs], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw + vec3(0.0, 0.0, 2.0), 0.0).rgb; vec3 lm_light_l1p1 = textureLod(sampler2DArray(lightmap_textures[ofs], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw + vec3(0.0, 0.0, 3.0), 0.0).rgb; - uint idx = draw_call.gi_offset >> 20; + uint idx = instances.data[instance_index].gi_offset >> 20; vec3 n = normalize(lightmaps.data[idx].normal_xform * normal); ambient_light += lm_light_l0 * 0.282095f; @@ -2159,7 +2274,7 @@ FRAGMENT_SHADER_CODE } #elif defined(USE_FORWARD_GI) - if (bool(draw_call.flags & INSTANCE_FLAGS_USE_SDFGI)) { //has lightmap capture + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_SDFGI)) { //has lightmap capture //make vertex orientation the world one, but still align to camera vec3 cam_pos = mat3(scene_data.camera_matrix) * vertex; @@ -2231,9 +2346,9 @@ FRAGMENT_SHADER_CODE } } - if (bool(draw_call.flags & INSTANCE_FLAGS_USE_GIPROBE)) { // process giprobes + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_GIPROBE)) { // process giprobes - uint index1 = draw_call.gi_offset & 0xFFFF; + uint index1 = instances.data[instance_index].gi_offset & 0xFFFF; vec3 ref_vec = normalize(reflect(normalize(vertex), normal)); //find arbitrary tangent and bitangent, then build a matrix vec3 v0 = abs(normal.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(0.0, 1.0, 0.0); @@ -2245,7 +2360,7 @@ FRAGMENT_SHADER_CODE vec4 spec_accum = vec4(0.0); gi_probe_compute(index1, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); - uint index2 = draw_call.gi_offset >> 16; + uint index2 = instances.data[instance_index].gi_offset >> 16; if (index2 != 0xFFFF) { gi_probe_compute(index2, vertex, normal, ref_vec, normal_mat, roughness * roughness, ambient_light, specular_light, spec_accum, amb_accum); @@ -2264,19 +2379,19 @@ FRAGMENT_SHADER_CODE } #elif !defined(LOW_END_MODE) - if (bool(draw_call.flags & INSTANCE_FLAGS_USE_GI_BUFFERS)) { //use GI buffers + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_GI_BUFFERS)) { //use GI buffers - ivec2 coord; + vec2 coord; if (scene_data.gi_upscale_for_msaa) { - ivec2 base_coord = ivec2(gl_FragCoord.xy); - ivec2 closest_coord = base_coord; - float closest_ang = dot(normal, texelFetch(sampler2D(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), base_coord, 0).xyz * 2.0 - 1.0); + vec2 base_coord = screen_uv; + vec2 closest_coord = base_coord; + float closest_ang = dot(normal, textureLod(sampler2D(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), base_coord, 0.0).xyz * 2.0 - 1.0); for (int i = 0; i < 4; i++) { - const ivec2 neighbours[4] = ivec2[](ivec2(-1, 0), ivec2(1, 0), ivec2(0, -1), ivec2(0, 1)); - ivec2 neighbour_coord = base_coord + neighbours[i]; - float neighbour_ang = dot(normal, texelFetch(sampler2D(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), neighbour_coord, 0).xyz * 2.0 - 1.0); + const vec2 neighbours[4] = vec2[](vec2(-1, 0), vec2(1, 0), vec2(0, -1), vec2(0, 1)); + vec2 neighbour_coord = base_coord + neighbours[i] * scene_data.screen_pixel_size; + float neighbour_ang = dot(normal, textureLod(sampler2D(normal_roughness_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), neighbour_coord, 0.0).xyz * 2.0 - 1.0); if (neighbour_ang > closest_ang) { closest_ang = neighbour_ang; closest_coord = neighbour_coord; @@ -2286,28 +2401,69 @@ FRAGMENT_SHADER_CODE coord = closest_coord; } else { - coord = ivec2(gl_FragCoord.xy); + coord = screen_uv; } - vec4 buffer_ambient = texelFetch(sampler2D(ambient_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), coord, 0); - vec4 buffer_reflection = texelFetch(sampler2D(reflection_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), coord, 0); + vec4 buffer_ambient = textureLod(sampler2D(ambient_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), coord, 0.0); + vec4 buffer_reflection = textureLod(sampler2D(reflection_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), coord, 0.0); ambient_light = mix(ambient_light, buffer_ambient.rgb, buffer_ambient.a); specular_light = mix(specular_light, buffer_reflection.rgb, buffer_reflection.a); } #endif +#ifndef LOW_END_MODE + if (scene_data.ssao_enabled) { + float ssao = texture(sampler2D(ao_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), screen_uv).r; + ao = min(ao, ssao); + ao_light_affect = mix(ao_light_affect, max(ao_light_affect, scene_data.ssao_light_affect), scene_data.ssao_ao_affect); + } +#endif //LOW_END_MODE + { // process reflections vec4 reflection_accum = vec4(0.0, 0.0, 0.0, 0.0); vec4 ambient_accum = vec4(0.0, 0.0, 0.0, 0.0); - uint reflection_probe_count = cluster_cell.z >> CLUSTER_COUNTER_SHIFT; - uint reflection_probe_pointer = cluster_cell.z & CLUSTER_POINTER_MASK; + uint cluster_reflection_offset = cluster_offset + scene_data.cluster_type_size * 3; - for (uint i = 0; i < reflection_probe_count; i++) { - uint ref_index = cluster_data.indices[reflection_probe_pointer + i]; - reflection_process(ref_index, vertex, normal, roughness, ambient_light, specular_light, ambient_accum, reflection_accum); + uint item_min; + uint item_max; + uint item_from; + uint item_to; + + cluster_get_item_range(cluster_reflection_offset + scene_data.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); + +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_reflection_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + uint reflection_index = 32 * i + bit; + + if (!bool(reflections.data[reflection_index].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } + + reflection_process(reflection_index, vertex, normal, roughness, ambient_light, specular_light, ambient_accum, reflection_accum); + } } if (reflection_accum.a > 0.0) { @@ -2321,6 +2477,16 @@ FRAGMENT_SHADER_CODE #endif } + //finalize ambient light here + ambient_light *= albedo.rgb; + ambient_light *= ao; + + // convert ao to direct light ao + ao = mix(1.0, ao, ao_light_affect); + + //this saves some VGPRs + vec3 f0 = F0(metallic, specular, albedo); + { #if defined(DIFFUSE_TOON) //simplify for toon, as @@ -2338,24 +2504,39 @@ FRAGMENT_SHADER_CODE float a004 = min(r.x * r.x, exp2(-9.28 * ndotv)) * r.x + r.y; vec2 env = vec2(-1.04, 1.04) * a004 + r.zw; - vec3 f0 = F0(metallic, specular, albedo); specular_light *= env.x * f0 + env.y; #endif } +#endif //GI !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) + +#if !defined(MODE_RENDER_DEPTH) + //this saves some VGPRs + uint orms = packUnorm4x8(vec4(ao, roughness, metallic, specular)); +#endif + +// LIGHTING +#if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) + { //directional light - for (uint i = 0; i < scene_data.directional_light_count; i++) { - if (!bool(directional_lights.data[i].mask & draw_call.layer_mask)) { - continue; //not masked + // Do shadow and lighting in two passes to reduce register pressure + uint shadow0 = 0; + uint shadow1 = 0; + + for (uint i = 0; i < 8; i++) { + if (i >= scene_data.directional_light_count) { + break; } - vec3 shadow_attenuation = vec3(1.0); + if (!bool(directional_lights.data[i].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } -#ifdef LIGHT_TRANSMITTANCE_USED - float transmittance_z = transmittance_depth; -#endif + float shadow = 1.0; +#ifdef USE_SOFT_SHADOWS + //version with soft shadows, more expensive if (directional_lights.data[i].shadow_enabled) { float depth_z = -vertex.z; @@ -2369,8 +2550,6 @@ FRAGMENT_SHADER_CODE normal_bias -= light_dir * dot(light_dir, normal_bias); \ m_var.xyz += normal_bias; - float shadow = 0.0; - if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { vec4 v = vec4(vertex, 1.0); @@ -2391,19 +2570,6 @@ FRAGMENT_SHADER_CODE shadow_color = directional_lights.data[i].shadow_color1.rgb; -#ifdef LIGHT_TRANSMITTANCE_USED - { - vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.x, 1.0); - vec4 trans_coord = directional_lights.data[i].shadow_matrix1 * trans_vertex; - trans_coord /= trans_coord.w; - - float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; - shadow_z *= directional_lights.data[i].shadow_z_range.x; - float z = trans_coord.z * directional_lights.data[i].shadow_z_range.x; - - transmittance_z = z - shadow_z; - } -#endif } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { vec4 v = vec4(vertex, 1.0); @@ -2423,19 +2589,6 @@ FRAGMENT_SHADER_CODE } shadow_color = directional_lights.data[i].shadow_color2.rgb; -#ifdef LIGHT_TRANSMITTANCE_USED - { - vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.y, 1.0); - vec4 trans_coord = directional_lights.data[i].shadow_matrix2 * trans_vertex; - trans_coord /= trans_coord.w; - - float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; - shadow_z *= directional_lights.data[i].shadow_z_range.y; - float z = trans_coord.z * directional_lights.data[i].shadow_z_range.y; - - transmittance_z = z - shadow_z; - } -#endif } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { vec4 v = vec4(vertex, 1.0); @@ -2455,19 +2608,6 @@ FRAGMENT_SHADER_CODE } shadow_color = directional_lights.data[i].shadow_color3.rgb; -#ifdef LIGHT_TRANSMITTANCE_USED - { - vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.z, 1.0); - vec4 trans_coord = directional_lights.data[i].shadow_matrix3 * trans_vertex; - trans_coord /= trans_coord.w; - - float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; - shadow_z *= directional_lights.data[i].shadow_z_range.z; - float z = trans_coord.z * directional_lights.data[i].shadow_z_range.z; - - transmittance_z = z - shadow_z; - } -#endif } else { vec4 v = vec4(vertex, 1.0); @@ -2488,20 +2628,6 @@ FRAGMENT_SHADER_CODE } shadow_color = directional_lights.data[i].shadow_color4.rgb; - -#ifdef LIGHT_TRANSMITTANCE_USED - { - vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.w, 1.0); - vec4 trans_coord = directional_lights.data[i].shadow_matrix4 * trans_vertex; - trans_coord /= trans_coord.w; - - float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; - shadow_z *= directional_lights.data[i].shadow_z_range.w; - float z = trans_coord.z * directional_lights.data[i].shadow_z_range.w; - - transmittance_z = z - shadow_z; - } -#endif } if (directional_lights.data[i].blend_splits) { @@ -2575,130 +2701,407 @@ FRAGMENT_SHADER_CODE shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance - shadow_attenuation = mix(shadow_color, vec3(1.0), shadow); +#undef BIAS_FUNC + } +#else + // Soft shadow disabled version + + if (directional_lights.data[i].shadow_enabled) { + float depth_z = -vertex.z; + + vec4 pssm_coord; + vec3 light_dir = directional_lights.data[i].direction; + vec3 base_normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))); + +#define BIAS_FUNC(m_var, m_idx) \ + m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ + vec3 normal_bias = base_normal_bias * directional_lights.data[i].shadow_normal_bias[m_idx]; \ + normal_bias -= light_dir * dot(light_dir, normal_bias); \ + m_var.xyz += normal_bias; + + if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { + vec4 v = vec4(vertex, 1.0); + + BIAS_FUNC(v, 0) + + pssm_coord = (directional_lights.data[i].shadow_matrix1 * v); +#ifdef LIGHT_TRANSMITTANCE_USED + { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.x, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix1 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.x; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.x; + + transmittance_z = z - shadow_z; + } +#endif + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { + vec4 v = vec4(vertex, 1.0); + + BIAS_FUNC(v, 1) + + pssm_coord = (directional_lights.data[i].shadow_matrix2 * v); +#ifdef LIGHT_TRANSMITTANCE_USED + { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.y, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix2 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.y; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.y; + + transmittance_z = z - shadow_z; + } +#endif + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { + vec4 v = vec4(vertex, 1.0); + + BIAS_FUNC(v, 2) + + pssm_coord = (directional_lights.data[i].shadow_matrix3 * v); +#ifdef LIGHT_TRANSMITTANCE_USED + { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.z, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix3 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.z; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.z; + + transmittance_z = z - shadow_z; + } +#endif + + } else { + vec4 v = vec4(vertex, 1.0); + + BIAS_FUNC(v, 3) + + pssm_coord = (directional_lights.data[i].shadow_matrix4 * v); +#ifdef LIGHT_TRANSMITTANCE_USED + { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.w, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix4 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.w; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.w; + + transmittance_z = z - shadow_z; + } +#endif + } + + pssm_coord /= pssm_coord.w; + + shadow = sample_directional_pcf_shadow(directional_shadow_atlas, scene_data.directional_shadow_pixel_size * directional_lights.data[i].soft_shadow_scale, pssm_coord); + + if (directional_lights.data[i].blend_splits) { + float pssm_blend; + + if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { + vec4 v = vec4(vertex, 1.0); + BIAS_FUNC(v, 1) + pssm_coord = (directional_lights.data[i].shadow_matrix2 * v); + pssm_blend = smoothstep(0.0, directional_lights.data[i].shadow_split_offsets.x, depth_z); + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { + vec4 v = vec4(vertex, 1.0); + BIAS_FUNC(v, 2) + pssm_coord = (directional_lights.data[i].shadow_matrix3 * v); + pssm_blend = smoothstep(directional_lights.data[i].shadow_split_offsets.x, directional_lights.data[i].shadow_split_offsets.y, depth_z); + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { + vec4 v = vec4(vertex, 1.0); + BIAS_FUNC(v, 3) + pssm_coord = (directional_lights.data[i].shadow_matrix4 * v); + pssm_blend = smoothstep(directional_lights.data[i].shadow_split_offsets.y, directional_lights.data[i].shadow_split_offsets.z, depth_z); + } else { + pssm_blend = 0.0; //if no blend, same coord will be used (divide by z will result in same value, and already cached) + } + + pssm_coord /= pssm_coord.w; + + float shadow2 = sample_directional_pcf_shadow(directional_shadow_atlas, scene_data.directional_shadow_pixel_size * directional_lights.data[i].soft_shadow_scale, pssm_coord); + shadow = mix(shadow, shadow2, pssm_blend); + } + + shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance #undef BIAS_FUNC } +#endif + + if (i < 4) { + shadow0 |= uint(clamp(shadow * 255.0, 0.0, 255.0)) << (i * 8); + } else { + shadow1 |= uint(clamp(shadow * 255.0, 0.0, 255.0)) << ((i - 4) * 8); + } + } + + for (uint i = 0; i < 8; i++) { + if (i >= scene_data.directional_light_count) { + break; + } + + if (!bool(directional_lights.data[i].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } + +#ifdef LIGHT_TRANSMITTANCE_USED + float transmittance_z = transmittance_depth; + + if (directional_lights.data[i].shadow_enabled) { + float depth_z = -vertex.z; + + if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.x, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix1 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.x; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.x; + + transmittance_z = z - shadow_z; + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.y, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix2 * trans_vertex; + trans_coord /= trans_coord.w; - light_compute(normal, directional_lights.data[i].direction, normalize(view), directional_lights.data[i].size, directional_lights.data[i].color * directional_lights.data[i].energy, 1.0, shadow_attenuation, albedo, roughness, metallic, specular, directional_lights.data[i].specular * specular_blob_intensity, + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.y; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.y; + + transmittance_z = z - shadow_z; + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.z, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix3 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.z; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.z; + + transmittance_z = z - shadow_z; + + } else { + vec4 trans_vertex = vec4(vertex - normalize(normal_interp) * directional_lights.data[i].shadow_transmittance_bias.w, 1.0); + vec4 trans_coord = directional_lights.data[i].shadow_matrix4 * trans_vertex; + trans_coord /= trans_coord.w; + + float shadow_z = textureLod(sampler2D(directional_shadow_atlas, material_samplers[SAMPLER_LINEAR_CLAMP]), trans_coord.xy, 0.0).r; + shadow_z *= directional_lights.data[i].shadow_transmittance_z_scale.w; + float z = trans_coord.z * directional_lights.data[i].shadow_transmittance_z_scale.w; + + transmittance_z = z - shadow_z; + } +#endif + + float shadow = 1.0; + + if (i < 4) { + shadow = float(shadow0 >> (i * 8) & 0xFF) / 255.0; + } else { + shadow = float(shadow1 >> ((i - 4) * 8) & 0xFF) / 255.0; + } + + blur_shadow(shadow); + + light_compute(normal, directional_lights.data[i].direction, normalize(view), directional_lights.data[i].color * directional_lights.data[i].energy, shadow, f0, orms, 1.0, #ifdef LIGHT_BACKLIGHT_USED - backlight, + backlight, #endif #ifdef LIGHT_TRANSMITTANCE_USED - transmittance_color, - transmittance_depth, - transmittance_curve, - transmittance_boost, - transmittance_z, + transmittance_color, + transmittance_depth, + transmittance_curve, + transmittance_boost, + transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim, rim_tint, + rim, rim_tint, albedo, #endif #ifdef LIGHT_CLEARCOAT_USED - clearcoat, clearcoat_gloss, + clearcoat, clearcoat_gloss, #endif #ifdef LIGHT_ANISOTROPY_USED - binormal, tangent, anisotropy, + binormal, tangent, anisotropy, +#endif +#ifdef USE_SOFT_SHADOW + directional_lights.data[i].size, #endif #ifdef USE_SHADOW_TO_OPACITY - alpha, + alpha, #endif - diffuse_light, - specular_light); + diffuse_light, + specular_light); + } } - } - { //omni lights + { //omni lights - uint omni_light_count = cluster_cell.x >> CLUSTER_COUNTER_SHIFT; - uint omni_light_pointer = cluster_cell.x & CLUSTER_POINTER_MASK; + uint cluster_omni_offset = cluster_offset; - for (uint i = 0; i < omni_light_count; i++) { - uint light_index = cluster_data.indices[omni_light_pointer + i]; + uint item_min; + uint item_max; + uint item_from; + uint item_to; - if (!bool(lights.data[light_index].mask & draw_call.layer_mask)) { - continue; //not masked - } + cluster_get_item_range(cluster_omni_offset + scene_data.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); + +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_omni_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + uint light_index = 32 * i + bit; - light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, albedo, roughness, metallic, specular, specular_blob_intensity, + if (!bool(omni_lights.data[light_index].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } + + float shadow = light_process_omni_shadow(light_index, vertex, view); + + shadow = blur_shadow(shadow); + + light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, #ifdef LIGHT_BACKLIGHT_USED - backlight, + backlight, #endif #ifdef LIGHT_TRANSMITTANCE_USED - transmittance_color, - transmittance_depth, - transmittance_curve, - transmittance_boost, + transmittance_color, + transmittance_depth, + transmittance_curve, + transmittance_boost, #endif #ifdef LIGHT_RIM_USED - rim, - rim_tint, + rim, + rim_tint, + albedo, #endif #ifdef LIGHT_CLEARCOAT_USED - clearcoat, clearcoat_gloss, + clearcoat, clearcoat_gloss, #endif #ifdef LIGHT_ANISOTROPY_USED - tangent, binormal, anisotropy, + tangent, binormal, anisotropy, #endif #ifdef USE_SHADOW_TO_OPACITY - alpha, + alpha, #endif - diffuse_light, specular_light); + diffuse_light, specular_light); + } + } } - } - { //spot lights - uint spot_light_count = cluster_cell.y >> CLUSTER_COUNTER_SHIFT; - uint spot_light_pointer = cluster_cell.y & CLUSTER_POINTER_MASK; + { //spot lights - for (uint i = 0; i < spot_light_count; i++) { - uint light_index = cluster_data.indices[spot_light_pointer + i]; + uint cluster_spot_offset = cluster_offset + scene_data.cluster_type_size; - if (!bool(lights.data[light_index].mask & draw_call.layer_mask)) { - continue; //not masked - } + uint item_min; + uint item_max; + uint item_from; + uint item_to; - light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, albedo, roughness, metallic, specular, specular_blob_intensity, + cluster_get_item_range(cluster_spot_offset + scene_data.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); + +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_spot_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + + uint light_index = 32 * i + bit; + + if (!bool(spot_lights.data[light_index].mask & instances.data[instance_index].layer_mask)) { + continue; //not masked + } + + float shadow = light_process_spot_shadow(light_index, vertex, view); + + shadow = blur_shadow(shadow); + + light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, #ifdef LIGHT_BACKLIGHT_USED - backlight, + backlight, #endif #ifdef LIGHT_TRANSMITTANCE_USED - transmittance_color, - transmittance_depth, - transmittance_curve, - transmittance_boost, + transmittance_color, + transmittance_depth, + transmittance_curve, + transmittance_boost, #endif #ifdef LIGHT_RIM_USED - rim, - rim_tint, + rim, + rim_tint, + albedo, #endif #ifdef LIGHT_CLEARCOAT_USED - clearcoat, clearcoat_gloss, + clearcoat, clearcoat_gloss, #endif #ifdef LIGHT_ANISOTROPY_USED - tangent, binormal, anisotropy, + tangent, binormal, anisotropy, #endif #ifdef USE_SHADOW_TO_OPACITY - alpha, + alpha, #endif - diffuse_light, specular_light); + diffuse_light, specular_light); + } + } } - } #ifdef USE_SHADOW_TO_OPACITY - alpha = min(alpha, clamp(length(ambient_light), 0.0, 1.0)); + alpha = min(alpha, clamp(length(ambient_light), 0.0, 1.0)); #if defined(ALPHA_SCISSOR_USED) - if (alpha < alpha_scissor) { - discard; - } + if (alpha < alpha_scissor) { + discard; + } #endif // ALPHA_SCISSOR_USED #ifdef USE_OPAQUE_PREPASS - if (alpha < opaque_prepass_threshold) { - discard; - } + if (alpha < opaque_prepass_threshold) { + discard; + } #endif // USE_OPAQUE_PREPASS @@ -2710,173 +3113,149 @@ FRAGMENT_SHADER_CODE #ifdef MODE_RENDER_SDF - { - vec3 local_pos = (scene_data.sdf_to_bounds * vec4(vertex, 1.0)).xyz; - ivec3 grid_pos = scene_data.sdf_offset + ivec3(local_pos * vec3(scene_data.sdf_size)); - - uint albedo16 = 0x1; //solid flag - albedo16 |= clamp(uint(albedo.r * 31.0), 0, 31) << 11; - albedo16 |= clamp(uint(albedo.g * 31.0), 0, 31) << 6; - albedo16 |= clamp(uint(albedo.b * 31.0), 0, 31) << 1; - - imageStore(albedo_volume_grid, grid_pos, uvec4(albedo16)); - - uint facing_bits = 0; - const vec3 aniso_dir[6] = vec3[]( - vec3(1, 0, 0), - vec3(0, 1, 0), - vec3(0, 0, 1), - vec3(-1, 0, 0), - vec3(0, -1, 0), - vec3(0, 0, -1)); - - vec3 cam_normal = mat3(scene_data.camera_matrix) * normalize(normal_interp); - - float closest_dist = -1e20; - - for (uint i = 0; i < 6; i++) { - float d = dot(cam_normal, aniso_dir[i]); - if (d > closest_dist) { - closest_dist = d; - facing_bits = (1 << i); + { + vec3 local_pos = (scene_data.sdf_to_bounds * vec4(vertex, 1.0)).xyz; + ivec3 grid_pos = scene_data.sdf_offset + ivec3(local_pos * vec3(scene_data.sdf_size)); + + uint albedo16 = 0x1; //solid flag + albedo16 |= clamp(uint(albedo.r * 31.0), 0, 31) << 11; + albedo16 |= clamp(uint(albedo.g * 31.0), 0, 31) << 6; + albedo16 |= clamp(uint(albedo.b * 31.0), 0, 31) << 1; + + imageStore(albedo_volume_grid, grid_pos, uvec4(albedo16)); + + uint facing_bits = 0; + const vec3 aniso_dir[6] = vec3[]( + vec3(1, 0, 0), + vec3(0, 1, 0), + vec3(0, 0, 1), + vec3(-1, 0, 0), + vec3(0, -1, 0), + vec3(0, 0, -1)); + + vec3 cam_normal = mat3(scene_data.camera_matrix) * normalize(normal_interp); + + float closest_dist = -1e20; + + for (uint i = 0; i < 6; i++) { + float d = dot(cam_normal, aniso_dir[i]); + if (d > closest_dist) { + closest_dist = d; + facing_bits = (1 << i); + } } - } - imageAtomicOr(geom_facing_grid, grid_pos, facing_bits); //store facing bits + imageAtomicOr(geom_facing_grid, grid_pos, facing_bits); //store facing bits - if (length(emission) > 0.001) { - float lumas[6]; - vec3 light_total = vec3(0); + if (length(emission) > 0.001) { + float lumas[6]; + vec3 light_total = vec3(0); - for (int i = 0; i < 6; i++) { - float strength = max(0.0, dot(cam_normal, aniso_dir[i])); - vec3 light = emission * strength; - light_total += light; - lumas[i] = max(light.r, max(light.g, light.b)); - } + for (int i = 0; i < 6; i++) { + float strength = max(0.0, dot(cam_normal, aniso_dir[i])); + vec3 light = emission * strength; + light_total += light; + lumas[i] = max(light.r, max(light.g, light.b)); + } - float luma_total = max(light_total.r, max(light_total.g, light_total.b)); + float luma_total = max(light_total.r, max(light_total.g, light_total.b)); - uint light_aniso = 0; + uint light_aniso = 0; - for (int i = 0; i < 6; i++) { - light_aniso |= min(31, uint((lumas[i] / luma_total) * 31.0)) << (i * 5); - } + for (int i = 0; i < 6; i++) { + light_aniso |= min(31, uint((lumas[i] / luma_total) * 31.0)) << (i * 5); + } - //compress to RGBE9995 to save space + //compress to RGBE9995 to save space - const float pow2to9 = 512.0f; - const float B = 15.0f; - const float N = 9.0f; - const float LN2 = 0.6931471805599453094172321215; + const float pow2to9 = 512.0f; + const float B = 15.0f; + const float N = 9.0f; + const float LN2 = 0.6931471805599453094172321215; - float cRed = clamp(light_total.r, 0.0, 65408.0); - float cGreen = clamp(light_total.g, 0.0, 65408.0); - float cBlue = clamp(light_total.b, 0.0, 65408.0); + float cRed = clamp(light_total.r, 0.0, 65408.0); + float cGreen = clamp(light_total.g, 0.0, 65408.0); + float cBlue = clamp(light_total.b, 0.0, 65408.0); - float cMax = max(cRed, max(cGreen, cBlue)); + float cMax = max(cRed, max(cGreen, cBlue)); - float expp = max(-B - 1.0f, floor(log(cMax) / LN2)) + 1.0f + B; + float expp = max(-B - 1.0f, floor(log(cMax) / LN2)) + 1.0f + B; - float sMax = floor((cMax / pow(2.0f, expp - B - N)) + 0.5f); + float sMax = floor((cMax / pow(2.0f, expp - B - N)) + 0.5f); - float exps = expp + 1.0f; + float exps = expp + 1.0f; - if (0.0 <= sMax && sMax < pow2to9) { - exps = expp; - } + if (0.0 <= sMax && sMax < pow2to9) { + exps = expp; + } - float sRed = floor((cRed / pow(2.0f, exps - B - N)) + 0.5f); - float sGreen = floor((cGreen / pow(2.0f, exps - B - N)) + 0.5f); - float sBlue = floor((cBlue / pow(2.0f, exps - B - N)) + 0.5f); - //store as 8985 to have 2 extra neighbour bits - uint light_rgbe = ((uint(sRed) & 0x1FF) >> 1) | ((uint(sGreen) & 0x1FF) << 8) | (((uint(sBlue) & 0x1FF) >> 1) << 17) | ((uint(exps) & 0x1F) << 25); + float sRed = floor((cRed / pow(2.0f, exps - B - N)) + 0.5f); + float sGreen = floor((cGreen / pow(2.0f, exps - B - N)) + 0.5f); + float sBlue = floor((cBlue / pow(2.0f, exps - B - N)) + 0.5f); + //store as 8985 to have 2 extra neighbour bits + uint light_rgbe = ((uint(sRed) & 0x1FF) >> 1) | ((uint(sGreen) & 0x1FF) << 8) | (((uint(sBlue) & 0x1FF) >> 1) << 17) | ((uint(exps) & 0x1F) << 25); - imageStore(emission_grid, grid_pos, uvec4(light_rgbe)); - imageStore(emission_aniso_grid, grid_pos, uvec4(light_aniso)); + imageStore(emission_grid, grid_pos, uvec4(light_rgbe)); + imageStore(emission_aniso_grid, grid_pos, uvec4(light_aniso)); + } } - } #endif #ifdef MODE_RENDER_MATERIAL - albedo_output_buffer.rgb = albedo; - albedo_output_buffer.a = alpha; + albedo_output_buffer.rgb = albedo; + albedo_output_buffer.a = alpha; - normal_output_buffer.rgb = normal * 0.5 + 0.5; - normal_output_buffer.a = 0.0; - depth_output_buffer.r = -vertex.z; + normal_output_buffer.rgb = normal * 0.5 + 0.5; + normal_output_buffer.a = 0.0; + depth_output_buffer.r = -vertex.z; -#if defined(AO_USED) - orm_output_buffer.r = ao; -#else - orm_output_buffer.r = 0.0; -#endif - orm_output_buffer.g = roughness; - orm_output_buffer.b = metallic; - orm_output_buffer.a = sss_strength; + orm_output_buffer.r = ao; + orm_output_buffer.g = roughness; + orm_output_buffer.b = metallic; + orm_output_buffer.a = sss_strength; - emission_output_buffer.rgb = emission; - emission_output_buffer.a = 0.0; + emission_output_buffer.rgb = emission; + emission_output_buffer.a = 0.0; #endif #ifdef MODE_RENDER_NORMAL_ROUGHNESS - normal_roughness_output_buffer = vec4(normal * 0.5 + 0.5, roughness); + normal_roughness_output_buffer = vec4(normal * 0.5 + 0.5, roughness); #ifdef MODE_RENDER_GIPROBE - if (bool(draw_call.flags & INSTANCE_FLAGS_USE_GIPROBE)) { // process giprobes - uint index1 = draw_call.gi_offset & 0xFFFF; - uint index2 = draw_call.gi_offset >> 16; - giprobe_buffer.x = index1 & 0xFF; - giprobe_buffer.y = index2 & 0xFF; - } else { - giprobe_buffer.x = 0xFF; - giprobe_buffer.y = 0xFF; - } + if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_GIPROBE)) { // process giprobes + uint index1 = instances.data[instance_index].gi_offset & 0xFFFF; + uint index2 = instances.data[instance_index].gi_offset >> 16; + giprobe_buffer.x = index1 & 0xFF; + giprobe_buffer.y = index2 & 0xFF; + } else { + giprobe_buffer.x = 0xFF; + giprobe_buffer.y = 0xFF; + } #endif -#endif //MODE_RENDER_NORMAL +#endif //MODE_RENDER_NORMAL_ROUGHNESS //nothing happens, so a tree-ssa optimizer will result in no fragment shader :) #else - specular_light *= scene_data.reflection_multiplier; - ambient_light *= albedo; //ambient must be multiplied by albedo at the end - -//ambient occlusion -#if defined(AO_USED) - -#ifndef LOW_END_MODE - if (scene_data.ssao_enabled && scene_data.ssao_ao_affect > 0.0) { - float ssao = texture(sampler2D(ao_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), screen_uv).r; - ao = mix(ao, min(ao, ssao), scene_data.ssao_ao_affect); - ao_light_affect = mix(ao_light_affect, max(ao_light_affect, scene_data.ssao_light_affect), scene_data.ssao_ao_affect); - } -#endif //LOW_END_MODE - - ambient_light = mix(scene_data.ao_color.rgb, ambient_light, ao); - ao_light_affect = mix(1.0, ao, ao_light_affect); - specular_light = mix(scene_data.ao_color.rgb, specular_light, ao_light_affect); - diffuse_light = mix(scene_data.ao_color.rgb, diffuse_light, ao_light_affect); -#else - -#ifndef LOW_END_MODE - if (scene_data.ssao_enabled) { - float ao = texture(sampler2D(ao_buffer, material_samplers[SAMPLER_LINEAR_CLAMP]), screen_uv).r; - ambient_light = mix(scene_data.ao_color.rgb, ambient_light, ao); - float ao_light_affect = mix(1.0, ao, scene_data.ssao_light_affect); - specular_light = mix(scene_data.ao_color.rgb, specular_light, ao_light_affect); - diffuse_light = mix(scene_data.ao_color.rgb, diffuse_light, ao_light_affect); - } -#endif //LOW_END_MODE + // multiply by albedo + diffuse_light *= albedo; // ambient must be multiplied by albedo at the end -#endif // AO_USED + // apply direct light AO + ao = unpackUnorm4x8(orms).x; + specular_light *= ao; + diffuse_light *= ao; - // base color remapping - diffuse_light *= 1.0 - metallic; // TODO: avoid all diffuse and ambient light calculations when metallic == 1 up to this point + // apply metallic + metallic = unpackUnorm4x8(orms).z; + diffuse_light *= 1.0 - metallic; ambient_light *= 1.0 - metallic; + //restore fog + fog = vec4(unpackHalf2x16(fog_rg), unpackHalf2x16(fog_ba)); + #ifdef MODE_MULTIPLE_RENDER_TARGETS #ifdef MODE_UNSHADED @@ -2892,25 +3271,8 @@ FRAGMENT_SHADER_CODE specular_buffer = vec4(specular_light, metallic); #endif - // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. - if (scene_data.fog_enabled) { - vec4 fog = fog_process(vertex); - diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); - specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); - } - -#ifndef LOW_END_MODE - if (scene_data.volumetric_fog_enabled) { - vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); - diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); - specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); - } -#endif // LOW_END_MODE - -#if defined(CUSTOM_FOG_USED) - diffuse_buffer.rgb = mix(diffuse_buffer.rgb, custom_fog.rgb, custom_fog.a); - specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), custom_fog.a); -#endif //CUSTOM_FOG_USED + diffuse_buffer.rgb = mix(diffuse_buffer.rgb, fog.rgb, fog.a); + specular_buffer.rgb = mix(specular_buffer.rgb, vec3(0.0), fog.a); #else //MODE_MULTIPLE_RENDER_TARGETS @@ -2922,22 +3284,10 @@ FRAGMENT_SHADER_CODE #endif //USE_NO_SHADING // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. - if (scene_data.fog_enabled) { - vec4 fog = fog_process(vertex); - frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); - } -#ifndef LOW_END_MODE - if (scene_data.volumetric_fog_enabled) { - vec4 fog = volumetric_fog_process(screen_uv, -vertex.z); - frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); - } -#endif - -#if defined(CUSTOM_FOG_USED) - frag_color.rgb = mix(frag_color.rgb, custom_fog.rgb, custom_fog.a); -#endif //CUSTOM_FOG_USED + frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); + ; #endif //MODE_MULTIPLE_RENDER_TARGETS #endif //MODE_RENDER_DEPTH -} + } diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl index 87ce74ba88..4ea05c9ccc 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl @@ -3,6 +3,15 @@ #define MAX_GI_PROBES 8 +#if defined(has_GL_KHR_shader_subgroup_ballot) && defined(has_GL_KHR_shader_subgroup_arithmetic) + +#extension GL_KHR_shader_subgroup_ballot : enable +#extension GL_KHR_shader_subgroup_arithmetic : enable + +#define USE_SUBGROUPS + +#endif + #include "cluster_data_inc.glsl" #if !defined(MODE_RENDER_DEPTH) || defined(MODE_RENDER_MATERIAL) || defined(MODE_RENDER_SDF) || defined(MODE_RENDER_NORMAL_ROUGHNESS) || defined(MODE_RENDER_GIPROBE) || defined(TANGENT_USED) || defined(NORMAL_MAP_USED) @@ -12,12 +21,10 @@ #endif layout(push_constant, binding = 0, std430) uniform DrawCall { - mat4 transform; - uint flags; - uint instance_uniforms_ofs; //base offset in global buffer for instance variables - uint gi_offset; //GI information when using lightmapping (VCT or lightmap index) - uint layer_mask; - vec4 lightmap_uv_scale; + uint instance_index; + uint uv_offset; + uint pad0; + uint pad1; } draw_call; @@ -36,91 +43,13 @@ draw_call; #define SAMPLER_NEAREST_WITH_MIPMAPS_ANISOTROPIC_REPEAT 10 #define SAMPLER_LINEAR_WITH_MIPMAPS_ANISOTROPIC_REPEAT 11 -layout(set = 0, binding = 1) uniform sampler material_samplers[12]; - -layout(set = 0, binding = 2) uniform sampler shadow_sampler; - #define SDFGI_MAX_CASCADES 8 -layout(set = 0, binding = 3, std140) uniform SceneData { - mat4 projection_matrix; - mat4 inv_projection_matrix; - - mat4 camera_matrix; - mat4 inv_camera_matrix; - - vec2 viewport_size; - vec2 screen_pixel_size; - - //use vec4s because std140 doesnt play nice with vec2s, z and w are wasted - vec4 directional_penumbra_shadow_kernel[32]; - vec4 directional_soft_shadow_kernel[32]; - vec4 penumbra_shadow_kernel[32]; - vec4 soft_shadow_kernel[32]; - - uint directional_penumbra_shadow_samples; - uint directional_soft_shadow_samples; - uint penumbra_shadow_samples; - uint soft_shadow_samples; - - vec4 ambient_light_color_energy; - - float ambient_color_sky_mix; - bool use_ambient_light; - bool use_ambient_cubemap; - bool use_reflection_cubemap; - - mat3 radiance_inverse_xform; - - vec2 shadow_atlas_pixel_size; - vec2 directional_shadow_pixel_size; - - uint directional_light_count; - float dual_paraboloid_side; - float z_far; - float z_near; - - bool ssao_enabled; - float ssao_light_affect; - float ssao_ao_affect; - bool roughness_limiter_enabled; - - float roughness_limiter_amount; - float roughness_limiter_limit; - uvec2 roughness_limiter_pad; - - vec4 ao_color; - - mat4 sdf_to_bounds; - - ivec3 sdf_offset; - bool material_uv2_mode; - - ivec3 sdf_size; - bool gi_upscale_for_msaa; - - bool volumetric_fog_enabled; - float volumetric_fog_inv_length; - float volumetric_fog_detail_spread; - uint volumetric_fog_pad; - - bool fog_enabled; - float fog_density; - float fog_height; - float fog_height_density; - - vec3 fog_light_color; - float fog_sun_scatter; - - float fog_aerial_perspective; - - float time; - float reflection_multiplier; // one normally, zero when rendering reflections +/* Set 1: Base Pass (never changes) */ - bool pancake_shadows; -} +layout(set = 0, binding = 1) uniform sampler material_samplers[12]; -scene_data; +layout(set = 0, binding = 2) uniform sampler shadow_sampler; #define INSTANCE_FLAGS_USE_GI_BUFFERS (1 << 6) #define INSTANCE_FLAGS_USE_SDFGI (1 << 7) @@ -139,17 +68,22 @@ scene_data; #define INSTANCE_FLAGS_SKELETON (1 << 19) #define INSTANCE_FLAGS_NON_UNIFORM_SCALE (1 << 20) -layout(set = 0, binding = 5, std430) restrict readonly buffer Lights { +layout(set = 0, binding = 3, std430) restrict readonly buffer OmniLights { + LightData data[]; +} +omni_lights; + +layout(set = 0, binding = 4, std430) restrict readonly buffer SpotLights { LightData data[]; } -lights; +spot_lights; -layout(set = 0, binding = 6) buffer restrict readonly ReflectionProbeData { +layout(set = 0, binding = 5) buffer restrict readonly ReflectionProbeData { ReflectionData data[]; } reflections; -layout(set = 0, binding = 7, std140) uniform DirectionalLights { +layout(set = 0, binding = 6, std140) uniform DirectionalLights { DirectionalLightData data[MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS]; } directional_lights; @@ -161,7 +95,7 @@ struct Lightmap { mat3 normal_xform; }; -layout(set = 0, binding = 10, std140) restrict readonly buffer Lightmaps { +layout(set = 0, binding = 7, std140) restrict readonly buffer Lightmaps { Lightmap data[]; } lightmaps; @@ -170,29 +104,20 @@ struct LightmapCapture { vec4 sh[9]; }; -layout(set = 0, binding = 11, std140) restrict readonly buffer LightmapCaptures { +layout(set = 0, binding = 8, std140) restrict readonly buffer LightmapCaptures { LightmapCapture data[]; } lightmap_captures; -layout(set = 0, binding = 12) uniform texture2D decal_atlas; -layout(set = 0, binding = 13) uniform texture2D decal_atlas_srgb; +layout(set = 0, binding = 9) uniform texture2D decal_atlas; +layout(set = 0, binding = 10) uniform texture2D decal_atlas_srgb; -layout(set = 0, binding = 14, std430) restrict readonly buffer Decals { +layout(set = 0, binding = 11, std430) restrict readonly buffer Decals { DecalData data[]; } decals; -layout(set = 0, binding = 15) uniform utexture3D cluster_texture; - -layout(set = 0, binding = 16, std430) restrict readonly buffer ClusterData { - uint indices[]; -} -cluster_data; - -layout(set = 0, binding = 17) uniform texture2D directional_shadow_atlas; - -layout(set = 0, binding = 18, std430) restrict readonly buffer GlobalVariableData { +layout(set = 0, binding = 12, std430) restrict readonly buffer GlobalVariableData { vec4 data[]; } global_variables; @@ -206,7 +131,7 @@ struct SDFGIProbeCascadeData { float to_cell; // 1/bounds * grid_size }; -layout(set = 0, binding = 19, std140) uniform SDFGI { +layout(set = 0, binding = 13, std140) uniform SDFGI { vec3 grid_size; uint max_cascades; @@ -236,40 +161,140 @@ sdfgi; #endif //LOW_END_MODE -// decal atlas +/* Set 2: Render Pass (changes per render pass) */ + +layout(set = 1, binding = 0, std140) uniform SceneData { + mat4 projection_matrix; + mat4 inv_projection_matrix; + + mat4 camera_matrix; + mat4 inv_camera_matrix; + + vec2 viewport_size; + vec2 screen_pixel_size; + + uint cluster_shift; + uint cluster_width; + uint cluster_type_size; + uint max_cluster_element_count_div_32; + + //use vec4s because std140 doesnt play nice with vec2s, z and w are wasted + vec4 directional_penumbra_shadow_kernel[32]; + vec4 directional_soft_shadow_kernel[32]; + vec4 penumbra_shadow_kernel[32]; + vec4 soft_shadow_kernel[32]; + + uint directional_penumbra_shadow_samples; + uint directional_soft_shadow_samples; + uint penumbra_shadow_samples; + uint soft_shadow_samples; + + vec4 ambient_light_color_energy; + + float ambient_color_sky_mix; + bool use_ambient_light; + bool use_ambient_cubemap; + bool use_reflection_cubemap; + + mat3 radiance_inverse_xform; + + vec2 shadow_atlas_pixel_size; + vec2 directional_shadow_pixel_size; + + uint directional_light_count; + float dual_paraboloid_side; + float z_far; + float z_near; -/* Set 1, Radiance */ + bool ssao_enabled; + float ssao_light_affect; + float ssao_ao_affect; + bool roughness_limiter_enabled; + + float roughness_limiter_amount; + float roughness_limiter_limit; + uvec2 roughness_limiter_pad; + + vec4 ao_color; + + mat4 sdf_to_bounds; + + ivec3 sdf_offset; + bool material_uv2_mode; + + ivec3 sdf_size; + bool gi_upscale_for_msaa; + + bool volumetric_fog_enabled; + float volumetric_fog_inv_length; + float volumetric_fog_detail_spread; + uint volumetric_fog_pad; + + bool fog_enabled; + float fog_density; + float fog_height; + float fog_height_density; + + vec3 fog_light_color; + float fog_sun_scatter; + + float fog_aerial_perspective; + + float time; + float reflection_multiplier; // one normally, zero when rendering reflections + + bool pancake_shadows; +} + +scene_data; + +struct InstanceData { + mat4 transform; + uint flags; + uint instance_uniforms_ofs; //base offset in global buffer for instance variables + uint gi_offset; //GI information when using lightmapping (VCT or lightmap index) + uint layer_mask; + vec4 lightmap_uv_scale; +}; + +layout(set = 1, binding = 1, std430) buffer restrict readonly InstanceDataBuffer { + InstanceData data[]; +} +instances; #ifdef USE_RADIANCE_CUBEMAP_ARRAY -layout(set = 1, binding = 0) uniform textureCubeArray radiance_cubemap; +layout(set = 1, binding = 2) uniform textureCubeArray radiance_cubemap; #else -layout(set = 1, binding = 0) uniform textureCube radiance_cubemap; +layout(set = 1, binding = 2) uniform textureCube radiance_cubemap; #endif -/* Set 2, Reflection and Shadow Atlases (view dependent) */ +layout(set = 1, binding = 3) uniform textureCubeArray reflection_atlas; -layout(set = 1, binding = 1) uniform textureCubeArray reflection_atlas; +layout(set = 1, binding = 4) uniform texture2D shadow_atlas; -layout(set = 1, binding = 2) uniform texture2D shadow_atlas; +layout(set = 1, binding = 5) uniform texture2D directional_shadow_atlas; -layout(set = 1, binding = 3) uniform texture2DArray lightmap_textures[MAX_LIGHTMAP_TEXTURES]; +layout(set = 1, binding = 6) uniform texture2DArray lightmap_textures[MAX_LIGHTMAP_TEXTURES]; -#ifndef LOW_END_MODE -layout(set = 1, binding = 4) uniform texture3D gi_probe_textures[MAX_GI_PROBES]; +#ifndef LOW_END_MOD +layout(set = 1, binding = 7) uniform texture3D gi_probe_textures[MAX_GI_PROBES]; #endif -/* Set 3, Render Buffers */ +layout(set = 1, binding = 8, std430) buffer restrict readonly ClusterBuffer { + uint data[]; +} +cluster_buffer; #ifdef MODE_RENDER_SDF -layout(r16ui, set = 1, binding = 5) uniform restrict writeonly uimage3D albedo_volume_grid; -layout(r32ui, set = 1, binding = 6) uniform restrict writeonly uimage3D emission_grid; -layout(r32ui, set = 1, binding = 7) uniform restrict writeonly uimage3D emission_aniso_grid; -layout(r32ui, set = 1, binding = 8) uniform restrict uimage3D geom_facing_grid; +layout(r16ui, set = 1, binding = 9) uniform restrict writeonly uimage3D albedo_volume_grid; +layout(r32ui, set = 1, binding = 10) uniform restrict writeonly uimage3D emission_grid; +layout(r32ui, set = 1, binding = 11) uniform restrict writeonly uimage3D emission_aniso_grid; +layout(r32ui, set = 1, binding = 12) uniform restrict uimage3D geom_facing_grid; //still need to be present for shaders that use it, so remap them to something #define depth_buffer shadow_atlas @@ -278,17 +303,17 @@ layout(r32ui, set = 1, binding = 8) uniform restrict uimage3D geom_facing_grid; #else -layout(set = 1, binding = 5) uniform texture2D depth_buffer; -layout(set = 1, binding = 6) uniform texture2D color_buffer; +layout(set = 1, binding = 9) uniform texture2D depth_buffer; +layout(set = 1, binding = 10) uniform texture2D color_buffer; #ifndef LOW_END_MODE -layout(set = 1, binding = 7) uniform texture2D normal_roughness_buffer; -layout(set = 1, binding = 8) uniform texture2D ao_buffer; -layout(set = 1, binding = 9) uniform texture2D ambient_buffer; -layout(set = 1, binding = 10) uniform texture2D reflection_buffer; -layout(set = 1, binding = 11) uniform texture2DArray sdfgi_lightprobe_texture; -layout(set = 1, binding = 12) uniform texture3D sdfgi_occlusion_cascades; +layout(set = 1, binding = 11) uniform texture2D normal_roughness_buffer; +layout(set = 1, binding = 12) uniform texture2D ao_buffer; +layout(set = 1, binding = 13) uniform texture2D ambient_buffer; +layout(set = 1, binding = 14) uniform texture2D reflection_buffer; +layout(set = 1, binding = 15) uniform texture2DArray sdfgi_lightprobe_texture; +layout(set = 1, binding = 16) uniform texture3D sdfgi_occlusion_cascades; struct GIProbeData { mat4 xform; @@ -306,22 +331,22 @@ struct GIProbeData { uint mipmaps; }; -layout(set = 1, binding = 13, std140) uniform GIProbes { +layout(set = 1, binding = 17, std140) uniform GIProbes { GIProbeData data[MAX_GI_PROBES]; } gi_probes; -layout(set = 1, binding = 14) uniform texture3D volumetric_fog_texture; +layout(set = 1, binding = 18) uniform texture3D volumetric_fog_texture; #endif // LOW_END_MODE #endif -/* Set 4 Skeleton & Instancing (Multimesh) */ +/* Set 2 Skeleton & Instancing (can change per item) */ layout(set = 2, binding = 0, std430) restrict readonly buffer Transforms { vec4 data[]; } transforms; -/* Set 5 User Material */ +/* Set 3 User Material */ diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl index 813ea29fa1..e4c3f3a84b 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_debug.glsl @@ -97,6 +97,8 @@ void main() { float blend = 0.0; #if 1 + // No interpolation + vec3 inv_dir = 1.0 / ray_dir; float rough = 0.5; @@ -161,114 +163,11 @@ void main() { hit_light *= (dot(max(vec3(0.0), (hit_normal * hit_aniso0)), vec3(1.0)) + dot(max(vec3(0.0), (-hit_normal * hit_aniso1)), vec3(1.0))); - if (blend > 0.0) { - light = mix(light, hit_light, blend); - blend = 0.0; - } else { - light = hit_light; - - //process blend - float blend_from = (float(params.probe_axis_size - 1) / 2.0) - 2.5; - float blend_to = blend_from + 2.0; - - vec3 cam_pos = params.cam_transform[3].xyz - cascades.data[i].offset; - cam_pos *= cascades.data[i].to_cell; - - pos += ray_dir * min(advance, max_advance); - vec3 inner_pos = pos - cam_pos; - - inner_pos = inner_pos * float(params.probe_axis_size - 1) / params.grid_size.x; - - float len = length(inner_pos); - - inner_pos = abs(normalize(inner_pos)); - len *= max(inner_pos.x, max(inner_pos.y, inner_pos.z)); - - if (len >= blend_from) { - blend = smoothstep(blend_from, blend_to, len); - - pos /= cascades.data[i].to_cell; - pos += cascades.data[i].offset; - ray_pos = pos; - hit = false; //continue trace for blend - - continue; - } - } + light = hit_light; break; } - light = mix(light, vec3(0.0), blend); - -#else - - vec3 inv_dir = 1.0 / ray_dir; - - bool hit = false; - vec4 light_accum = vec4(0.0); - - float blend_size = (params.grid_size.x / float(params.probe_axis_size - 1)) * 0.5; - - float radius_sizes[MAX_CASCADES]; - for (uint i = 0; i < params.max_cascades; i++) { - radius_sizes[i] = (1.0 / cascades.data[i].to_cell) * (params.grid_size.x * 0.5 - blend_size); - } - - float max_distance = radius_sizes[params.max_cascades - 1]; - float advance = 0; - while (advance < max_distance) { - for (uint i = 0; i < params.max_cascades; i++) { - if (advance < radius_sizes[i]) { - vec3 pos = (ray_pos + ray_dir * advance) - cascades.data[i].offset; - pos *= cascades.data[i].to_cell * pos_to_uvw; - - float distance = texture(sampler3D(sdf_cascades[i], linear_sampler), pos).r * 255.0 - 1.0; - - vec4 hit_light = vec4(0.0); - if (distance < 1.0) { - hit_light.a = max(0.0, 1.0 - distance); - hit_light.rgb = texture(sampler3D(light_cascades[i], linear_sampler), pos).rgb; - hit_light.rgb *= hit_light.a; - } - - distance /= cascades.data[i].to_cell; - - if (i < (params.max_cascades - 1)) { - pos = (ray_pos + ray_dir * advance) - cascades.data[i + 1].offset; - pos *= cascades.data[i + 1].to_cell * pos_to_uvw; - - float distance2 = texture(sampler3D(sdf_cascades[i + 1], linear_sampler), pos).r * 255.0 - 1.0; - - vec4 hit_light2 = vec4(0.0); - if (distance2 < 1.0) { - hit_light2.a = max(0.0, 1.0 - distance2); - hit_light2.rgb = texture(sampler3D(light_cascades[i + 1], linear_sampler), pos).rgb; - hit_light2.rgb *= hit_light2.a; - } - - float prev_radius = i == 0 ? 0.0 : radius_sizes[i - 1]; - float blend = (advance - prev_radius) / (radius_sizes[i] - prev_radius); - - distance2 /= cascades.data[i + 1].to_cell; - - hit_light = mix(hit_light, hit_light2, blend); - distance = mix(distance, distance2, blend); - } - - light_accum += hit_light; - advance += distance; - break; - } - } - - if (light_accum.a > 0.98) { - break; - } - } - - light = light_accum.rgb / light_accum.a; - #endif imageStore(screen_buffer, screen_pos, vec4(linear_to_srgb(light), 1.0)); diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl index 30dbf5871f..dc7238abed 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_direct_light.glsl @@ -67,8 +67,8 @@ struct Light { float attenuation; uint type; - float spot_angle; - float spot_attenuation; + float cos_spot_angle; + float inv_spot_attenuation; float radius; vec4 shadow_color; @@ -80,6 +80,7 @@ layout(set = 0, binding = 9, std140) buffer restrict readonly Lights { lights; layout(set = 0, binding = 10) uniform texture2DArray lightprobe_texture; +layout(set = 0, binding = 11) uniform texture3D occlusion_texture; layout(push_constant, binding = 0, std430) uniform Params { vec3 grid_size; @@ -91,9 +92,9 @@ layout(push_constant, binding = 0, std430) uniform Params { uint process_increment; int probe_axis_size; - bool multibounce; + float bounce_feedback; float y_mult; - uint pad; + bool use_occlusion; } params; @@ -125,7 +126,10 @@ void main() { uint voxel_index = uint(gl_GlobalInvocationID.x); //used for skipping voxels every N frames - voxel_index = params.process_offset + voxel_index * params.process_increment; + if (params.process_increment > 1) { + voxel_index *= params.process_increment; + voxel_index += params.process_offset; + } if (voxel_index >= dispatch_data.total_count) { return; @@ -143,10 +147,96 @@ void main() { uint voxel_albedo = process_voxels.data[voxel_index].albedo; vec3 albedo = vec3(uvec3(voxel_albedo >> 10, voxel_albedo >> 5, voxel_albedo) & uvec3(0x1F)) / float(0x1F); - vec3 light_accum[6]; - + vec3 light_accum[6] = vec3[](vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0), vec3(0.0)); uint valid_aniso = (voxel_albedo >> 15) & 0x3F; + const vec3 aniso_dir[6] = vec3[]( + vec3(1, 0, 0), + vec3(0, 1, 0), + vec3(0, 0, 1), + vec3(-1, 0, 0), + vec3(0, -1, 0), + vec3(0, 0, -1)); + + // Add indirect light first, in order to save computation resources +#ifdef MODE_PROCESS_DYNAMIC + if (params.bounce_feedback > 0.001) { + vec3 feedback = (params.bounce_feedback < 1.0) ? (albedo * params.bounce_feedback) : mix(albedo, vec3(1.0), params.bounce_feedback - 1.0); + vec3 pos = (vec3(positioni) + vec3(0.5)) * float(params.probe_axis_size - 1) / params.grid_size; + ivec3 probe_base_pos = ivec3(pos); + + float weight_accum[6] = float[](0, 0, 0, 0, 0, 0); + + ivec3 tex_pos = ivec3(probe_base_pos.xy, int(params.cascade)); + tex_pos.x += probe_base_pos.z * int(params.probe_axis_size); + + tex_pos.xy = tex_pos.xy * (OCT_SIZE + 2) + ivec2(1); + + vec3 base_tex_posf = vec3(tex_pos); + vec2 tex_pixel_size = 1.0 / vec2(ivec2((OCT_SIZE + 2) * params.probe_axis_size * params.probe_axis_size, (OCT_SIZE + 2) * params.probe_axis_size)); + vec3 probe_uv_offset = vec3(ivec3(OCT_SIZE + 2, OCT_SIZE + 2, (OCT_SIZE + 2) * params.probe_axis_size)) * tex_pixel_size.xyx; + + for (uint j = 0; j < 8; j++) { + ivec3 offset = (ivec3(j) >> ivec3(0, 1, 2)) & ivec3(1, 1, 1); + ivec3 probe_posi = probe_base_pos; + probe_posi += offset; + + // Compute weight + + vec3 probe_pos = vec3(probe_posi); + vec3 probe_to_pos = pos - probe_pos; + vec3 probe_dir = normalize(-probe_to_pos); + + // Compute lightprobe texture position + + vec3 trilinear = vec3(1.0) - abs(probe_to_pos); + + for (uint k = 0; k < 6; k++) { + if (bool(valid_aniso & (1 << k))) { + vec3 n = aniso_dir[k]; + float weight = trilinear.x * trilinear.y * trilinear.z * max(0, dot(n, probe_dir)); + + if (weight > 0.0 && params.use_occlusion) { + ivec3 occ_indexv = abs((cascades.data[params.cascade].probe_world_offset + probe_posi) & ivec3(1, 1, 1)) * ivec3(1, 2, 4); + vec4 occ_mask = mix(vec4(0.0), vec4(1.0), equal(ivec4(occ_indexv.x | occ_indexv.y), ivec4(0, 1, 2, 3))); + + vec3 occ_pos = (vec3(positioni) + aniso_dir[k] + vec3(0.5)) / params.grid_size; + occ_pos.z += float(params.cascade); + if (occ_indexv.z != 0) { //z bit is on, means index is >=4, so make it switch to the other half of textures + occ_pos.x += 1.0; + } + occ_pos *= vec3(0.5, 1.0, 1.0 / float(params.max_cascades)); //renormalize + float occlusion = dot(textureLod(sampler3D(occlusion_texture, linear_sampler), occ_pos, 0.0), occ_mask); + + weight *= occlusion; + } + + if (weight > 0.0) { + vec3 tex_posf = base_tex_posf + vec3(octahedron_encode(n) * float(OCT_SIZE), 0.0); + tex_posf.xy *= tex_pixel_size; + + vec3 pos_uvw = tex_posf; + pos_uvw.xy += vec2(offset.xy) * probe_uv_offset.xy; + pos_uvw.x += float(offset.z) * probe_uv_offset.z; + vec3 indirect_light = textureLod(sampler2DArray(lightprobe_texture, linear_sampler), pos_uvw, 0.0).rgb; + + light_accum[k] += indirect_light * weight; + weight_accum[k] += weight; + } + } + } + } + + for (uint k = 0; k < 6; k++) { + if (weight_accum[k] > 0.0) { + light_accum[k] /= weight_accum[k]; + light_accum[k] *= feedback; + } + } + } + +#endif + { uint rgbe = process_voxels.data[voxel_index].light; @@ -162,18 +252,10 @@ void main() { uint aniso = process_voxels.data[voxel_index].light_aniso; for (uint i = 0; i < 6; i++) { float strength = ((aniso >> (i * 5)) & 0x1F) / float(0x1F); - light_accum[i] = l * strength; + light_accum[i] += l * strength; } } - const vec3 aniso_dir[6] = vec3[]( - vec3(1, 0, 0), - vec3(0, 1, 0), - vec3(0, 0, 1), - vec3(-1, 0, 0), - vec3(0, -1, 0), - vec3(0, 0, -1)); - // Raytrace light vec3 pos_to_uvw = 1.0 / params.grid_size; @@ -203,13 +285,16 @@ void main() { rel_vec.y /= params.y_mult; attenuation = get_omni_attenuation(light_distance, 1.0 / lights.data[i].radius, lights.data[i].attenuation); - float angle = acos(dot(normalize(rel_vec), -lights.data[i].direction)); - if (angle > lights.data[i].spot_angle) { - attenuation = 0.0; - } else { - float d = clamp(angle / lights.data[i].spot_angle, 0, 1); - attenuation *= pow(1.0 - d, lights.data[i].spot_attenuation); + float cos_spot_angle = lights.data[i].cos_spot_angle; + float cos_angle = dot(-direction, lights.data[i].direction); + + if (cos_angle < cos_spot_angle) { + continue; } + + float scos = max(cos_angle, cos_spot_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - cos_spot_angle)); + attenuation *= 1.0 - pow(spot_rim, lights.data[i].inv_spot_attenuation); } break; } @@ -292,65 +377,6 @@ void main() { } } - // Add indirect light - - if (params.multibounce) { - vec3 pos = (vec3(positioni) + vec3(0.5)) * float(params.probe_axis_size - 1) / params.grid_size; - ivec3 probe_base_pos = ivec3(pos); - - vec4 probe_accum[6] = vec4[](vec4(0.0), vec4(0.0), vec4(0.0), vec4(0.0), vec4(0.0), vec4(0.0)); - float weight_accum[6] = float[](0, 0, 0, 0, 0, 0); - - ivec3 tex_pos = ivec3(probe_base_pos.xy, int(params.cascade)); - tex_pos.x += probe_base_pos.z * int(params.probe_axis_size); - - tex_pos.xy = tex_pos.xy * (OCT_SIZE + 2) + ivec2(1); - - vec3 base_tex_posf = vec3(tex_pos); - vec2 tex_pixel_size = 1.0 / vec2(ivec2((OCT_SIZE + 2) * params.probe_axis_size * params.probe_axis_size, (OCT_SIZE + 2) * params.probe_axis_size)); - vec3 probe_uv_offset = (ivec3(OCT_SIZE + 2, OCT_SIZE + 2, (OCT_SIZE + 2) * params.probe_axis_size)) * tex_pixel_size.xyx; - - for (uint j = 0; j < 8; j++) { - ivec3 offset = (ivec3(j) >> ivec3(0, 1, 2)) & ivec3(1, 1, 1); - ivec3 probe_posi = probe_base_pos; - probe_posi += offset; - - // Compute weight - - vec3 probe_pos = vec3(probe_posi); - vec3 probe_to_pos = pos - probe_pos; - vec3 probe_dir = normalize(-probe_to_pos); - - // Compute lightprobe texture position - - vec3 trilinear = vec3(1.0) - abs(probe_to_pos); - - for (uint k = 0; k < 6; k++) { - if (bool(valid_aniso & (1 << k))) { - vec3 n = aniso_dir[k]; - float weight = trilinear.x * trilinear.y * trilinear.z * max(0.005, dot(n, probe_dir)); - - vec3 tex_posf = base_tex_posf + vec3(octahedron_encode(n) * float(OCT_SIZE), 0.0); - tex_posf.xy *= tex_pixel_size; - - vec3 pos_uvw = tex_posf; - pos_uvw.xy += vec2(offset.xy) * probe_uv_offset.xy; - pos_uvw.x += float(offset.z) * probe_uv_offset.z; - vec4 indirect_light = textureLod(sampler2DArray(lightprobe_texture, linear_sampler), pos_uvw, 0.0); - - probe_accum[k] += indirect_light * weight; - weight_accum[k] += weight; - } - } - } - - for (uint k = 0; k < 6; k++) { - if (weight_accum[k] > 0.0) { - light_accum[k] += probe_accum[k].rgb * albedo / weight_accum[k]; - } - } - } - // Store the light in the light texture float lumas[6]; diff --git a/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl b/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl index d516ab22c3..007e4c113a 100644 --- a/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl +++ b/servers/rendering/renderer_rd/shaders/sdfgi_integrate.glsl @@ -39,8 +39,11 @@ layout(rgba32i, set = 0, binding = 13) uniform restrict iimage2D lightprobe_aver layout(rgba16f, set = 0, binding = 14) uniform restrict writeonly image2DArray lightprobe_ambient_texture; +#ifdef USE_CUBEMAP_ARRAY +layout(set = 1, binding = 0) uniform textureCubeArray sky_irradiance; +#else layout(set = 1, binding = 0) uniform textureCube sky_irradiance; - +#endif layout(set = 1, binding = 1) uniform sampler linear_sampler_mipmaps; #define HISTORY_BITS 10 @@ -136,12 +139,24 @@ uint rgbe_encode(vec3 color) { return (uint(sRed) & 0x1FF) | ((uint(sGreen) & 0x1FF) << 9) | ((uint(sBlue) & 0x1FF) << 18) | ((uint(exps) & 0x1F) << 27); } +struct SH { +#if (SH_SIZE == 16) + float c[48]; +#else + float c[28]; +#endif +}; + +shared SH sh_accum[64]; //8x8 + void main() { ivec2 pos = ivec2(gl_GlobalInvocationID.xy); if (any(greaterThanEqual(pos, params.image_size))) { //too large, do nothing return; } + uint probe_index = gl_LocalInvocationID.x + gl_LocalInvocationID.y * 8; + #ifdef MODE_PROCESS float probe_cell_size = float(params.grid_size.x / float(params.probe_axis_size - 1)) / cascades.data[params.cascade].to_cell; @@ -154,27 +169,9 @@ void main() { vec3 probe_pos = cascades.data[params.cascade].offset + vec3(probe_cell) * probe_cell_size; vec3 pos_to_uvw = 1.0 / params.grid_size; - vec4 probe_sh_accum[SH_SIZE] = vec4[]( - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0) -#if (SH_SIZE == 16) - , - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0), - vec4(0.0) -#endif - ); + for (uint i = 0; i < SH_SIZE * 3; i++) { + sh_accum[probe_index].c[i] = 0.0; + } // quickly ensure each probe has a different "offset" for the vogel function, based on integer world position uvec3 h3 = hash3(uvec3(params.world_offset + probe_cell)); @@ -195,14 +192,12 @@ void main() { vec3 inv_dir = 1.0 / ray_dir; bool hit = false; - vec3 hit_normal; - vec3 hit_light; - vec3 hit_aniso0; - vec3 hit_aniso1; + uint hit_cascade; float bias = params.ray_bias; vec3 abs_ray_dir = abs(ray_dir); ray_pos += ray_dir * 1.0 / max(abs_ray_dir.x, max(abs_ray_dir.y, abs_ray_dir.z)) * bias / cascades.data[params.cascade].to_cell; + vec3 uvw; for (uint j = params.cascade; j < params.max_cascades; j++) { //convert to local bounds @@ -221,14 +216,12 @@ void main() { float advance = 0.0; - vec3 uvw; - while (advance < max_advance) { //read how much to advance from SDF uvw = (pos + ray_dir * advance) * pos_to_uvw; float distance = texture(sampler3D(sdf_cascades[j], linear_sampler), uvw).r * 255.0 - 1.0; - if (distance < 0.001) { + if (distance < 0.05) { //consider hit hit = true; break; @@ -238,17 +231,7 @@ void main() { } if (hit) { - const float EPSILON = 0.001; - hit_normal = normalize(vec3( - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw + vec3(EPSILON, 0.0, 0.0)).r - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw - vec3(EPSILON, 0.0, 0.0)).r, - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw + vec3(0.0, EPSILON, 0.0)).r - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw - vec3(0.0, EPSILON, 0.0)).r, - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw + vec3(0.0, 0.0, EPSILON)).r - texture(sampler3D(sdf_cascades[j], linear_sampler), uvw - vec3(0.0, 0.0, EPSILON)).r)); - - hit_light = texture(sampler3D(light_cascades[j], linear_sampler), uvw).rgb; - vec4 aniso0 = texture(sampler3D(aniso0_cascades[j], linear_sampler), uvw); - hit_aniso0 = aniso0.rgb; - hit_aniso1 = vec3(aniso0.a, texture(sampler3D(aniso1_cascades[j], linear_sampler), uvw).rg); - + hit_cascade = j; break; } @@ -261,11 +244,32 @@ void main() { vec4 light; if (hit) { - //one liner magic - light.rgb = hit_light * (dot(max(vec3(0.0), (hit_normal * hit_aniso0)), vec3(1.0)) + dot(max(vec3(0.0), (-hit_normal * hit_aniso1)), vec3(1.0))); - light.a = 1.0; + //avoid reading different texture from different threads + for (uint j = params.cascade; j < params.max_cascades; j++) { + if (j == hit_cascade) { + const float EPSILON = 0.001; + vec3 hit_normal = normalize(vec3( + texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw + vec3(EPSILON, 0.0, 0.0)).r - texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw - vec3(EPSILON, 0.0, 0.0)).r, + texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw + vec3(0.0, EPSILON, 0.0)).r - texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw - vec3(0.0, EPSILON, 0.0)).r, + texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw + vec3(0.0, 0.0, EPSILON)).r - texture(sampler3D(sdf_cascades[hit_cascade], linear_sampler), uvw - vec3(0.0, 0.0, EPSILON)).r)); + + vec3 hit_light = texture(sampler3D(light_cascades[hit_cascade], linear_sampler), uvw).rgb; + vec4 aniso0 = texture(sampler3D(aniso0_cascades[hit_cascade], linear_sampler), uvw); + vec3 hit_aniso0 = aniso0.rgb; + vec3 hit_aniso1 = vec3(aniso0.a, texture(sampler3D(aniso1_cascades[hit_cascade], linear_sampler), uvw).rg); + + //one liner magic + light.rgb = hit_light * (dot(max(vec3(0.0), (hit_normal * hit_aniso0)), vec3(1.0)) + dot(max(vec3(0.0), (-hit_normal * hit_aniso1)), vec3(1.0))); + light.a = 1.0; + } + } + } else if (params.sky_mode == SKY_MODE_SKY) { +#ifdef USE_CUBEMAP_ARRAY + light.rgb = textureLod(samplerCubeArray(sky_irradiance, linear_sampler_mipmaps), vec4(ray_dir, 0.0), 2.0).rgb; //use second mipmap because we dont usually throw a lot of rays, so this compensates +#else light.rgb = textureLod(samplerCube(sky_irradiance, linear_sampler_mipmaps), ray_dir, 2.0).rgb; //use second mipmap because we dont usually throw a lot of rays, so this compensates +#endif light.rgb *= params.sky_energy; light.a = 0.0; @@ -278,33 +282,33 @@ void main() { } vec3 ray_dir2 = ray_dir * ray_dir; - float c[SH_SIZE] = float[]( - - 0.282095, //l0 - 0.488603 * ray_dir.y, //l1n1 - 0.488603 * ray_dir.z, //l1n0 - 0.488603 * ray_dir.x, //l1p1 - 1.092548 * ray_dir.x * ray_dir.y, //l2n2 - 1.092548 * ray_dir.y * ray_dir.z, //l2n1 - 0.315392 * (3.0 * ray_dir2.z - 1.0), //l20 - 1.092548 * ray_dir.x * ray_dir.z, //l2p1 - 0.546274 * (ray_dir2.x - ray_dir2.y) //l2p2 + +#define SH_ACCUM(m_idx, m_value) \ + { \ + vec3 l = light.rgb * (m_value); \ + sh_accum[probe_index].c[m_idx * 3 + 0] += l.r; \ + sh_accum[probe_index].c[m_idx * 3 + 1] += l.g; \ + sh_accum[probe_index].c[m_idx * 3 + 2] += l.b; \ + } + SH_ACCUM(0, 0.282095); //l0 + SH_ACCUM(1, 0.488603 * ray_dir.y); //l1n1 + SH_ACCUM(2, 0.488603 * ray_dir.z); //l1n0 + SH_ACCUM(3, 0.488603 * ray_dir.x); //l1p1 + SH_ACCUM(4, 1.092548 * ray_dir.x * ray_dir.y); //l2n2 + SH_ACCUM(5, 1.092548 * ray_dir.y * ray_dir.z); //l2n1 + SH_ACCUM(6, 0.315392 * (3.0 * ray_dir2.z - 1.0)); //l20 + SH_ACCUM(7, 1.092548 * ray_dir.x * ray_dir.z); //l2p1 + SH_ACCUM(8, 0.546274 * (ray_dir2.x - ray_dir2.y)); //l2p2 #if (SH_SIZE == 16) - , - 0.590043 * ray_dir.y * (3.0f * ray_dir2.x - ray_dir2.y), - 2.890611 * ray_dir.y * ray_dir.x * ray_dir.z, - 0.646360 * ray_dir.y * (-1.0f + 5.0f * ray_dir2.z), - 0.373176 * (5.0f * ray_dir2.z * ray_dir.z - 3.0f * ray_dir.z), - 0.457045 * ray_dir.x * (-1.0f + 5.0f * ray_dir2.z), - 1.445305 * (ray_dir2.x - ray_dir2.y) * ray_dir.z, - 0.590043 * ray_dir.x * (ray_dir2.x - 3.0f * ray_dir2.y) + SH_ACCUM(9, 0.590043 * ray_dir.y * (3.0f * ray_dir2.x - ray_dir2.y)); + SH_ACCUM(10, 2.890611 * ray_dir.y * ray_dir.x * ray_dir.z); + SH_ACCUM(11, 0.646360 * ray_dir.y * (-1.0f + 5.0f * ray_dir2.z)); + SH_ACCUM(12, 0.373176 * (5.0f * ray_dir2.z * ray_dir.z - 3.0f * ray_dir.z)); + SH_ACCUM(13, 0.457045 * ray_dir.x * (-1.0f + 5.0f * ray_dir2.z)); + SH_ACCUM(14, 1.445305 * (ray_dir2.x - ray_dir2.y) * ray_dir.z); + SH_ACCUM(15, 0.590043 * ray_dir.x * (ray_dir2.x - 3.0f * ray_dir2.y)); #endif - ); - - for (uint j = 0; j < SH_SIZE; j++) { - probe_sh_accum[j] += light * c[j]; - } } for (uint i = 0; i < SH_SIZE; i++) { @@ -312,7 +316,7 @@ void main() { ivec3 prev_pos = ivec3(pos.x, pos.y * SH_SIZE + i, int(params.history_index)); ivec2 average_pos = prev_pos.xy; - vec4 value = probe_sh_accum[i] * 4.0 / float(params.ray_count); + vec4 value = vec4(sh_accum[probe_index].c[i * 3 + 0], sh_accum[probe_index].c[i * 3 + 1], sh_accum[probe_index].c[i * 3 + 2], 1.0) * 4.0 / float(params.ray_count); ivec4 ivalue = clamp(ivec4(value * float(1 << HISTORY_BITS)), -32768, 32767); //clamp to 16 bits, so higher values don't break average @@ -344,37 +348,11 @@ void main() { ivec2 oct_pos = (pos / OCT_SIZE) * (OCT_SIZE + 2) + ivec2(1); ivec2 local_pos = pos % OCT_SIZE; - //fill the spherical harmonic - vec4 sh[SH_SIZE]; - - for (uint i = 0; i < SH_SIZE; i++) { - // store in history texture - ivec2 average_pos = sh_pos + ivec2(0, i); - ivec4 average = imageLoad(lightprobe_average_texture, average_pos); - - sh[i] = (vec4(average) / float(params.history_size)) / float(1 << HISTORY_BITS); - } - //compute the octahedral normal for this texel vec3 normal = octahedron_encode(vec2(local_pos) / float(OCT_SIZE)); - /* + // read the spherical harmonic - const float c1 = 0.429043; - const float c2 = 0.511664; - const float c3 = 0.743125; - const float c4 = 0.886227; - const float c5 = 0.247708; - vec4 light = (c1 * sh[8] * (normal.x * normal.x - normal.y * normal.y) + - c3 * sh[6] * normal.z * normal.z + - c4 * sh[0] - - c5 * sh[6] + - 2.0 * c1 * sh[4] * normal.x * normal.y + - 2.0 * c1 * sh[7] * normal.x * normal.z + - 2.0 * c1 * sh[5] * normal.y * normal.z + - 2.0 * c2 * sh[3] * normal.x + - 2.0 * c2 * sh[1] * normal.y + - 2.0 * c2 * sh[2] * normal.z); -*/ + vec3 normal2 = normal * normal; float c[SH_SIZE] = float[]( @@ -426,7 +404,14 @@ void main() { vec3 radiance = vec3(0.0); for (uint i = 0; i < SH_SIZE; i++) { - vec3 m = sh[i].rgb * c[i] * 4.0; + // store in history texture + ivec2 average_pos = sh_pos + ivec2(0, i); + ivec4 average = imageLoad(lightprobe_average_texture, average_pos); + + vec4 sh = (vec4(average) / float(params.history_size)) / float(1 << HISTORY_BITS); + + vec3 m = sh.rgb * c[i] * 4.0; + irradiance += m * l_mult[i]; radiance += m; } @@ -515,13 +500,15 @@ void main() { //can't scroll, must look for position in parent cascade //to global coords - float probe_cell_size = float(params.grid_size.x / float(params.probe_axis_size - 1)) / cascades.data[params.cascade].to_cell; + float cell_to_probe = float(params.grid_size.x / float(params.probe_axis_size - 1)); + + float probe_cell_size = cell_to_probe / cascades.data[params.cascade].to_cell; vec3 probe_pos = cascades.data[params.cascade].offset + vec3(probe_cell) * probe_cell_size; //to parent local coords + float probe_cell_size_next = cell_to_probe / cascades.data[params.cascade + 1].to_cell; probe_pos -= cascades.data[params.cascade + 1].offset; - probe_pos *= cascades.data[params.cascade + 1].to_cell; - probe_pos = probe_pos * float(params.probe_axis_size - 1) / float(params.grid_size.x); + probe_pos /= probe_cell_size_next; ivec3 probe_posi = ivec3(probe_pos); //add up all light, no need to use occlusion here, since occlusion will do its work afterwards @@ -574,20 +561,28 @@ void main() { } } else { - // clear and let it re-raytrace, only for the last cascade, which happens very un-often - //scroll + //scroll at the edge of the highest cascade, just copy what is there, + //since its the closest we have anyway + for (uint j = 0; j < params.history_size; j++) { + ivec2 tex_pos; + tex_pos = probe_cell.xy; + tex_pos.x += probe_cell.z * int(params.probe_axis_size); + for (int i = 0; i < SH_SIZE; i++) { // copy from history texture + ivec3 src_pos = ivec3(tex_pos.x, tex_pos.y * SH_SIZE + i, int(j)); ivec3 dst_pos = ivec3(pos.x, pos.y * SH_SIZE + i, int(j)); - imageStore(lightprobe_history_scroll_texture, dst_pos, ivec4(0)); + ivec4 value = imageLoad(lightprobe_history_texture, dst_pos); + imageStore(lightprobe_history_scroll_texture, dst_pos, value); } } for (int i = 0; i < SH_SIZE; i++) { // copy from average texture - ivec2 dst_pos = ivec2(pos.x, pos.y * SH_SIZE + i); - imageStore(lightprobe_average_scroll_texture, dst_pos, ivec4(0)); + ivec2 spos = ivec2(pos.x, pos.y * SH_SIZE + i); + ivec4 average = imageLoad(lightprobe_average_texture, spos); + imageStore(lightprobe_average_scroll_texture, spos, average); } } diff --git a/servers/rendering/renderer_rd/shaders/shadow_reduce.glsl b/servers/rendering/renderer_rd/shaders/shadow_reduce.glsl deleted file mode 100644 index 29443ae7db..0000000000 --- a/servers/rendering/renderer_rd/shaders/shadow_reduce.glsl +++ /dev/null @@ -1,105 +0,0 @@ -#[compute] - -#version 450 - -VERSION_DEFINES - -#define BLOCK_SIZE 8 - -layout(local_size_x = BLOCK_SIZE, local_size_y = BLOCK_SIZE, local_size_z = 1) in; - -#ifdef MODE_REDUCE - -shared float tmp_data[BLOCK_SIZE * BLOCK_SIZE]; -const uint swizzle_table[BLOCK_SIZE] = uint[](0, 4, 2, 6, 1, 5, 3, 7); -const uint unswizzle_table[BLOCK_SIZE] = uint[](0, 0, 0, 1, 0, 2, 1, 3); - -#endif - -layout(r32f, set = 0, binding = 0) uniform restrict readonly image2D source_depth; -layout(r32f, set = 0, binding = 1) uniform restrict writeonly image2D dst_depth; - -layout(push_constant, binding = 1, std430) uniform Params { - ivec2 source_size; - ivec2 source_offset; - uint min_size; - uint gaussian_kernel_version; - ivec2 filter_dir; -} -params; - -void main() { -#ifdef MODE_REDUCE - - uvec2 pos = gl_LocalInvocationID.xy; - - ivec2 image_offset = params.source_offset; - ivec2 image_pos = image_offset + ivec2(gl_GlobalInvocationID.xy); - uint dst_t = swizzle_table[pos.y] * BLOCK_SIZE + swizzle_table[pos.x]; - tmp_data[dst_t] = imageLoad(source_depth, min(image_pos, params.source_size - ivec2(1))).r; - ivec2 image_size = params.source_size; - - uint t = pos.y * BLOCK_SIZE + pos.x; - - //neighbours - uint size = BLOCK_SIZE; - - do { - groupMemoryBarrier(); - barrier(); - - size >>= 1; - image_size >>= 1; - image_offset >>= 1; - - if (all(lessThan(pos, uvec2(size)))) { - uint nx = t + size; - uint ny = t + (BLOCK_SIZE * size); - uint nxy = ny + size; - - tmp_data[t] += tmp_data[nx]; - tmp_data[t] += tmp_data[ny]; - tmp_data[t] += tmp_data[nxy]; - tmp_data[t] /= 4.0; - } - - } while (size > params.min_size); - - if (all(lessThan(pos, uvec2(size)))) { - image_pos = ivec2(unswizzle_table[size + pos.x], unswizzle_table[size + pos.y]); - image_pos += image_offset + ivec2(gl_WorkGroupID.xy) * int(size); - - image_size = max(ivec2(1), image_size); //in case image size became 0 - - if (all(lessThan(image_pos, uvec2(image_size)))) { - imageStore(dst_depth, image_pos, vec4(tmp_data[t])); - } - } -#endif - -#ifdef MODE_FILTER - - ivec2 image_pos = params.source_offset + ivec2(gl_GlobalInvocationID.xy); - if (any(greaterThanEqual(image_pos, params.source_size))) { - return; - } - - ivec2 clamp_min = ivec2(params.source_offset); - ivec2 clamp_max = ivec2(params.source_size) - 1; - - //gaussian kernel, size 9, sigma 4 - const int kernel_size = 9; - const float gaussian_kernel[kernel_size * 3] = float[]( - 0.000229, 0.005977, 0.060598, 0.241732, 0.382928, 0.241732, 0.060598, 0.005977, 0.000229, - 0.028532, 0.067234, 0.124009, 0.179044, 0.20236, 0.179044, 0.124009, 0.067234, 0.028532, - 0.081812, 0.101701, 0.118804, 0.130417, 0.134535, 0.130417, 0.118804, 0.101701, 0.081812); - float accum = 0.0; - for (int i = 0; i < kernel_size; i++) { - ivec2 ofs = clamp(image_pos + params.filter_dir * (i - kernel_size / 2), clamp_min, clamp_max); - accum += imageLoad(source_depth, ofs).r * gaussian_kernel[params.gaussian_kernel_version + i]; - } - - imageStore(dst_depth, image_pos, vec4(accum)); - -#endif -} diff --git a/servers/rendering/renderer_rd/shaders/skeleton.glsl b/servers/rendering/renderer_rd/shaders/skeleton.glsl index b19f5a9ad3..680d1045cd 100644 --- a/servers/rendering/renderer_rd/shaders/skeleton.glsl +++ b/servers/rendering/renderer_rd/shaders/skeleton.glsl @@ -100,7 +100,7 @@ void main() { for (uint i = 0; i < params.blend_shape_count; i++) { float w = blend_shape_weights.data[i]; - if (w > 0.0001) { + if (abs(w) > 0.0001) { uint base_offset = (params.vertex_count * i + index) * params.vertex_stride; blend_vertex += uintBitsToFloat(uvec3(src_blend_shapes.data[base_offset + 0], src_blend_shapes.data[base_offset + 1], src_blend_shapes.data[base_offset + 2])) * w; diff --git a/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl b/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl index 498a6ddb5b..ce8a459b24 100644 --- a/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl +++ b/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl @@ -4,6 +4,15 @@ VERSION_DEFINES +/* Do not use subgroups here, seems there is not much advantage and causes glitches +#if defined(has_GL_KHR_shader_subgroup_ballot) && defined(has_GL_KHR_shader_subgroup_arithmetic) +#extension GL_KHR_shader_subgroup_ballot: enable +#extension GL_KHR_shader_subgroup_arithmetic: enable + +#define USE_SUBGROUPS +#endif +*/ + #if defined(MODE_FOG) || defined(MODE_FILTER) layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; @@ -23,22 +32,25 @@ layout(local_size_x = 4, local_size_y = 4, local_size_z = 4) in; layout(set = 0, binding = 1) uniform texture2D shadow_atlas; layout(set = 0, binding = 2) uniform texture2D directional_shadow_atlas; -layout(set = 0, binding = 3, std430) restrict readonly buffer Lights { +layout(set = 0, binding = 3, std430) restrict readonly buffer OmniLights { LightData data[]; } -lights; +omni_lights; -layout(set = 0, binding = 4, std140) uniform DirectionalLights { +layout(set = 0, binding = 4, std430) restrict readonly buffer SpotLights { + LightData data[]; +} +spot_lights; + +layout(set = 0, binding = 5, std140) uniform DirectionalLights { DirectionalLightData data[MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS]; } directional_lights; -layout(set = 0, binding = 5) uniform utexture3D cluster_texture; - -layout(set = 0, binding = 6, std430) restrict readonly buffer ClusterData { - uint indices[]; +layout(set = 0, binding = 6, std430) buffer restrict readonly ClusterBuffer { + uint data[]; } -cluster_data; +cluster_buffer; layout(set = 0, binding = 7) uniform sampler linear_sampler; @@ -132,7 +144,7 @@ layout(set = 1, binding = 2) uniform texture3D sdfgi_occlusion_texture; #endif //SDFGI -layout(push_constant, binding = 0, std430) uniform Params { +layout(set = 0, binding = 14, std140) uniform Params { vec2 fog_frustum_size_begin; vec2 fog_frustum_size_end; @@ -150,12 +162,24 @@ layout(push_constant, binding = 0, std430) uniform Params { float detail_spread; float gi_inject; uint max_gi_probes; - uint pad; + uint cluster_type_size; + + vec2 screen_size; + uint cluster_shift; + uint cluster_width; + + uint max_cluster_element_count_div_32; + bool use_temporal_reprojection; + uint temporal_frame; + float temporal_blend; mat3x4 cam_rotation; + mat4 to_prev_view; } params; +layout(set = 0, binding = 15) uniform texture3D prev_density_texture; + float get_depth_at_pos(float cell_depth_size, int z) { float d = float(z) * cell_depth_size + cell_depth_size * 0.5; //center of voxels d = pow(d, params.detail_spread); @@ -178,6 +202,42 @@ float get_omni_attenuation(float distance, float inv_range, float decay) { return nd * pow(max(distance, 0.0001), -decay); } +void cluster_get_item_range(uint p_offset, out uint item_min, out uint item_max, out uint item_from, out uint item_to) { + uint item_min_max = cluster_buffer.data[p_offset]; + item_min = item_min_max & 0xFFFF; + item_max = item_min_max >> 16; + ; + + item_from = item_min >> 5; + item_to = (item_max == 0) ? 0 : ((item_max - 1) >> 5) + 1; //side effect of how it is stored, as item_max 0 means no elements +} + +uint cluster_get_range_clip_mask(uint i, uint z_min, uint z_max) { + int local_min = clamp(int(z_min) - int(i) * 32, 0, 31); + int mask_width = min(int(z_max) - int(z_min), 32 - local_min); + return bitfieldInsert(uint(0), uint(0xFFFFFFFF), local_min, mask_width); +} + +#define TEMPORAL_FRAMES 16 + +const vec3 halton_map[TEMPORAL_FRAMES] = vec3[]( + vec3(0.5, 0.33333333, 0.2), + vec3(0.25, 0.66666667, 0.4), + vec3(0.75, 0.11111111, 0.6), + vec3(0.125, 0.44444444, 0.8), + vec3(0.625, 0.77777778, 0.04), + vec3(0.375, 0.22222222, 0.24), + vec3(0.875, 0.55555556, 0.44), + vec3(0.0625, 0.88888889, 0.64), + vec3(0.5625, 0.03703704, 0.84), + vec3(0.3125, 0.37037037, 0.08), + vec3(0.8125, 0.7037037, 0.28), + vec3(0.1875, 0.14814815, 0.48), + vec3(0.6875, 0.48148148, 0.68), + vec3(0.4375, 0.81481481, 0.88), + vec3(0.9375, 0.25925926, 0.12), + vec3(0.03125, 0.59259259, 0.32)); + void main() { vec3 fog_cell_size = 1.0 / vec3(params.fog_volume_size); @@ -193,6 +253,12 @@ void main() { //posf += mix(vec3(0.0),vec3(1.0),0.3) * hash3f(uvec3(pos)) * 2.0 - 1.0; vec3 fog_unit_pos = posf * fog_cell_size + fog_cell_size * 0.5; //center of voxels + + uvec2 screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); + uvec2 cluster_pos = screen_pos >> params.cluster_shift; + uint cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); + //positions in screen are too spread apart, no hopes for optimizing with subgroups + fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); vec3 view_pos; @@ -200,6 +266,47 @@ void main() { view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; view_pos.y = -view_pos.y; + vec4 reprojected_density = vec4(0.0); + float reproject_amount = 0.0; + + if (params.use_temporal_reprojection) { + vec3 prev_view = (params.to_prev_view * vec4(view_pos, 1.0)).xyz; + //undo transform into prev view + prev_view.y = -prev_view.y; + //z back to unit size + prev_view.z /= -params.fog_frustum_end; + //xy back to unit size + prev_view.xy /= mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(prev_view.z)); + prev_view.xy = prev_view.xy * 0.5 + 0.5; + //z back to unspread value + prev_view.z = pow(prev_view.z, 1.0 / params.detail_spread); + + if (all(greaterThan(prev_view, vec3(0.0))) && all(lessThan(prev_view, vec3(1.0)))) { + //reprojectinon fits + + reprojected_density = textureLod(sampler3D(prev_density_texture, linear_sampler), prev_view, 0.0); + reproject_amount = params.temporal_blend; + + // Since we can reproject, now we must jitter the current view pos. + // This is done here because cells that can't reproject should not jitter. + + fog_unit_pos = posf * fog_cell_size + fog_cell_size * halton_map[params.temporal_frame]; //center of voxels, offset by halton table + + screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); + cluster_pos = screen_pos >> params.cluster_shift; + cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); + //positions in screen are too spread apart, no hopes for optimizing with subgroups + + fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); + + view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(fog_unit_pos.z)); + view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; + view_pos.y = -view_pos.y; + } + } + + uint cluster_z = uint(clamp((abs(view_pos.z) / params.z_far) * 32.0, 0.0, 31.0)); + vec3 total_light = params.light_color; float total_density = params.base_density; @@ -266,108 +373,160 @@ void main() { //compute lights from cluster - vec3 cluster_pos; - cluster_pos.xy = fog_unit_pos.xy; - cluster_pos.z = clamp((abs(view_pos.z) - params.z_near) / (params.z_far - params.z_near), 0.0, 1.0); + { //omni lights - uvec4 cluster_cell = texture(usampler3D(cluster_texture, linear_sampler), cluster_pos); + uint cluster_omni_offset = cluster_offset; - uint omni_light_count = cluster_cell.x >> CLUSTER_COUNTER_SHIFT; - uint omni_light_pointer = cluster_cell.x & CLUSTER_POINTER_MASK; + uint item_min; + uint item_max; + uint item_from; + uint item_to; - for (uint i = 0; i < omni_light_count; i++) { - uint light_index = cluster_data.indices[omni_light_pointer + i]; + cluster_get_item_range(cluster_omni_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); - vec3 light_pos = lights.data[i].position; - float d = distance(lights.data[i].position, view_pos); - vec3 shadow_attenuation = vec3(1.0); +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_omni_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + uint light_index = 32 * i + bit; - if (d * lights.data[i].inv_radius < 1.0) { - vec2 attenuation_energy = unpackHalf2x16(lights.data[i].attenuation_energy); - vec4 color_specular = unpackUnorm4x8(lights.data[i].color_specular); + //if (!bool(omni_omni_lights.data[light_index].mask & draw_call.layer_mask)) { + // continue; //not masked + //} - float attenuation = get_omni_attenuation(d, lights.data[i].inv_radius, attenuation_energy.x); + vec3 light_pos = omni_lights.data[light_index].position; + float d = distance(omni_lights.data[light_index].position, view_pos); + float shadow_attenuation = 1.0; - vec3 light = attenuation_energy.y * color_specular.rgb / M_PI; + if (d * omni_lights.data[light_index].inv_radius < 1.0) { + float attenuation = get_omni_attenuation(d, omni_lights.data[light_index].inv_radius, omni_lights.data[light_index].attenuation); - vec4 shadow_color_enabled = unpackUnorm4x8(lights.data[i].shadow_color_enabled); + vec3 light = omni_lights.data[light_index].color / M_PI; - if (shadow_color_enabled.a > 0.5) { - //has shadow - vec4 v = vec4(view_pos, 1.0); + if (omni_lights.data[light_index].shadow_enabled) { + //has shadow + vec4 v = vec4(view_pos, 1.0); - vec4 splane = (lights.data[i].shadow_matrix * v); - float shadow_len = length(splane.xyz); //need to remember shadow len from here + vec4 splane = (omni_lights.data[light_index].shadow_matrix * v); + float shadow_len = length(splane.xyz); //need to remember shadow len from here - splane.xyz = normalize(splane.xyz); - vec4 clamp_rect = lights.data[i].atlas_rect; + splane.xyz = normalize(splane.xyz); + vec4 clamp_rect = omni_lights.data[light_index].atlas_rect; - if (splane.z >= 0.0) { - splane.z += 1.0; + if (splane.z >= 0.0) { + splane.z += 1.0; - clamp_rect.y += clamp_rect.w; + clamp_rect.y += clamp_rect.w; - } else { - splane.z = 1.0 - splane.z; - } + } else { + splane.z = 1.0 - splane.z; + } - splane.xy /= splane.z; + splane.xy /= splane.z; - splane.xy = splane.xy * 0.5 + 0.5; - splane.z = shadow_len * lights.data[i].inv_radius; - splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; - splane.w = 1.0; //needed? i think it should be 1 already + splane.xy = splane.xy * 0.5 + 0.5; + splane.z = shadow_len * omni_lights.data[light_index].inv_radius; + splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; + splane.w = 1.0; //needed? i think it should be 1 already - float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; - float shadow = exp(min(0.0, (depth - splane.z)) / lights.data[i].inv_radius * lights.data[i].shadow_volumetric_fog_fade); + float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; - shadow_attenuation = mix(shadow_color_enabled.rgb, vec3(1.0), shadow); + shadow_attenuation = exp(min(0.0, (depth - splane.z)) / omni_lights.data[light_index].inv_radius * omni_lights.data[light_index].shadow_volumetric_fog_fade); + } + total_light += light * attenuation * shadow_attenuation; + } } - total_light += light * attenuation * shadow_attenuation; } } - uint spot_light_count = cluster_cell.y >> CLUSTER_COUNTER_SHIFT; - uint spot_light_pointer = cluster_cell.y & CLUSTER_POINTER_MASK; + { //spot lights - for (uint i = 0; i < spot_light_count; i++) { - uint light_index = cluster_data.indices[spot_light_pointer + i]; + uint cluster_spot_offset = cluster_offset + params.cluster_type_size; - vec3 light_pos = lights.data[i].position; - vec3 light_rel_vec = lights.data[i].position - view_pos; - float d = length(light_rel_vec); - vec3 shadow_attenuation = vec3(1.0); + uint item_min; + uint item_max; + uint item_from; + uint item_to; + + cluster_get_item_range(cluster_spot_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); - if (d * lights.data[i].inv_radius < 1.0) { - vec2 attenuation_energy = unpackHalf2x16(lights.data[i].attenuation_energy); - vec4 color_specular = unpackUnorm4x8(lights.data[i].color_specular); +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif - float attenuation = get_omni_attenuation(d, lights.data[i].inv_radius, attenuation_energy.x); + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_spot_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif - vec3 spot_dir = lights.data[i].direction; - vec2 spot_att_angle = unpackHalf2x16(lights.data[i].cone_attenuation_angle); - float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_att_angle.y); - float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_att_angle.y)); - attenuation *= 1.0 - pow(spot_rim, spot_att_angle.x); + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif - vec3 light = attenuation_energy.y * color_specular.rgb / M_PI; + //if (!bool(omni_lights.data[light_index].mask & draw_call.layer_mask)) { + // continue; //not masked + //} - vec4 shadow_color_enabled = unpackUnorm4x8(lights.data[i].shadow_color_enabled); + uint light_index = 32 * i + bit; - if (shadow_color_enabled.a > 0.5) { - //has shadow - vec4 v = vec4(view_pos, 1.0); + vec3 light_pos = spot_lights.data[light_index].position; + vec3 light_rel_vec = spot_lights.data[light_index].position - view_pos; + float d = length(light_rel_vec); + float shadow_attenuation = 1.0; - vec4 splane = (lights.data[i].shadow_matrix * v); - splane /= splane.w; + if (d * spot_lights.data[light_index].inv_radius < 1.0) { + float attenuation = get_omni_attenuation(d, spot_lights.data[light_index].inv_radius, spot_lights.data[light_index].attenuation); - float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; - float shadow = exp(min(0.0, (depth - splane.z)) / lights.data[i].inv_radius * lights.data[i].shadow_volumetric_fog_fade); + vec3 spot_dir = spot_lights.data[light_index].direction; + float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[light_index].cone_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[light_index].cone_angle)); + attenuation *= 1.0 - pow(spot_rim, spot_lights.data[light_index].cone_attenuation); - shadow_attenuation = mix(shadow_color_enabled.rgb, vec3(1.0), shadow); - } + vec3 light = spot_lights.data[light_index].color / M_PI; + + if (spot_lights.data[light_index].shadow_enabled) { + //has shadow + vec4 v = vec4(view_pos, 1.0); + + vec4 splane = (spot_lights.data[light_index].shadow_matrix * v); + splane /= splane.w; - total_light += light * attenuation * shadow_attenuation; + float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; + + shadow_attenuation = exp(min(0.0, (depth - splane.z)) / spot_lights.data[light_index].inv_radius * spot_lights.data[light_index].shadow_volumetric_fog_fade); + } + + total_light += light * attenuation * shadow_attenuation; + } + } } } @@ -470,7 +629,11 @@ void main() { #endif - imageStore(density_map, pos, vec4(total_light, total_density)); + vec4 final_density = vec4(total_light, total_density); + + final_density = mix(final_density, reprojected_density, reproject_amount); + + imageStore(density_map, pos, final_density); #endif #ifdef MODE_FOG diff --git a/servers/rendering/renderer_scene.h b/servers/rendering/renderer_scene.h index c483898fed..b546001843 100644 --- a/servers/rendering/renderer_scene.h +++ b/servers/rendering/renderer_scene.h @@ -36,7 +36,8 @@ class RendererScene { public: - virtual RID camera_create() = 0; + virtual RID camera_allocate() = 0; + virtual void camera_initialize(RID p_rid) = 0; virtual void camera_set_perspective(RID p_camera, float p_fovy_degrees, float p_z_near, float p_z_far) = 0; virtual void camera_set_orthogonal(RID p_camera, float p_size, float p_z_near, float p_z_far) = 0; @@ -48,7 +49,8 @@ public: virtual void camera_set_use_vertical_aspect(RID p_camera, bool p_enable) = 0; virtual bool is_camera(RID p_camera) const = 0; - virtual RID scenario_create() = 0; + virtual RID scenario_allocate() = 0; + virtual void scenario_initialize(RID p_rid) = 0; virtual void scenario_set_debug(RID p_scenario, RS::ScenarioDebugMode p_debug_mode) = 0; virtual void scenario_set_environment(RID p_scenario, RID p_environment) = 0; @@ -58,7 +60,8 @@ public: virtual bool is_scenario(RID p_scenario) const = 0; virtual RID scenario_get_environment(RID p_scenario) = 0; - virtual RID instance_create() = 0; + virtual RID instance_allocate() = 0; + virtual void instance_initialize(RID p_rid) = 0; virtual void instance_set_base(RID p_instance, RID p_base) = 0; virtual void instance_set_scenario(RID p_instance, RID p_scenario) = 0; @@ -95,11 +98,13 @@ public: virtual Variant instance_geometry_get_shader_parameter(RID p_instance, const StringName &p_parameter) const = 0; virtual Variant instance_geometry_get_shader_parameter_default_value(RID p_instance, const StringName &p_parameter) const = 0; - virtual void directional_shadow_atlas_set_size(int p_size) = 0; + virtual void directional_shadow_atlas_set_size(int p_size, bool p_16_bits = false) = 0; /* SKY API */ - virtual RID sky_create() = 0; + virtual RID sky_allocate() = 0; + virtual void sky_initialize(RID p_rid) = 0; + virtual void sky_set_radiance_size(RID p_sky, int p_radiance_size) = 0; virtual void sky_set_mode(RID p_sky, RS::SkyMode p_samples) = 0; virtual void sky_set_material(RID p_sky, RID p_material) = 0; @@ -107,7 +112,8 @@ public: /* ENVIRONMENT API */ - virtual RID environment_create() = 0; + virtual RID environment_allocate() = 0; + virtual void environment_initialize(RID p_rid) = 0; virtual void environment_set_background(RID p_env, RS::EnvironmentBG p_bg) = 0; virtual void environment_set_sky(RID p_env, RID p_sky) = 0; @@ -122,12 +128,10 @@ public: virtual void environment_glow_set_use_bicubic_upscale(bool p_enable) = 0; virtual void environment_glow_set_use_high_quality(bool p_enable) = 0; - virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, RS::EnvVolumetricFogShadowFilter p_shadow_filter) = 0; + virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) = 0; virtual void environment_set_volumetric_fog_volume_size(int p_size, int p_depth) = 0; virtual void environment_set_volumetric_fog_filter_active(bool p_enable) = 0; - virtual void environment_set_volumetric_fog_directional_shadow_shrink_size(int p_shrink_size) = 0; - virtual void environment_set_volumetric_fog_positional_shadow_shrink_size(int p_shrink_size) = 0; virtual void environment_set_ssr(RID p_env, bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance) = 0; virtual void environment_set_ssr_roughness_quality(RS::EnvironmentSSRRoughnessQuality p_quality) = 0; @@ -136,10 +140,11 @@ public: virtual void environment_set_ssao_quality(RS::EnvironmentSSAOQuality p_quality, bool p_half_size, float p_adaptive_target, int p_blur_passes, float p_fadeout_from, float p_fadeout_to) = 0; - virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, bool p_use_multibounce, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; + virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; virtual void environment_set_sdfgi_ray_count(RS::EnvironmentSDFGIRayCount p_ray_count) = 0; virtual void environment_set_sdfgi_frames_to_converge(RS::EnvironmentSDFGIFramesToConverge p_frames) = 0; + virtual void environment_set_sdfgi_frames_to_update_light(RS::EnvironmentSDFGIFramesToUpdateLight p_update) = 0; virtual void environment_set_tonemap(RID p_env, RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) = 0; @@ -160,7 +165,8 @@ public: /* Camera Effects */ - virtual RID camera_effects_create() = 0; + virtual RID camera_effects_allocate() = 0; + virtual void camera_effects_initialize(RID p_rid) = 0; virtual void camera_effects_set_dof_blur_quality(RS::DOFBlurQuality p_quality, bool p_use_jitter) = 0; virtual void camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokehShape p_shape) = 0; @@ -172,14 +178,17 @@ public: virtual void directional_shadow_quality_set(RS::ShadowQuality p_quality) = 0; virtual RID shadow_atlas_create() = 0; - virtual void shadow_atlas_set_size(RID p_atlas, int p_size) = 0; + virtual void shadow_atlas_set_size(RID p_atlas, int p_size, bool p_use_16_bits = false) = 0; virtual void shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision) = 0; /* Render Buffers */ virtual RID render_buffers_create() = 0; + virtual void render_buffers_configure(RID p_render_buffers, RID p_render_target, int p_width, int p_height, RS::ViewportMSAA p_msaa, RS::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_debanding) = 0; + virtual void gi_set_use_half_resolution(bool p_enable) = 0; + virtual void set_debug_draw_mode(RS::ViewportDebugDraw p_debug_draw) = 0; virtual TypedArray<Image> bake_render_uv2(RID p_base, const Vector<RID> &p_material_overrides, const Size2i &p_image_size) = 0; diff --git a/servers/rendering/renderer_scene_cull.cpp b/servers/rendering/renderer_scene_cull.cpp index d3979521b1..e8155e4025 100644 --- a/servers/rendering/renderer_scene_cull.cpp +++ b/servers/rendering/renderer_scene_cull.cpp @@ -39,9 +39,11 @@ /* CAMERA API */ -RID RendererSceneCull::camera_create() { - Camera *camera = memnew(Camera); - return camera_owner.make_rid(camera); +RID RendererSceneCull::camera_allocate() { + return camera_owner.allocate_rid(); +} +void RendererSceneCull::camera_initialize(RID p_rid) { + camera_owner.initialize_rid(p_rid, memnew(Camera)); } void RendererSceneCull::camera_set_perspective(RID p_camera, float p_fovy_degrees, float p_z_near, float p_z_far) { @@ -290,11 +292,12 @@ void RendererSceneCull::_instance_unpair(Instance *p_A, Instance *p_B) { } } -RID RendererSceneCull::scenario_create() { +RID RendererSceneCull::scenario_allocate() { + return scenario_owner.allocate_rid(); +} +void RendererSceneCull::scenario_initialize(RID p_rid) { Scenario *scenario = memnew(Scenario); - ERR_FAIL_COND_V(!scenario, RID()); - RID scenario_rid = scenario_owner.make_rid(scenario); - scenario->self = scenario_rid; + scenario->self = p_rid; scenario->reflection_probe_shadow_atlas = scene_render->shadow_atlas_create(); scene_render->shadow_atlas_set_size(scenario->reflection_probe_shadow_atlas, 1024); //make enough shadows for close distance, don't bother with rest @@ -307,7 +310,7 @@ RID RendererSceneCull::scenario_create() { scenario->instance_aabbs.set_page_pool(&instance_aabb_page_pool); scenario->instance_data.set_page_pool(&instance_data_page_pool); - return scenario_rid; + scenario_owner.initialize_rid(p_rid, scenario); } void RendererSceneCull::scenario_set_debug(RID p_scenario, RS::ScenarioDebugMode p_debug_mode) { @@ -367,14 +370,14 @@ void RendererSceneCull::_instance_queue_update(Instance *p_instance, bool p_upda _instance_update_list.add(&p_instance->update_item); } -RID RendererSceneCull::instance_create() { +RID RendererSceneCull::instance_allocate() { + return instance_owner.allocate_rid(); +} +void RendererSceneCull::instance_initialize(RID p_rid) { Instance *instance = memnew(Instance); - ERR_FAIL_COND_V(!instance, RID()); + instance->self = p_rid; - RID instance_rid = instance_owner.make_rid(instance); - instance->self = instance_rid; - - return instance_rid; + instance_owner.initialize_rid(p_rid, instance); } void RendererSceneCull::_instance_update_mesh_instance(Instance *p_instance) { @@ -436,7 +439,7 @@ void RendererSceneCull::instance_set_base(RID p_instance, RID p_base) { case RS::INSTANCE_LIGHT: { InstanceLightData *light = static_cast<InstanceLightData *>(instance->base_data); - if (scenario && RSG::storage->light_get_type(instance->base) != RS::LIGHT_DIRECTIONAL && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { + if (scenario && instance->visible && RSG::storage->light_get_type(instance->base) != RS::LIGHT_DIRECTIONAL && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { scenario->dynamic_lights.erase(light->instance); } @@ -783,6 +786,17 @@ void RendererSceneCull::instance_set_visible(RID p_instance, bool p_visible) { _unpair_instance(instance); } + if (instance->base_type == RS::INSTANCE_LIGHT) { + InstanceLightData *light = static_cast<InstanceLightData *>(instance->base_data); + if (instance->scenario && RSG::storage->light_get_type(instance->base) != RS::LIGHT_DIRECTIONAL && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { + if (p_visible) { + instance->scenario->dynamic_lights.push_back(light->instance); + } else { + instance->scenario->dynamic_lights.erase(light->instance); + } + } + } + if (instance->base_type == RS::INSTANCE_PARTICLES_COLLISION) { InstanceParticlesCollisionData *collision = static_cast<InstanceParticlesCollisionData *>(instance->base_data); RSG::storage->particles_collision_instance_set_active(collision->instance, p_visible); @@ -1150,13 +1164,13 @@ void RendererSceneCull::_update_instance(Instance *p_instance) { RS::LightBakeMode bake_mode = RSG::storage->light_get_bake_mode(p_instance->base); if (RSG::storage->light_get_type(p_instance->base) != RS::LIGHT_DIRECTIONAL && bake_mode != light->bake_mode) { - if (p_instance->scenario && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { + if (p_instance->visible && p_instance->scenario && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { p_instance->scenario->dynamic_lights.erase(light->instance); } light->bake_mode = bake_mode; - if (p_instance->scenario && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { + if (p_instance->visible && p_instance->scenario && light->bake_mode == RS::LIGHT_BAKE_DYNAMIC) { p_instance->scenario->dynamic_lights.push_back(light->instance); } } @@ -1242,7 +1256,8 @@ void RendererSceneCull::_update_instance(Instance *p_instance) { scene_render->geometry_instance_set_transform(geom->geometry_instance, p_instance->transform, p_instance->aabb, p_instance->transformed_aabb); } - if (p_instance->scenario == nullptr || !p_instance->visible || Math::is_zero_approx(p_instance->transform.basis.determinant())) { + // note: we had to remove is equal approx check here, it meant that det == 0.000004 won't work, which is the case for some of our scenes. + if (p_instance->scenario == nullptr || !p_instance->visible || p_instance->transform.basis.determinant() == 0) { p_instance->prev_transformed_aabb = p_instance->transformed_aabb; return; } @@ -1894,6 +1909,9 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons RS::LightOmniShadowMode shadow_mode = RSG::storage->light_omni_get_shadow_mode(p_instance->base); if (shadow_mode == RS::LIGHT_OMNI_SHADOW_DUAL_PARABOLOID || !scene_render->light_instances_can_render_shadow_cube()) { + if (max_shadows_used + 2 > MAX_UPDATE_SHADOWS) { + return true; + } for (int i = 0; i < 2; i++) { //using this one ensures that raster deferred will have it RENDER_TIMESTAMP("Culling Shadow Paraboloid" + itos(i)); @@ -1910,7 +1928,6 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons planes.write[4] = light_transform.xform(Plane(Vector3(0, -1, z).normalized(), radius)); planes.write[5] = light_transform.xform(Plane(Vector3(0, 0, -z), 0)); - geometry_instances_to_shadow_render.clear(); instance_shadow_cull_result.clear(); Vector<Vector3> points = Geometry3D::compute_convex_mesh_points(&planes[0], planes.size()); @@ -1931,6 +1948,8 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons Plane near_plane(light_transform.origin, light_transform.basis.get_axis(2) * z); + RendererSceneRender::RenderShadowData &shadow_data = render_shadow_data[max_shadows_used++]; + for (int j = 0; j < (int)instance_shadow_cull_result.size(); j++) { Instance *instance = instance_shadow_cull_result[j]; if (!instance->visible || !((1 << instance->base_type) & RS::INSTANCE_GEOMETRY_MASK) || !static_cast<InstanceGeometryData *>(instance->base_data)->can_cast_shadows) { @@ -1945,16 +1964,21 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons } } - geometry_instances_to_shadow_render.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); + shadow_data.instances.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); } RSG::storage->update_mesh_instances(); scene_render->light_instance_set_shadow_transform(light->instance, CameraMatrix(), light_transform, radius, 0, i, 0); - scene_render->render_shadow(light->instance, p_shadow_atlas, i, geometry_instances_to_shadow_render); + shadow_data.light = light->instance; + shadow_data.pass = i; } } else { //shadow cube + if (max_shadows_used + 6 > MAX_UPDATE_SHADOWS) { + return true; + } + real_t radius = RSG::storage->light_get_param(p_instance->base, RS::LIGHT_PARAM_RANGE); CameraMatrix cm; cm.set_perspective(90, 1, 0.01, radius); @@ -1984,7 +2008,6 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons Vector<Plane> planes = cm.get_projection_planes(xform); - geometry_instances_to_shadow_render.clear(); instance_shadow_cull_result.clear(); Vector<Vector3> points = Geometry3D::compute_convex_mesh_points(&planes[0], planes.size()); @@ -2003,7 +2026,7 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons p_scenario->indexers[Scenario::INDEXER_GEOMETRY].convex_query(planes.ptr(), planes.size(), points.ptr(), points.size(), cull_convex); - Plane near_plane(xform.origin, -xform.basis.get_axis(2)); + RendererSceneRender::RenderShadowData &shadow_data = render_shadow_data[max_shadows_used++]; for (int j = 0; j < (int)instance_shadow_cull_result.size(); j++) { Instance *instance = instance_shadow_cull_result[j]; @@ -2018,22 +2041,28 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons } } - geometry_instances_to_shadow_render.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); + shadow_data.instances.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); } RSG::storage->update_mesh_instances(); scene_render->light_instance_set_shadow_transform(light->instance, cm, xform, radius, 0, i, 0); - scene_render->render_shadow(light->instance, p_shadow_atlas, i, geometry_instances_to_shadow_render); + + shadow_data.light = light->instance; + shadow_data.pass = i; } //restore the regular DP matrix - scene_render->light_instance_set_shadow_transform(light->instance, CameraMatrix(), light_transform, radius, 0, 0, 0); + //scene_render->light_instance_set_shadow_transform(light->instance, CameraMatrix(), light_transform, radius, 0, 0, 0); } } break; case RS::LIGHT_SPOT: { RENDER_TIMESTAMP("Culling Spot Light"); + if (max_shadows_used + 1 > MAX_UPDATE_SHADOWS) { + return true; + } + real_t radius = RSG::storage->light_get_param(p_instance->base, RS::LIGHT_PARAM_RANGE); real_t angle = RSG::storage->light_get_param(p_instance->base, RS::LIGHT_PARAM_SPOT_ANGLE); @@ -2042,7 +2071,6 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons Vector<Plane> planes = cm.get_projection_planes(light_transform); - geometry_instances_to_shadow_render.clear(); instance_shadow_cull_result.clear(); Vector<Vector3> points = Geometry3D::compute_convex_mesh_points(&planes[0], planes.size()); @@ -2061,7 +2089,7 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons p_scenario->indexers[Scenario::INDEXER_GEOMETRY].convex_query(planes.ptr(), planes.size(), points.ptr(), points.size(), cull_convex); - Plane near_plane(light_transform.origin, -light_transform.basis.get_axis(2)); + RendererSceneRender::RenderShadowData &shadow_data = render_shadow_data[max_shadows_used++]; for (int j = 0; j < (int)instance_shadow_cull_result.size(); j++) { Instance *instance = instance_shadow_cull_result[j]; @@ -2076,13 +2104,14 @@ bool RendererSceneCull::_light_instance_update_shadow(Instance *p_instance, cons RSG::storage->mesh_instance_check_for_update(instance->mesh_instance); } } - geometry_instances_to_shadow_render.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); + shadow_data.instances.push_back(static_cast<InstanceGeometryData *>(instance->base_data)->geometry_instance); } RSG::storage->update_mesh_instances(); scene_render->light_instance_set_shadow_transform(light->instance, cm, light_transform, radius, 0, 0, 0); - scene_render->render_shadow(light->instance, p_shadow_atlas, 0, geometry_instances_to_shadow_render); + shadow_data.light = light->instance; + shadow_data.pass = 0; } break; } @@ -2135,14 +2164,13 @@ void RendererSceneCull::render_camera(RID p_render_buffers, RID p_camera, RID p_ RID environment = _render_get_environment(p_camera, p_scenario); - _prepare_scene(camera->transform, camera_matrix, ortho, camera->vaspect, p_render_buffers, environment, camera->visible_layers, p_scenario, p_shadow_atlas, RID(), p_screen_lod_threshold); - _render_scene(p_render_buffers, camera->transform, camera_matrix, ortho, environment, camera->effects, p_scenario, p_shadow_atlas, RID(), -1, p_screen_lod_threshold); + _render_scene(camera->transform, camera_matrix, ortho, camera->vaspect, p_render_buffers, environment, camera->effects, camera->visible_layers, p_scenario, p_shadow_atlas, RID(), -1, p_screen_lod_threshold); #endif } void RendererSceneCull::render_camera(RID p_render_buffers, Ref<XRInterface> &p_interface, XRInterface::Eyes p_eye, RID p_camera, RID p_scenario, Size2 p_viewport_size, float p_screen_lod_threshold, RID p_shadow_atlas) { // render for AR/VR interface - +#if 0 Camera *camera = camera_owner.getornull(p_camera); ERR_FAIL_COND(!camera); @@ -2222,6 +2250,7 @@ void RendererSceneCull::render_camera(RID p_render_buffers, Ref<XRInterface> &p_ // And render our scene... _render_scene(p_render_buffers, cam_transform, camera_matrix, false, environment, camera->effects, p_scenario, p_shadow_atlas, RID(), -1, p_screen_lod_threshold); +#endif }; void RendererSceneCull::_frustum_cull_threaded(uint32_t p_thread, FrustumCullData *cull_data) { @@ -2440,7 +2469,7 @@ void RendererSceneCull::_frustum_cull(FrustumCullData &cull_data, FrustumCullRes } } -void RendererSceneCull::_prepare_scene(const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, bool p_cam_vaspect, RID p_render_buffers, RID p_environment, uint32_t p_visible_layers, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, float p_screen_lod_threshold, bool p_using_shadows) { +void RendererSceneCull::_render_scene(const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, bool p_cam_vaspect, RID p_render_buffers, RID p_environment, RID p_force_camera_effects, uint32_t p_visible_layers, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold, bool p_using_shadows) { // Note, in stereo rendering: // - p_cam_transform will be a transform in the middle of our two eyes // - p_cam_projection is a wider frustrum that encompasses both eyes @@ -2454,6 +2483,7 @@ void RendererSceneCull::_prepare_scene(const Transform p_cam_transform, const Ca scene_render->set_scene_pass(render_pass); if (p_render_buffers.is_valid()) { + //no rendering code here, this is only to set up what needs to be done, request regions, etc. scene_render->sdfgi_update(p_render_buffers, p_environment, p_cam_transform.origin); //update conditions for SDFGI (whether its used or not) } @@ -2584,62 +2614,28 @@ void RendererSceneCull::_prepare_scene(const Transform p_cam_transform, const Ca //render shadows - for (uint32_t i = 0; i < cull.shadow_count; i++) { - for (uint32_t j = 0; j < cull.shadows[i].cascade_count; j++) { - const Cull::Shadow::Cascade &c = cull.shadows[i].cascades[j]; - // print_line("shadow " + itos(i) + " cascade " + itos(j) + " elements: " + itos(c.cull_result.size())); - scene_render->light_instance_set_shadow_transform(cull.shadows[i].light_instance, c.projection, c.transform, c.zfar, c.split, j, c.shadow_texel_size, c.bias_scale, c.range_begin, c.uv_scale); - scene_render->render_shadow(cull.shadows[i].light_instance, p_shadow_atlas, j, frustum_cull_result.directional_shadows[i].cascade_geometry_instances[j], near_plane, p_cam_projection.get_lod_multiplier(), p_screen_lod_threshold); - } - } + max_shadows_used = 0; - //render SDFGI + if (p_using_shadows) { //setup shadow maps - { - if (cull.sdfgi.region_count > 0) { - //update regions - for (uint32_t i = 0; i < cull.sdfgi.region_count; i++) { - scene_render->render_sdfgi(p_render_buffers, i, frustum_cull_result.sdfgi_region_geometry_instances[i]); - } - //check if static lights were culled - bool static_lights_culled = false; - for (uint32_t i = 0; i < cull.sdfgi.cascade_light_count; i++) { - if (frustum_cull_result.sdfgi_cascade_lights[i].size()) { - static_lights_culled = true; - break; - } - } + // Directional Shadows - if (static_lights_culled) { - scene_render->render_sdfgi_static_lights(p_render_buffers, cull.sdfgi.cascade_light_count, cull.sdfgi.cascade_light_index, frustum_cull_result.sdfgi_cascade_lights); + for (uint32_t i = 0; i < cull.shadow_count; i++) { + for (uint32_t j = 0; j < cull.shadows[i].cascade_count; j++) { + const Cull::Shadow::Cascade &c = cull.shadows[i].cascades[j]; + // print_line("shadow " + itos(i) + " cascade " + itos(j) + " elements: " + itos(c.cull_result.size())); + scene_render->light_instance_set_shadow_transform(cull.shadows[i].light_instance, c.projection, c.transform, c.zfar, c.split, j, c.shadow_texel_size, c.bias_scale, c.range_begin, c.uv_scale); + if (max_shadows_used == MAX_UPDATE_SHADOWS) { + continue; + } + render_shadow_data[max_shadows_used].light = cull.shadows[i].light_instance; + render_shadow_data[max_shadows_used].pass = j; + render_shadow_data[max_shadows_used].instances.merge_unordered(frustum_cull_result.directional_shadows[i].cascade_geometry_instances[j]); + max_shadows_used++; } } - if (p_render_buffers.is_valid()) { - scene_render->sdfgi_update_probes(p_render_buffers, p_environment, directional_lights, scenario->dynamic_lights.ptr(), scenario->dynamic_lights.size()); - } - } - - //light_samplers_culled=0; - - /* - print_line("OT: "+rtos( (OS::get_singleton()->get_ticks_usec()-t)/1000.0)); - print_line("OTO: "+itos(p_scenario->octree.get_octant_count())); - print_line("OTE: "+itos(p_scenario->octree.get_elem_count())); - print_line("OTP: "+itos(p_scenario->octree.get_pair_count())); - */ - - /* STEP 3 - PROCESS PORTALS, VALIDATE ROOMS */ - //removed, will replace with culling - - /* STEP 4 - REMOVE FURTHER CULLED OBJECTS, ADD LIGHTS */ - - /* STEP 5 - PROCESS POSITIONAL LIGHTS */ - - if (p_using_shadows) { //setup shadow maps - - //SortArray<Instance*,_InstanceLightsort> sorter; - //sorter.sort(light_cull_result,light_cull_count); + // Positional Shadowss for (uint32_t i = 0; i < (uint32_t)frustum_cull_result.lights.size(); i++) { Instance *ins = frustum_cull_result.lights[i]; @@ -2726,19 +2722,78 @@ void RendererSceneCull::_prepare_scene(const Transform p_cam_transform, const Ca bool redraw = scene_render->shadow_atlas_update_light(p_shadow_atlas, light->instance, coverage, light->last_version); - if (redraw) { + if (redraw && max_shadows_used < MAX_UPDATE_SHADOWS) { //must redraw! RENDER_TIMESTAMP(">Rendering Light " + itos(i)); light->shadow_dirty = _light_instance_update_shadow(ins, p_cam_transform, p_cam_projection, p_cam_orthogonal, p_cam_vaspect, p_shadow_atlas, scenario, p_screen_lod_threshold); RENDER_TIMESTAMP("<Rendering Light " + itos(i)); + } else { + light->shadow_dirty = redraw; + } + } + } + + //render SDFGI + + { + sdfgi_update_data.update_static = false; + + if (cull.sdfgi.region_count > 0) { + //update regions + for (uint32_t i = 0; i < cull.sdfgi.region_count; i++) { + render_sdfgi_data[i].instances.merge_unordered(frustum_cull_result.sdfgi_region_geometry_instances[i]); + render_sdfgi_data[i].region = i; + } + //check if static lights were culled + bool static_lights_culled = false; + for (uint32_t i = 0; i < cull.sdfgi.cascade_light_count; i++) { + if (frustum_cull_result.sdfgi_cascade_lights[i].size()) { + static_lights_culled = true; + break; + } + } + + if (static_lights_culled) { + sdfgi_update_data.static_cascade_count = cull.sdfgi.cascade_light_count; + sdfgi_update_data.static_cascade_indices = cull.sdfgi.cascade_light_index; + sdfgi_update_data.static_positional_lights = frustum_cull_result.sdfgi_cascade_lights; + sdfgi_update_data.update_static = true; } } + + if (p_render_buffers.is_valid()) { + sdfgi_update_data.directional_lights = &directional_lights; + sdfgi_update_data.positional_light_instances = scenario->dynamic_lights.ptr(); + sdfgi_update_data.positional_light_count = scenario->dynamic_lights.size(); + } } //append the directional lights to the lights culled for (int i = 0; i < directional_lights.size(); i++) { frustum_cull_result.light_instances.push_back(directional_lights[i]); } + + RID camera_effects; + if (p_force_camera_effects.is_valid()) { + camera_effects = p_force_camera_effects; + } else { + camera_effects = scenario->camera_effects; + } + /* PROCESS GEOMETRY AND DRAW SCENE */ + + RENDER_TIMESTAMP("Render Scene "); + scene_render->render_scene(p_render_buffers, p_cam_transform, p_cam_projection, p_cam_orthogonal, frustum_cull_result.geometry_instances, frustum_cull_result.light_instances, frustum_cull_result.reflections, frustum_cull_result.gi_probes, frustum_cull_result.decals, frustum_cull_result.lightmaps, p_environment, camera_effects, p_shadow_atlas, p_reflection_probe.is_valid() ? RID() : scenario->reflection_atlas, p_reflection_probe, p_reflection_probe_pass, p_screen_lod_threshold, render_shadow_data, max_shadows_used, render_sdfgi_data, cull.sdfgi.region_count, &sdfgi_update_data); + + for (uint32_t i = 0; i < max_shadows_used; i++) { + render_shadow_data[i].instances.clear(); + } + max_shadows_used = 0; + + for (uint32_t i = 0; i < cull.sdfgi.region_count; i++) { + render_sdfgi_data[i].instances.clear(); + } + + // virtual void render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold,const RenderShadowData *p_render_shadows,int p_render_shadow_count,const RenderSDFGIData *p_render_sdfgi_regions,int p_render_sdfgi_region_count,const RenderSDFGIStaticLightData *p_render_sdfgi_static_lights=nullptr) = 0; } RID RendererSceneCull::_render_get_environment(RID p_camera, RID p_scenario) { @@ -2762,21 +2817,6 @@ RID RendererSceneCull::_render_get_environment(RID p_camera, RID p_scenario) { return RID(); } -void RendererSceneCull::_render_scene(RID p_render_buffers, const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, RID p_environment, RID p_force_camera_effects, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold) { - Scenario *scenario = scenario_owner.getornull(p_scenario); - - RID camera_effects; - if (p_force_camera_effects.is_valid()) { - camera_effects = p_force_camera_effects; - } else { - camera_effects = scenario->camera_effects; - } - /* PROCESS GEOMETRY AND DRAW SCENE */ - - RENDER_TIMESTAMP("Render Scene "); - scene_render->render_scene(p_render_buffers, p_cam_transform, p_cam_projection, p_cam_orthogonal, frustum_cull_result.geometry_instances, frustum_cull_result.light_instances, frustum_cull_result.reflections, frustum_cull_result.gi_probes, frustum_cull_result.decals, frustum_cull_result.lightmaps, p_environment, camera_effects, p_shadow_atlas, p_reflection_probe.is_valid() ? RID() : scenario->reflection_atlas, p_reflection_probe, p_reflection_probe_pass, p_screen_lod_threshold); -} - void RendererSceneCull::render_empty_scene(RID p_render_buffers, RID p_scenario, RID p_shadow_atlas) { #ifndef _3D_DISABLED @@ -2789,7 +2829,7 @@ void RendererSceneCull::render_empty_scene(RID p_render_buffers, RID p_scenario, environment = scenario->fallback_environment; } RENDER_TIMESTAMP("Render Empty Scene "); - scene_render->render_scene(p_render_buffers, Transform(), CameraMatrix(), true, PagedArray<RendererSceneRender::GeometryInstance *>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), RID(), RID(), p_shadow_atlas, scenario->reflection_atlas, RID(), 0, 0); + scene_render->render_scene(p_render_buffers, Transform(), CameraMatrix(), true, PagedArray<RendererSceneRender::GeometryInstance *>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), PagedArray<RID>(), RID(), RID(), p_shadow_atlas, scenario->reflection_atlas, RID(), 0, 0, nullptr, 0, nullptr, 0, nullptr); #endif } @@ -2852,8 +2892,7 @@ bool RendererSceneCull::_render_reflection_probe_step(Instance *p_instance, int } RENDER_TIMESTAMP("Render Reflection Probe, Step " + itos(p_step)); - _prepare_scene(xform, cm, false, false, RID(), RID(), RSG::storage->reflection_probe_get_cull_mask(p_instance->base), p_instance->scenario->self, shadow_atlas, reflection_probe->instance, lod_threshold, use_shadows); - _render_scene(RID(), xform, cm, false, RID(), RID(), p_instance->scenario->self, shadow_atlas, reflection_probe->instance, p_step, lod_threshold); + _render_scene(xform, cm, false, false, RID(), RID(), RID(), RSG::storage->reflection_probe_get_cull_mask(p_instance->base), p_instance->scenario->self, shadow_atlas, reflection_probe->instance, p_step, lod_threshold, use_shadows); } else { //do roughness postprocess step until it believes it's done @@ -3481,7 +3520,12 @@ RendererSceneCull::RendererSceneCull() { instance_cull_result.set_page_pool(&instance_cull_page_pool); instance_shadow_cull_result.set_page_pool(&instance_cull_page_pool); - geometry_instances_to_shadow_render.set_page_pool(&geometry_instance_cull_page_pool); + for (uint32_t i = 0; i < MAX_UPDATE_SHADOWS; i++) { + render_shadow_data[i].instances.set_page_pool(&geometry_instance_cull_page_pool); + } + for (uint32_t i = 0; i < SDFGI_MAX_CASCADES * SDFGI_MAX_REGIONS_PER_CASCADE; i++) { + render_sdfgi_data[i].instances.set_page_pool(&geometry_instance_cull_page_pool); + } frustum_cull_result.init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); frustum_cull_result_threads.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); @@ -3489,8 +3533,8 @@ RendererSceneCull::RendererSceneCull() { frustum_cull_result_threads[i].init(&rid_cull_page_pool, &geometry_instance_cull_page_pool, &instance_cull_page_pool); } - indexer_update_iterations = GLOBAL_GET("rendering/spatial_indexer/update_iterations_per_frame"); - thread_cull_threshold = GLOBAL_GET("rendering/spatial_indexer/threaded_cull_minimum_instances"); + indexer_update_iterations = GLOBAL_GET("rendering/limits/spatial_indexer/update_iterations_per_frame"); + thread_cull_threshold = GLOBAL_GET("rendering/limits/spatial_indexer/threaded_cull_minimum_instances"); thread_cull_threshold = MAX(thread_cull_threshold, (uint32_t)RendererThreadPool::singleton->thread_work_pool.get_thread_count()); //make sure there is at least one thread per CPU } @@ -3498,7 +3542,12 @@ RendererSceneCull::~RendererSceneCull() { instance_cull_result.reset(); instance_shadow_cull_result.reset(); - geometry_instances_to_shadow_render.reset(); + for (uint32_t i = 0; i < MAX_UPDATE_SHADOWS; i++) { + render_shadow_data[i].instances.reset(); + } + for (uint32_t i = 0; i < SDFGI_MAX_CASCADES * SDFGI_MAX_REGIONS_PER_CASCADE; i++) { + render_sdfgi_data[i].instances.reset(); + } frustum_cull_result.reset(); for (uint32_t i = 0; i < frustum_cull_result_threads.size(); i++) { diff --git a/servers/rendering/renderer_scene_cull.h b/servers/rendering/renderer_scene_cull.h index 796fb14743..32f4334288 100644 --- a/servers/rendering/renderer_scene_cull.h +++ b/servers/rendering/renderer_scene_cull.h @@ -54,7 +54,8 @@ public: enum { SDFGI_MAX_CASCADES = 8, SDFGI_MAX_REGIONS_PER_CASCADE = 3, - MAX_INSTANCE_PAIRS = 32 + MAX_INSTANCE_PAIRS = 32, + MAX_UPDATE_SHADOWS = 512 }; uint64_t render_pass; @@ -93,9 +94,11 @@ public: } }; - mutable RID_PtrOwner<Camera> camera_owner; + mutable RID_PtrOwner<Camera, true> camera_owner; + + virtual RID camera_allocate(); + virtual void camera_initialize(RID p_rid); - virtual RID camera_create(); virtual void camera_set_perspective(RID p_camera, float p_fovy_degrees, float p_z_near, float p_z_far); virtual void camera_set_orthogonal(RID p_camera, float p_size, float p_z_near, float p_z_far); virtual void camera_set_frustum(RID p_camera, float p_size, Vector2 p_offset, float p_z_near, float p_z_far); @@ -295,14 +298,15 @@ public: int indexer_update_iterations = 0; - mutable RID_PtrOwner<Scenario> scenario_owner; + mutable RID_PtrOwner<Scenario, true> scenario_owner; static void _instance_pair(Instance *p_A, Instance *p_B); static void _instance_unpair(Instance *p_A, Instance *p_B); void _instance_update_mesh_instance(Instance *p_instance); - virtual RID scenario_create(); + virtual RID scenario_allocate(); + virtual void scenario_initialize(RID p_rid); virtual void scenario_set_debug(RID p_scenario, RS::ScenarioDebugMode p_debug_mode); virtual void scenario_set_environment(RID p_scenario, RID p_environment); @@ -696,7 +700,6 @@ public: PagedArray<Instance *> instance_cull_result; PagedArray<Instance *> instance_shadow_cull_result; - PagedArray<RendererSceneRender::GeometryInstance *> geometry_instances_to_shadow_render; struct FrustumCullResult { PagedArray<RendererSceneRender::GeometryInstance *> geometry_instances; @@ -795,6 +798,7 @@ public: lightmaps.set_page_pool(p_rid_pool); reflections.set_page_pool(p_rid_pool); decals.set_page_pool(p_rid_pool); + gi_probes.set_page_pool(p_rid_pool); mesh_instances.set_page_pool(p_rid_pool); for (int i = 0; i < RendererSceneRender::MAX_DIRECTIONAL_LIGHTS; i++) { for (int j = 0; j < RendererSceneRender::MAX_DIRECTIONAL_LIGHT_CASCADES; j++) { @@ -815,13 +819,20 @@ public: FrustumCullResult frustum_cull_result; LocalVector<FrustumCullResult> frustum_cull_result_threads; + RendererSceneRender::RenderShadowData render_shadow_data[MAX_UPDATE_SHADOWS]; + uint32_t max_shadows_used = 0; + + RendererSceneRender::RenderSDFGIData render_sdfgi_data[SDFGI_MAX_CASCADES * SDFGI_MAX_REGIONS_PER_CASCADE]; + RendererSceneRender::RenderSDFGIUpdateData sdfgi_update_data; + uint32_t thread_cull_threshold = 200; - RID_PtrOwner<Instance> instance_owner; + RID_PtrOwner<Instance, true> instance_owner; uint32_t geometry_instance_pair_mask; // used in traditional forward, unnecesary on clustered - virtual RID instance_create(); + virtual RID instance_allocate(); + virtual void instance_initialize(RID p_rid); virtual void instance_set_base(RID p_instance, RID p_base); virtual void instance_set_scenario(RID p_instance, RID p_scenario); @@ -923,8 +934,7 @@ public: void _frustum_cull(FrustumCullData &cull_data, FrustumCullResult &cull_result, uint64_t p_from, uint64_t p_to); bool _render_reflection_probe_step(Instance *p_instance, int p_step); - void _prepare_scene(const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, bool p_cam_vaspect, RID p_render_buffers, RID p_environment, uint32_t p_visible_layers, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, float p_screen_lod_threshold, bool p_using_shadows = true); - void _render_scene(RID p_render_buffers, const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, RID p_environment, RID p_force_camera_effects, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold); + void _render_scene(const Transform p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_orthogonal, bool p_cam_vaspect, RID p_render_buffers, RID p_environment, RID p_force_camera_effects, uint32_t p_visible_layers, RID p_scenario, RID p_shadow_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold, bool p_using_shadows = true); void render_empty_scene(RID p_render_buffers, RID p_scenario, RID p_shadow_atlas); void render_camera(RID p_render_buffers, RID p_camera, RID p_scenario, Size2 p_viewport_size, float p_screen_lod_threshold, RID p_shadow_atlas); @@ -946,18 +956,21 @@ public: #define PASSBASE scene_render - PASS1(directional_shadow_atlas_set_size, int) + PASS2(directional_shadow_atlas_set_size, int, bool) PASS1(gi_probe_set_quality, RS::GIProbeQuality) /* SKY API */ - PASS0R(RID, sky_create) + PASS0R(RID, sky_allocate) + PASS1(sky_initialize, RID) + PASS2(sky_set_radiance_size, RID, int) PASS2(sky_set_mode, RID, RS::SkyMode) PASS2(sky_set_material, RID, RID) PASS4R(Ref<Image>, sky_bake_panorama, RID, float, bool, const Size2i &) - PASS0R(RID, environment_create) + PASS0R(RID, environment_allocate) + PASS1(environment_initialize, RID) PASS1RC(bool, is_environment, RID) @@ -985,16 +998,15 @@ public: PASS7(environment_set_adjustment, RID, bool, float, float, float, bool, RID) PASS9(environment_set_fog, RID, bool, const Color &, float, float, float, float, float, float) - PASS9(environment_set_volumetric_fog, RID, bool, float, const Color &, float, float, float, float, RS::EnvVolumetricFogShadowFilter) + PASS10(environment_set_volumetric_fog, RID, bool, float, const Color &, float, float, float, float, bool, float) PASS2(environment_set_volumetric_fog_volume_size, int, int) PASS1(environment_set_volumetric_fog_filter_active, bool) - PASS1(environment_set_volumetric_fog_directional_shadow_shrink_size, int) - PASS1(environment_set_volumetric_fog_positional_shadow_shrink_size, int) - PASS11(environment_set_sdfgi, RID, bool, RS::EnvironmentSDFGICascades, float, RS::EnvironmentSDFGIYScale, bool, bool, bool, float, float, float) + PASS11(environment_set_sdfgi, RID, bool, RS::EnvironmentSDFGICascades, float, RS::EnvironmentSDFGIYScale, bool, float, bool, float, float, float) PASS1(environment_set_sdfgi_ray_count, RS::EnvironmentSDFGIRayCount) PASS1(environment_set_sdfgi_frames_to_converge, RS::EnvironmentSDFGIFramesToConverge) + PASS1(environment_set_sdfgi_frames_to_update_light, RS::EnvironmentSDFGIFramesToUpdateLight) PASS1RC(RS::EnvironmentBG, environment_get_background, RID) PASS1RC(int, environment_get_canvas_max_layer, RID) @@ -1007,7 +1019,8 @@ public: /* CAMERA EFFECTS */ - PASS0R(RID, camera_effects_create) + PASS0R(RID, camera_effects_allocate) + PASS1(camera_effects_initialize, RID) PASS2(camera_effects_set_dof_blur_quality, RS::DOFBlurQuality, bool) PASS1(camera_effects_set_dof_blur_bokeh_shape, RS::DOFBokehShape) @@ -1024,10 +1037,11 @@ public: PASS0R(RID, render_buffers_create) PASS7(render_buffers_configure, RID, RID, int, int, RS::ViewportMSAA, RS::ViewportScreenSpaceAA, bool) + PASS1(gi_set_use_half_resolution, bool) /* Shadow Atlas */ PASS0R(RID, shadow_atlas_create) - PASS2(shadow_atlas_set_size, RID, int) + PASS3(shadow_atlas_set_size, RID, int, bool) PASS3(shadow_atlas_set_quadrant_subdivision, RID, int, int) PASS1(set_debug_draw_mode, RS::ViewportDebugDraw) diff --git a/servers/rendering/renderer_scene_render.h b/servers/rendering/renderer_scene_render.h index 85353c400d..1dea3580b6 100644 --- a/servers/rendering/renderer_scene_render.h +++ b/servers/rendering/renderer_scene_render.h @@ -71,13 +71,12 @@ public: /* SHADOW ATLAS API */ - virtual RID - shadow_atlas_create() = 0; - virtual void shadow_atlas_set_size(RID p_atlas, int p_size) = 0; + virtual RID shadow_atlas_create() = 0; + virtual void shadow_atlas_set_size(RID p_atlas, int p_size, bool p_16_bits = false) = 0; virtual void shadow_atlas_set_quadrant_subdivision(RID p_atlas, int p_quadrant, int p_subdivision) = 0; virtual bool shadow_atlas_update_light(RID p_atlas, RID p_light_intance, float p_coverage, uint64_t p_light_version) = 0; - virtual void directional_shadow_atlas_set_size(int p_size) = 0; + virtual void directional_shadow_atlas_set_size(int p_size, bool p_16_bits = false) = 0; virtual int get_directional_light_shadow_size(RID p_light_intance) = 0; virtual void set_directional_shadow_count(int p_count) = 0; @@ -87,11 +86,12 @@ public: virtual int sdfgi_get_pending_region_count(RID p_render_buffers) const = 0; virtual AABB sdfgi_get_pending_region_bounds(RID p_render_buffers, int p_region) const = 0; virtual uint32_t sdfgi_get_pending_region_cascade(RID p_render_buffers, int p_region) const = 0; - virtual void sdfgi_update_probes(RID p_render_buffers, RID p_environment, const Vector<RID> &p_directional_lights, const RID *p_positional_light_instances, uint32_t p_positional_light_count) = 0; /* SKY API */ - virtual RID sky_create() = 0; + virtual RID sky_allocate() = 0; + virtual void sky_initialize(RID p_rid) = 0; + virtual void sky_set_radiance_size(RID p_sky, int p_radiance_size) = 0; virtual void sky_set_mode(RID p_sky, RS::SkyMode p_samples) = 0; virtual void sky_set_material(RID p_sky, RID p_material) = 0; @@ -99,7 +99,8 @@ public: /* ENVIRONMENT API */ - virtual RID environment_create() = 0; + virtual RID environment_allocate() = 0; + virtual void environment_initialize(RID p_rid) = 0; virtual void environment_set_background(RID p_env, RS::EnvironmentBG p_bg) = 0; virtual void environment_set_sky(RID p_env, RID p_sky) = 0; @@ -118,12 +119,9 @@ public: virtual void environment_glow_set_use_bicubic_upscale(bool p_enable) = 0; virtual void environment_glow_set_use_high_quality(bool p_enable) = 0; - virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, RS::EnvVolumetricFogShadowFilter p_shadow_filter) = 0; - + virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) = 0; virtual void environment_set_volumetric_fog_volume_size(int p_size, int p_depth) = 0; virtual void environment_set_volumetric_fog_filter_active(bool p_enable) = 0; - virtual void environment_set_volumetric_fog_directional_shadow_shrink_size(int p_shrink_size) = 0; - virtual void environment_set_volumetric_fog_positional_shadow_shrink_size(int p_shrink_size) = 0; virtual void environment_set_ssr(RID p_env, bool p_enable, int p_max_steps, float p_fade_int, float p_fade_out, float p_depth_tolerance) = 0; virtual void environment_set_ssr_roughness_quality(RS::EnvironmentSSRRoughnessQuality p_quality) = 0; @@ -132,10 +130,11 @@ public: virtual void environment_set_ssao_quality(RS::EnvironmentSSAOQuality p_quality, bool p_half_size, float p_adaptive_target, int p_blur_passes, float p_fadeout_from, float p_fadeout_to) = 0; - virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, bool p_use_multibounce, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; + virtual void environment_set_sdfgi(RID p_env, bool p_enable, RS::EnvironmentSDFGICascades p_cascades, float p_min_cell_size, RS::EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; virtual void environment_set_sdfgi_ray_count(RS::EnvironmentSDFGIRayCount p_ray_count) = 0; virtual void environment_set_sdfgi_frames_to_converge(RS::EnvironmentSDFGIFramesToConverge p_frames) = 0; + virtual void environment_set_sdfgi_frames_to_update_light(RS::EnvironmentSDFGIFramesToUpdateLight p_update) = 0; virtual void environment_set_tonemap(RID p_env, RS::EnvironmentToneMapper p_tone_mapper, float p_exposure, float p_white, bool p_auto_exposure, float p_min_luminance, float p_max_luminance, float p_auto_exp_speed, float p_auto_exp_scale) = 0; @@ -149,7 +148,8 @@ public: virtual RS::EnvironmentBG environment_get_background(RID p_env) const = 0; virtual int environment_get_canvas_max_layer(RID p_env) const = 0; - virtual RID camera_effects_create() = 0; + virtual RID camera_effects_allocate() = 0; + virtual void camera_effects_initialize(RID p_rid) = 0; virtual void camera_effects_set_dof_blur_quality(RS::DOFBlurQuality p_quality, bool p_use_jitter) = 0; virtual void camera_effects_set_dof_blur_bokeh_shape(RS::DOFBokehShape p_shape) = 0; @@ -194,12 +194,31 @@ public: virtual void gi_probe_set_quality(RS::GIProbeQuality) = 0; - virtual void render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold) = 0; + struct RenderShadowData { + RID light; + int pass = 0; + PagedArray<GeometryInstance *> instances; + }; + + struct RenderSDFGIData { + int region = 0; + PagedArray<GeometryInstance *> instances; + }; + + struct RenderSDFGIUpdateData { + bool update_static = false; + uint32_t static_cascade_count; + uint32_t *static_cascade_indices; + PagedArray<RID> *static_positional_lights; + + const Vector<RID> *directional_lights; + const RID *positional_light_instances; + uint32_t positional_light_count; + }; + + virtual void render_scene(RID p_render_buffers, const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, const PagedArray<RID> &p_lights, const PagedArray<RID> &p_reflection_probes, const PagedArray<RID> &p_gi_probes, const PagedArray<RID> &p_decals, const PagedArray<RID> &p_lightmaps, RID p_environment, RID p_camera_effects, RID p_shadow_atlas, RID p_reflection_atlas, RID p_reflection_probe, int p_reflection_probe_pass, float p_screen_lod_threshold, const RenderShadowData *p_render_shadows, int p_render_shadow_count, const RenderSDFGIData *p_render_sdfgi_regions, int p_render_sdfgi_region_count, const RenderSDFGIUpdateData *p_sdfgi_update_data = nullptr) = 0; - virtual void render_shadow(RID p_light, RID p_shadow_atlas, int p_pass, const PagedArray<GeometryInstance *> &p_instances, const Plane &p_camera_plane = Plane(), float p_lod_distance_multiplier = 0, float p_screen_lod_threshold = 0.0) = 0; virtual void render_material(const Transform &p_cam_transform, const CameraMatrix &p_cam_projection, bool p_cam_ortogonal, const PagedArray<GeometryInstance *> &p_instances, RID p_framebuffer, const Rect2i &p_region) = 0; - virtual void render_sdfgi(RID p_render_buffers, int p_region, const PagedArray<GeometryInstance *> &p_instances) = 0; - virtual void render_sdfgi_static_lights(RID p_render_buffers, uint32_t p_cascade_count, const uint32_t *p_cascade_indices, const PagedArray<RID> *p_positional_lights) = 0; virtual void render_particle_collider_heightfield(RID p_collider, const Transform &p_transform, const PagedArray<GeometryInstance *> &p_instances) = 0; virtual void set_scene_pass(uint64_t p_pass) = 0; @@ -208,6 +227,7 @@ public: virtual RID render_buffers_create() = 0; virtual void render_buffers_configure(RID p_render_buffers, RID p_render_target, int p_width, int p_height, RS::ViewportMSAA p_msaa, RS::ViewportScreenSpaceAA p_screen_space_aa, bool p_use_debanding) = 0; + virtual void gi_set_use_half_resolution(bool p_enable) = 0; virtual void screen_space_roughness_limiter_set_active(bool p_enable, float p_amount, float p_limit) = 0; virtual bool screen_space_roughness_limiter_is_active() const = 0; diff --git a/servers/rendering/renderer_storage.h b/servers/rendering/renderer_storage.h index 3e53f7130a..22cf6acb19 100644 --- a/servers/rendering/renderer_storage.h +++ b/servers/rendering/renderer_storage.h @@ -98,6 +98,7 @@ public: while (to_clean_up.size()) { to_clean_up.front()->get().first->instances.erase(to_clean_up.front()->get().second); + dependencies.erase(to_clean_up.front()->get().first); to_clean_up.pop_front(); } } @@ -118,12 +119,15 @@ public: Set<Dependency *> dependencies; }; + virtual bool can_create_resources_async() const = 0; /* TEXTURE API */ - virtual RID texture_2d_create(const Ref<Image> &p_image) = 0; - virtual RID texture_2d_layered_create(const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type) = 0; - virtual RID texture_3d_create(Image::Format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) = 0; - virtual RID texture_proxy_create(RID p_base) = 0; //all slices, then all the mipmaps, must be coherent + virtual RID texture_allocate() = 0; + + virtual void texture_2d_initialize(RID p_texture, const Ref<Image> &p_image) = 0; + virtual void texture_2d_layered_initialize(RID p_texture, const Vector<Ref<Image>> &p_layers, RS::TextureLayeredType p_layered_type) = 0; + virtual void texture_3d_initialize(RID p_texture, Image::Format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) = 0; + virtual void texture_proxy_initialize(RID p_texture, RID p_base) = 0; //all slices, then all the mipmaps, must be coherent virtual void texture_2d_update_immediate(RID p_texture, const Ref<Image> &p_image, int p_layer = 0) = 0; //mostly used for video and streaming virtual void texture_2d_update(RID p_texture, const Ref<Image> &p_image, int p_layer = 0) = 0; @@ -131,9 +135,9 @@ public: virtual void texture_proxy_update(RID p_proxy, RID p_base) = 0; //these two APIs can be used together or in combination with the others. - virtual RID texture_2d_placeholder_create() = 0; - virtual RID texture_2d_layered_placeholder_create(RenderingServer::TextureLayeredType p_layered_type) = 0; - virtual RID texture_3d_placeholder_create() = 0; + virtual void texture_2d_placeholder_initialize(RID p_texture) = 0; + virtual void texture_2d_layered_placeholder_initialize(RID p_texture, RenderingServer::TextureLayeredType p_layered_type) = 0; + virtual void texture_3d_placeholder_initialize(RID p_texture) = 0; virtual Ref<Image> texture_2d_get(RID p_texture) const = 0; virtual Ref<Image> texture_2d_layer_get(RID p_texture, int p_layer) const = 0; @@ -160,7 +164,9 @@ public: /* CANVAS TEXTURE API */ - virtual RID canvas_texture_create() = 0; + virtual RID canvas_texture_allocate() = 0; + virtual void canvas_texture_initialize(RID p_rid) = 0; + virtual void canvas_texture_set_channel(RID p_canvas_texture, RS::CanvasTextureChannel p_channel, RID p_texture) = 0; virtual void canvas_texture_set_shading_parameters(RID p_canvas_texture, const Color &p_base_color, float p_shininess) = 0; @@ -169,7 +175,8 @@ public: /* SHADER API */ - virtual RID shader_create() = 0; + virtual RID shader_allocate() = 0; + virtual void shader_initialize(RID p_rid) = 0; virtual void shader_set_code(RID p_shader, const String &p_code) = 0; virtual String shader_get_code(RID p_shader) const = 0; @@ -179,9 +186,12 @@ public: virtual RID shader_get_default_texture_param(RID p_shader, const StringName &p_name) const = 0; virtual Variant shader_get_param_default(RID p_material, const StringName &p_param) const = 0; + virtual RS::ShaderNativeSourceCode shader_get_native_source_code(RID p_shader) const = 0; + /* COMMON MATERIAL API */ - virtual RID material_create() = 0; + virtual RID material_allocate() = 0; + virtual void material_initialize(RID p_rid) = 0; virtual void material_set_render_priority(RID p_material, int priority) = 0; virtual void material_set_shader(RID p_shader_material, RID p_shader) = 0; @@ -206,7 +216,8 @@ public: /* MESH API */ - virtual RID mesh_create() = 0; + virtual RID mesh_allocate() = 0; + virtual void mesh_initialize(RID p_rid) = 0; virtual void mesh_set_blend_shape_count(RID p_mesh, int p_blend_shape_count) = 0; @@ -232,6 +243,8 @@ public: virtual AABB mesh_get_aabb(RID p_mesh, RID p_skeleton = RID()) = 0; + virtual void mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) = 0; + virtual void mesh_clear(RID p_mesh) = 0; virtual bool mesh_needs_instance(RID p_mesh, bool p_has_skeleton) = 0; @@ -246,9 +259,10 @@ public: /* MULTIMESH API */ - virtual RID multimesh_create() = 0; + virtual RID multimesh_allocate() = 0; + virtual void multimesh_initialize(RID p_rid) = 0; - virtual void multimesh_allocate(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false) = 0; + virtual void multimesh_allocate_data(RID p_multimesh, int p_instances, RS::MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false) = 0; virtual int multimesh_get_instance_count(RID p_multimesh) const = 0; @@ -275,7 +289,9 @@ public: /* IMMEDIATE API */ - virtual RID immediate_create() = 0; + virtual RID immediate_allocate() = 0; + virtual void immediate_initialize(RID p_rid) = 0; + virtual void immediate_begin(RID p_immediate, RS::PrimitiveType p_rimitive, RID p_texture = RID()) = 0; virtual void immediate_vertex(RID p_immediate, const Vector3 &p_vertex) = 0; virtual void immediate_normal(RID p_immediate, const Vector3 &p_normal) = 0; @@ -291,8 +307,10 @@ public: /* SKELETON API */ - virtual RID skeleton_create() = 0; - virtual void skeleton_allocate(RID p_skeleton, int p_bones, bool p_2d_skeleton = false) = 0; + virtual RID skeleton_allocate() = 0; + virtual void skeleton_initialize(RID p_rid) = 0; + + virtual void skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton = false) = 0; virtual int skeleton_get_bone_count(RID p_skeleton) const = 0; virtual void skeleton_bone_set_transform(RID p_skeleton, int p_bone, const Transform &p_transform) = 0; virtual Transform skeleton_bone_get_transform(RID p_skeleton, int p_bone) const = 0; @@ -302,11 +320,14 @@ public: /* Light API */ - virtual RID light_create(RS::LightType p_type) = 0; + virtual RID directional_light_allocate() = 0; + virtual void directional_light_initialize(RID p_rid) = 0; + + virtual RID omni_light_allocate() = 0; + virtual void omni_light_initialize(RID p_rid) = 0; - RID directional_light_create() { return light_create(RS::LIGHT_DIRECTIONAL); } - RID omni_light_create() { return light_create(RS::LIGHT_OMNI); } - RID spot_light_create() { return light_create(RS::LIGHT_SPOT); } + virtual RID spot_light_allocate() = 0; + virtual void spot_light_initialize(RID p_rid) = 0; virtual void light_set_color(RID p_light, const Color &p_color) = 0; virtual void light_set_param(RID p_light, RS::LightParam p_param, float p_value) = 0; @@ -344,7 +365,8 @@ public: /* PROBE API */ - virtual RID reflection_probe_create() = 0; + virtual RID reflection_probe_allocate() = 0; + virtual void reflection_probe_initialize(RID p_rid) = 0; virtual void reflection_probe_set_update_mode(RID p_probe, RS::ReflectionProbeUpdateMode p_mode) = 0; virtual void reflection_probe_set_resolution(RID p_probe, int p_resolution) = 0; @@ -375,7 +397,9 @@ public: /* DECAL API */ - virtual RID decal_create() = 0; + virtual RID decal_allocate() = 0; + virtual void decal_initialize(RID p_rid) = 0; + virtual void decal_set_extents(RID p_decal, const Vector3 &p_extents) = 0; virtual void decal_set_texture(RID p_decal, RS::DecalTexture p_type, RID p_texture) = 0; virtual void decal_set_emission_energy(RID p_decal, float p_energy) = 0; @@ -390,9 +414,10 @@ public: /* GI PROBE API */ - virtual RID gi_probe_create() = 0; + virtual RID gi_probe_allocate() = 0; + virtual void gi_probe_initialize(RID p_rid) = 0; - virtual void gi_probe_allocate(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) = 0; + virtual void gi_probe_allocate_data(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) = 0; virtual AABB gi_probe_get_bounds(RID p_gi_probe) const = 0; virtual Vector3i gi_probe_get_octree_size(RID p_gi_probe) const = 0; @@ -435,9 +460,10 @@ public: virtual uint32_t gi_probe_get_version(RID p_probe) = 0; - /* LIGHTMAP CAPTURE */ + /* LIGHTMAP */ - virtual RID lightmap_create() = 0; + virtual RID lightmap_allocate() = 0; + virtual void lightmap_initialize(RID p_rid) = 0; virtual void lightmap_set_textures(RID p_lightmap, RID p_light, bool p_uses_spherical_haromics) = 0; virtual void lightmap_set_probe_bounds(RID p_lightmap, const AABB &p_bounds) = 0; @@ -455,7 +481,8 @@ public: /* PARTICLES */ - virtual RID particles_create() = 0; + virtual RID particles_allocate() = 0; + virtual void particles_initialize(RID p_rid) = 0; virtual void particles_set_emitting(RID p_particles, bool p_emitting) = 0; virtual bool particles_get_emitting(RID p_particles) = 0; @@ -502,7 +529,9 @@ public: /* PARTICLES COLLISION */ - virtual RID particles_collision_create() = 0; + virtual RID particles_collision_allocate() = 0; + virtual void particles_collision_initialize(RID p_rid) = 0; + virtual void particles_collision_set_collision_type(RID p_particles_collision, RS::ParticlesCollisionType p_type) = 0; virtual void particles_collision_set_cull_mask(RID p_particles_collision, uint32_t p_cull_mask) = 0; virtual void particles_collision_set_sphere_radius(RID p_particles_collision, float p_radius) = 0; //for spheres @@ -579,7 +608,7 @@ public: virtual void render_info_end_capture() = 0; virtual int get_captured_render_info(RS::RenderInfo p_info) = 0; - virtual int get_render_info(RS::RenderInfo p_info) = 0; + virtual uint64_t get_render_info(RS::RenderInfo p_info) = 0; virtual String get_video_adapter_name() const = 0; virtual String get_video_adapter_vendor() const = 0; diff --git a/servers/rendering/renderer_viewport.cpp b/servers/rendering/renderer_viewport.cpp index 9956e4050b..a5d5033c18 100644 --- a/servers/rendering/renderer_viewport.cpp +++ b/servers/rendering/renderer_viewport.cpp @@ -457,7 +457,7 @@ void RendererViewport::draw_viewports() { } if (Engine::get_singleton()->is_editor_hint()) { - set_default_clear_color(GLOBAL_GET("rendering/environment/default_clear_color")); + set_default_clear_color(GLOBAL_GET("rendering/environment/defaults/default_clear_color")); } //sort viewports @@ -608,19 +608,20 @@ void RendererViewport::draw_viewports() { } } -RID RendererViewport::viewport_create() { - Viewport *viewport = memnew(Viewport); - - RID rid = viewport_owner.make_rid(viewport); +RID RendererViewport::viewport_allocate() { + return viewport_owner.allocate_rid(); +} - viewport->self = rid; +void RendererViewport::viewport_initialize(RID p_rid) { + Viewport *viewport = memnew(Viewport); + viewport->self = p_rid; viewport->hide_scenario = false; viewport->hide_canvas = false; viewport->render_target = RSG::storage->render_target_create(); viewport->shadow_atlas = RSG::scene->shadow_atlas_create(); viewport->viewport_render_direct_to_screen = false; - return rid; + viewport_owner.initialize_rid(p_rid, viewport); } void RendererViewport::viewport_set_use_xr(RID p_viewport, bool p_use_xr) { @@ -831,13 +832,14 @@ void RendererViewport::viewport_set_canvas_stacking(RID p_viewport, RID p_canvas viewport->canvas_map[p_canvas].sublayer = p_sublayer; } -void RendererViewport::viewport_set_shadow_atlas_size(RID p_viewport, int p_size) { +void RendererViewport::viewport_set_shadow_atlas_size(RID p_viewport, int p_size, bool p_16_bits) { Viewport *viewport = viewport_owner.getornull(p_viewport); ERR_FAIL_COND(!viewport); viewport->shadow_atlas_size = p_size; + viewport->shadow_atlas_16_bits = p_16_bits; - RSG::scene->shadow_atlas_set_size(viewport->shadow_atlas, viewport->shadow_atlas_size); + RSG::scene->shadow_atlas_set_size(viewport->shadow_atlas, viewport->shadow_atlas_size, viewport->shadow_atlas_16_bits); } void RendererViewport::viewport_set_shadow_atlas_quadrant_subdivision(RID p_viewport, int p_quadrant, int p_subdiv) { @@ -1018,5 +1020,10 @@ void RendererViewport::set_default_clear_color(const Color &p_color) { RSG::storage->set_default_clear_color(p_color); } +//workaround for setting this on thread +void RendererViewport::call_set_use_vsync(bool p_enable) { + DisplayServer::get_singleton()->_set_use_vsync(p_enable); +} + RendererViewport::RendererViewport() { } diff --git a/servers/rendering/renderer_viewport.h b/servers/rendering/renderer_viewport.h index c3ff52a836..f5ed543e8d 100644 --- a/servers/rendering/renderer_viewport.h +++ b/servers/rendering/renderer_viewport.h @@ -81,6 +81,7 @@ public: RID shadow_atlas; int shadow_atlas_size; + bool shadow_atlas_16_bits = false; bool sdf_active; @@ -164,7 +165,7 @@ public: uint64_t draw_viewports_pass = 0; - mutable RID_PtrOwner<Viewport> viewport_owner; + mutable RID_PtrOwner<Viewport, true> viewport_owner; struct ViewportSort { _FORCE_INLINE_ bool operator()(const Viewport *p_left, const Viewport *p_right) const { @@ -185,7 +186,8 @@ private: void _draw_viewport(Viewport *p_viewport, XRInterface::Eyes p_eye = XRInterface::EYE_MONO); public: - RID viewport_create(); + RID viewport_allocate(); + void viewport_initialize(RID p_rid); void viewport_set_use_xr(RID p_viewport, bool p_use_xr); @@ -217,7 +219,7 @@ public: void viewport_set_global_canvas_transform(RID p_viewport, const Transform2D &p_transform); void viewport_set_canvas_stacking(RID p_viewport, RID p_canvas, int p_layer, int p_sublayer); - void viewport_set_shadow_atlas_size(RID p_viewport, int p_size); + void viewport_set_shadow_atlas_size(RID p_viewport, int p_size, bool p_16_bits = false); void viewport_set_shadow_atlas_quadrant_subdivision(RID p_viewport, int p_quadrant, int p_subdiv); void viewport_set_msaa(RID p_viewport, RS::ViewportMSAA p_msaa); @@ -248,6 +250,9 @@ public: bool free(RID p_rid); + //workaround for setting this on thread + void call_set_use_vsync(bool p_enable); + RendererViewport(); virtual ~RendererViewport() {} }; diff --git a/servers/rendering/rendering_device.cpp b/servers/rendering/rendering_device.cpp index 73c86a0a1d..e6ad001807 100644 --- a/servers/rendering/rendering_device.cpp +++ b/servers/rendering/rendering_device.cpp @@ -59,7 +59,7 @@ Vector<uint8_t> RenderingDevice::shader_compile_from_source(ShaderStage p_stage, ERR_FAIL_COND_V(!compile_function, Vector<uint8_t>()); - return compile_function(p_stage, p_source_code, p_language, r_error); + return compile_function(p_stage, p_source_code, p_language, r_error, &device_capabilities); } RID RenderingDevice::_texture_create(const Ref<RDTextureFormat> &p_format, const Ref<RDTextureView> &p_view, const TypedArray<PackedByteArray> &p_data) { @@ -174,8 +174,8 @@ RID RenderingDevice::_uniform_set_create(const Array &p_uniforms, RID p_shader, return uniform_set_create(uniforms, p_shader, p_shader_set); } -Error RenderingDevice::_buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const Vector<uint8_t> &p_data, bool p_sync_with_draw) { - return buffer_update(p_buffer, p_offset, p_size, p_data.ptr(), p_sync_with_draw); +Error RenderingDevice::_buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const Vector<uint8_t> &p_data, uint32_t p_post_barrier) { + return buffer_update(p_buffer, p_offset, p_size, p_data.ptr(), p_post_barrier); } RID RenderingDevice::_render_pipeline_create(RID p_shader, FramebufferFormatID p_framebuffer_format, VertexFormatID p_vertex_format, RenderPrimitive p_render_primitive, const Ref<RDPipelineRasterizationState> &p_rasterization_state, const Ref<RDPipelineMultisampleState> &p_multisample_state, const Ref<RDPipelineDepthStencilState> &p_depth_stencil_state, const Ref<RDPipelineColorBlendState> &p_blend_state, int p_dynamic_state_flags) { @@ -240,16 +240,12 @@ void RenderingDevice::_compute_list_set_push_constant(ComputeListID p_list, cons compute_list_set_push_constant(p_list, p_data.ptr(), p_data_size); } -void RenderingDevice::compute_list_dispatch_threads(ComputeListID p_list, uint32_t p_x_threads, uint32_t p_y_threads, uint32_t p_z_threads, uint32_t p_x_local_group, uint32_t p_y_local_group, uint32_t p_z_local_group) { - compute_list_dispatch(p_list, (p_x_threads - 1) / p_x_local_group + 1, (p_y_threads - 1) / p_y_local_group + 1, (p_z_threads - 1) / p_z_local_group + 1); -} - void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("texture_create", "format", "view", "data"), &RenderingDevice::_texture_create, DEFVAL(Array())); ClassDB::bind_method(D_METHOD("texture_create_shared", "view", "with_texture"), &RenderingDevice::_texture_create_shared); ClassDB::bind_method(D_METHOD("texture_create_shared_from_slice", "view", "with_texture", "layer", "mipmap", "slice_type"), &RenderingDevice::_texture_create_shared_from_slice, DEFVAL(TEXTURE_SLICE_2D)); - ClassDB::bind_method(D_METHOD("texture_update", "texture", "layer", "data", "sync_with_draw"), &RenderingDevice::texture_update, DEFVAL(false)); + ClassDB::bind_method(D_METHOD("texture_update", "texture", "layer", "data", "post_barrier"), &RenderingDevice::texture_update, DEFVAL(BARRIER_MASK_ALL)); ClassDB::bind_method(D_METHOD("texture_get_data", "texture", "layer"), &RenderingDevice::texture_get_data); ClassDB::bind_method(D_METHOD("texture_is_format_supported_for_usage", "format", "usage_flags"), &RenderingDevice::texture_is_format_supported_for_usage); @@ -257,15 +253,15 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("texture_is_shared", "texture"), &RenderingDevice::texture_is_shared); ClassDB::bind_method(D_METHOD("texture_is_valid", "texture"), &RenderingDevice::texture_is_valid); - ClassDB::bind_method(D_METHOD("texture_copy", "from_texture", "to_texture", "from_pos", "to_pos", "size", "src_mipmap", "dst_mipmap", "src_layer", "dst_layer", "sync_with_draw"), &RenderingDevice::texture_copy, DEFVAL(false)); - ClassDB::bind_method(D_METHOD("texture_clear", "texture", "color", "base_mipmap", "mipmap_count", "base_layer", "layer_count", "sync_with_draw"), &RenderingDevice::texture_clear, DEFVAL(false)); - ClassDB::bind_method(D_METHOD("texture_resolve_multisample", "from_texture", "to_texture", "sync_with_draw"), &RenderingDevice::texture_resolve_multisample, DEFVAL(false)); + ClassDB::bind_method(D_METHOD("texture_copy", "from_texture", "to_texture", "from_pos", "to_pos", "size", "src_mipmap", "dst_mipmap", "src_layer", "dst_layer", "post_barrier"), &RenderingDevice::texture_copy, DEFVAL(BARRIER_MASK_ALL)); + ClassDB::bind_method(D_METHOD("texture_clear", "texture", "color", "base_mipmap", "mipmap_count", "base_layer", "layer_count", "post_barrier"), &RenderingDevice::texture_clear, DEFVAL(BARRIER_MASK_ALL)); + ClassDB::bind_method(D_METHOD("texture_resolve_multisample", "from_texture", "to_texture", "post_barrier"), &RenderingDevice::texture_resolve_multisample, DEFVAL(BARRIER_MASK_ALL)); ClassDB::bind_method(D_METHOD("framebuffer_format_create", "attachments"), &RenderingDevice::_framebuffer_format_create); - ClassDB::bind_method(D_METHOD("framebuffer_format_create_empty", "size"), &RenderingDevice::framebuffer_format_create_empty); + ClassDB::bind_method(D_METHOD("framebuffer_format_create_empty", "samples"), &RenderingDevice::framebuffer_format_create_empty, DEFVAL(TEXTURE_SAMPLES_1)); ClassDB::bind_method(D_METHOD("framebuffer_format_get_texture_samples", "format"), &RenderingDevice::framebuffer_format_get_texture_samples); ClassDB::bind_method(D_METHOD("framebuffer_create", "textures", "validate_with_format"), &RenderingDevice::_framebuffer_create, DEFVAL(INVALID_FORMAT_ID)); - ClassDB::bind_method(D_METHOD("framebuffer_create_empty", "size", "validate_with_format"), &RenderingDevice::framebuffer_create_empty, DEFVAL(INVALID_FORMAT_ID)); + ClassDB::bind_method(D_METHOD("framebuffer_create_empty", "size", "samples", "validate_with_format"), &RenderingDevice::framebuffer_create_empty, DEFVAL(TEXTURE_SAMPLES_1), DEFVAL(INVALID_FORMAT_ID)); ClassDB::bind_method(D_METHOD("framebuffer_get_format", "framebuffer"), &RenderingDevice::framebuffer_get_format); ClassDB::bind_method(D_METHOD("sampler_create", "state"), &RenderingDevice::_sampler_create); @@ -273,7 +269,7 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("vertex_buffer_create", "size_bytes", "data", "use_as_storage"), &RenderingDevice::vertex_buffer_create, DEFVAL(Vector<uint8_t>()), DEFVAL(false)); ClassDB::bind_method(D_METHOD("vertex_format_create", "vertex_descriptions"), &RenderingDevice::_vertex_format_create); - ClassDB::bind_method(D_METHOD("index_buffer_create", "size_indices", "format", "data"), &RenderingDevice::index_buffer_create, DEFVAL(Vector<uint8_t>()), DEFVAL(false)); + ClassDB::bind_method(D_METHOD("index_buffer_create", "size_indices", "format", "data", "use_restart_indices"), &RenderingDevice::index_buffer_create, DEFVAL(Vector<uint8_t>()), DEFVAL(false)); ClassDB::bind_method(D_METHOD("index_array_create", "index_buffer", "index_offset", "index_count"), &RenderingDevice::index_array_create); ClassDB::bind_method(D_METHOD("shader_compile_from_source", "shader_source", "allow_cache"), &RenderingDevice::_shader_compile_from_source, DEFVAL(true)); @@ -287,7 +283,8 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("uniform_set_create", "uniforms", "shader", "shader_set"), &RenderingDevice::_uniform_set_create); ClassDB::bind_method(D_METHOD("uniform_set_is_valid", "uniform_set"), &RenderingDevice::uniform_set_is_valid); - ClassDB::bind_method(D_METHOD("buffer_update", "buffer", "offset", "size_bytes", "data", "sync_with_draw"), &RenderingDevice::_buffer_update, DEFVAL(true)); + ClassDB::bind_method(D_METHOD("buffer_update", "buffer", "offset", "size_bytes", "data", "post_barrier"), &RenderingDevice::_buffer_update, DEFVAL(BARRIER_MASK_ALL)); + ClassDB::bind_method(D_METHOD("buffer_clear", "buffer", "offset", "size_bytes", "post_barrier"), &RenderingDevice::buffer_clear, DEFVAL(BARRIER_MASK_ALL)); ClassDB::bind_method(D_METHOD("buffer_get_data", "buffer"), &RenderingDevice::buffer_get_data); ClassDB::bind_method(D_METHOD("render_pipeline_create", "shader", "framebuffer_format", "vertex_format", "primitive", "rasterization_state", "multisample_state", "stencil_state", "color_blend_state", "dynamic_state_flags"), &RenderingDevice::_render_pipeline_create, DEFVAL(0)); @@ -316,19 +313,19 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("draw_list_enable_scissor", "draw_list", "rect"), &RenderingDevice::draw_list_enable_scissor, DEFVAL(Rect2i())); ClassDB::bind_method(D_METHOD("draw_list_disable_scissor", "draw_list"), &RenderingDevice::draw_list_disable_scissor); - ClassDB::bind_method(D_METHOD("draw_list_end"), &RenderingDevice::draw_list_end); + ClassDB::bind_method(D_METHOD("draw_list_end", "post_barrier"), &RenderingDevice::draw_list_end, DEFVAL(BARRIER_MASK_ALL)); - ClassDB::bind_method(D_METHOD("compute_list_begin"), &RenderingDevice::compute_list_begin); + ClassDB::bind_method(D_METHOD("compute_list_begin", "allow_draw_overlap"), &RenderingDevice::compute_list_begin, DEFVAL(false)); ClassDB::bind_method(D_METHOD("compute_list_bind_compute_pipeline", "compute_list", "compute_pipeline"), &RenderingDevice::compute_list_bind_compute_pipeline); ClassDB::bind_method(D_METHOD("compute_list_set_push_constant", "compute_list", "buffer", "size_bytes"), &RenderingDevice::_compute_list_set_push_constant); ClassDB::bind_method(D_METHOD("compute_list_bind_uniform_set", "compute_list", "uniform_set", "set_index"), &RenderingDevice::compute_list_bind_uniform_set); ClassDB::bind_method(D_METHOD("compute_list_dispatch", "compute_list", "x_groups", "y_groups", "z_groups"), &RenderingDevice::compute_list_dispatch); ClassDB::bind_method(D_METHOD("compute_list_add_barrier", "compute_list"), &RenderingDevice::compute_list_add_barrier); - ClassDB::bind_method(D_METHOD("compute_list_end"), &RenderingDevice::compute_list_end); + ClassDB::bind_method(D_METHOD("compute_list_end", "post_barrier"), &RenderingDevice::compute_list_end, DEFVAL(BARRIER_MASK_ALL)); ClassDB::bind_method(D_METHOD("free", "rid"), &RenderingDevice::free); - ClassDB::bind_method(D_METHOD("capture_timestamp", "name", "sync_to_draw"), &RenderingDevice::capture_timestamp); + ClassDB::bind_method(D_METHOD("capture_timestamp", "name"), &RenderingDevice::capture_timestamp); ClassDB::bind_method(D_METHOD("get_captured_timestamps_count"), &RenderingDevice::get_captured_timestamps_count); ClassDB::bind_method(D_METHOD("get_captured_timestamps_frame"), &RenderingDevice::get_captured_timestamps_frame); ClassDB::bind_method(D_METHOD("get_captured_timestamp_gpu_time", "index"), &RenderingDevice::get_captured_timestamp_gpu_time); @@ -340,8 +337,27 @@ void RenderingDevice::_bind_methods() { ClassDB::bind_method(D_METHOD("submit"), &RenderingDevice::submit); ClassDB::bind_method(D_METHOD("sync"), &RenderingDevice::sync); + ClassDB::bind_method(D_METHOD("barrier", "from", "to"), &RenderingDevice::barrier, DEFVAL(BARRIER_MASK_ALL), DEFVAL(BARRIER_MASK_ALL)); + ClassDB::bind_method(D_METHOD("full_barrier"), &RenderingDevice::full_barrier); + ClassDB::bind_method(D_METHOD("create_local_device"), &RenderingDevice::create_local_device); + ClassDB::bind_method(D_METHOD("set_resource_name", "id", "name"), &RenderingDevice::set_resource_name); + + ClassDB::bind_method(D_METHOD("draw_command_begin_label", "name", "color"), &RenderingDevice::draw_command_begin_label); + ClassDB::bind_method(D_METHOD("draw_command_insert_label", "name", "color"), &RenderingDevice::draw_command_insert_label); + ClassDB::bind_method(D_METHOD("draw_command_end_label"), &RenderingDevice::draw_command_end_label); + + ClassDB::bind_method(D_METHOD("get_device_vendor_name"), &RenderingDevice::get_device_vendor_name); + ClassDB::bind_method(D_METHOD("get_device_name"), &RenderingDevice::get_device_name); + ClassDB::bind_method(D_METHOD("get_device_pipeline_cache_uuid"), &RenderingDevice::get_device_pipeline_cache_uuid); + + BIND_CONSTANT(BARRIER_MASK_RASTER); + BIND_CONSTANT(BARRIER_MASK_COMPUTE); + BIND_CONSTANT(BARRIER_MASK_TRANSFER); + BIND_CONSTANT(BARRIER_MASK_ALL); + BIND_CONSTANT(BARRIER_MASK_NO_BARRIER); + BIND_ENUM_CONSTANT(DATA_FORMAT_R4G4_UNORM_PACK8); BIND_ENUM_CONSTANT(DATA_FORMAT_R4G4B4A4_UNORM_PACK16); BIND_ENUM_CONSTANT(DATA_FORMAT_B4G4R4A4_UNORM_PACK16); @@ -744,6 +760,8 @@ void RenderingDevice::_bind_methods() { BIND_ENUM_CONSTANT(DYNAMIC_STATE_STENCIL_REFERENCE); BIND_ENUM_CONSTANT(INITIAL_ACTION_CLEAR); //start rendering and clear the framebuffer (supply params) + BIND_ENUM_CONSTANT(INITIAL_ACTION_CLEAR_REGION); //start rendering and clear the framebuffer (supply params) + BIND_ENUM_CONSTANT(INITIAL_ACTION_CLEAR_REGION_CONTINUE); //continue rendering and clear the framebuffer (supply params) BIND_ENUM_CONSTANT(INITIAL_ACTION_KEEP); //start rendering); but keep attached color texture contents (depth will be cleared) BIND_ENUM_CONSTANT(INITIAL_ACTION_DROP); //start rendering); ignore what is there); just write above it BIND_ENUM_CONSTANT(INITIAL_ACTION_CONTINUE); //continue rendering (framebuffer must have been left in "continue" state as final action previously) diff --git a/servers/rendering/rendering_device.h b/servers/rendering/rendering_device.h index 787805ea6a..2de0549e8d 100644 --- a/servers/rendering/rendering_device.h +++ b/servers/rendering/rendering_device.h @@ -51,6 +51,13 @@ class RDPipelineColorBlendState; class RenderingDevice : public Object { GDCLASS(RenderingDevice, Object) public: + enum DeviceFamily { + DEVICE_UNKNOWN, + DEVICE_OPENGL, + DEVICE_VULKAN, + DEVICE_DIRECTX + }; + enum ShaderStage { SHADER_STAGE_VERTEX, SHADER_STAGE_FRAGMENT, @@ -70,7 +77,29 @@ public: SHADER_LANGUAGE_HLSL }; - typedef Vector<uint8_t> (*ShaderCompileFunction)(ShaderStage p_stage, const String &p_source_code, ShaderLanguage p_language, String *r_error); + enum SubgroupOperations { + SUBGROUP_BASIC_BIT = 1, + SUBGROUP_VOTE_BIT = 2, + SUBGROUP_ARITHMETIC_BIT = 4, + SUBGROUP_BALLOT_BIT = 8, + SUBGROUP_SHUFFLE_BIT = 16, + SUBGROUP_SHUFFLE_RELATIVE_BIT = 32, + SUBGROUP_CLUSTERED_BIT = 64, + SUBGROUP_QUAD_BIT = 128, + }; + + struct Capabilities { + // main device info + DeviceFamily device_family = DEVICE_UNKNOWN; + uint32_t version_major = 1.0; + uint32_t version_minor = 0.0; + // subgroup capabilities + uint32_t subgroup_size = 0; + uint32_t subgroup_in_shaders = 0; // Set flags using SHADER_STAGE_VERTEX_BIT, SHADER_STAGE_FRAGMENT_BIT, etc. + uint32_t subgroup_operations = 0; // Set flags, using SubgroupOperations + }; + + typedef Vector<uint8_t> (*ShaderCompileFunction)(ShaderStage p_stage, const String &p_source_code, ShaderLanguage p_language, String *r_error, const Capabilities *p_capabilities); typedef Vector<uint8_t> (*ShaderCacheFunction)(ShaderStage p_stage, const String &p_source_code, ShaderLanguage p_language); private: @@ -82,6 +111,8 @@ private: protected: static void _bind_methods(); + Capabilities device_capabilities; + public: //base numeric ID for all types enum { @@ -336,6 +367,18 @@ public: }; /*****************/ + /**** BARRIER ****/ + /*****************/ + + enum BarrierMask { + BARRIER_MASK_RASTER = 1, + BARRIER_MASK_COMPUTE = 2, + BARRIER_MASK_TRANSFER = 4, + BARRIER_MASK_NO_BARRIER = 8, + BARRIER_MASK_ALL = BARRIER_MASK_RASTER | BARRIER_MASK_COMPUTE | BARRIER_MASK_TRANSFER + }; + + /*****************/ /**** TEXTURE ****/ /*****************/ @@ -438,16 +481,16 @@ public: virtual RID texture_create_shared_from_slice(const TextureView &p_view, RID p_with_texture, uint32_t p_layer, uint32_t p_mipmap, TextureSliceType p_slice_type = TEXTURE_SLICE_2D) = 0; - virtual Error texture_update(RID p_texture, uint32_t p_layer, const Vector<uint8_t> &p_data, bool p_sync_with_draw = false) = 0; //this function can be used from any thread and it takes effect at the beginning of the frame, unless sync with draw is used, which is used to mix updates with draw calls + virtual Error texture_update(RID p_texture, uint32_t p_layer, const Vector<uint8_t> &p_data, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; virtual Vector<uint8_t> texture_get_data(RID p_texture, uint32_t p_layer) = 0; // CPU textures will return immediately, while GPU textures will most likely force a flush virtual bool texture_is_format_supported_for_usage(DataFormat p_format, uint32_t p_usage) const = 0; virtual bool texture_is_shared(RID p_texture) = 0; virtual bool texture_is_valid(RID p_texture) = 0; - virtual Error texture_copy(RID p_from_texture, RID p_to_texture, const Vector3 &p_from, const Vector3 &p_to, const Vector3 &p_size, uint32_t p_src_mipmap, uint32_t p_dst_mipmap, uint32_t p_src_layer, uint32_t p_dst_layer, bool p_sync_with_draw = false) = 0; - virtual Error texture_clear(RID p_texture, const Color &p_color, uint32_t p_base_mipmap, uint32_t p_mipmaps, uint32_t p_base_layer, uint32_t p_layers, bool p_sync_with_draw = false) = 0; - virtual Error texture_resolve_multisample(RID p_from_texture, RID p_to_texture, bool p_sync_with_draw = false) = 0; + virtual Error texture_copy(RID p_from_texture, RID p_to_texture, const Vector3 &p_from, const Vector3 &p_to, const Vector3 &p_size, uint32_t p_src_mipmap, uint32_t p_dst_mipmap, uint32_t p_src_layer, uint32_t p_dst_layer, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; + virtual Error texture_clear(RID p_texture, const Color &p_color, uint32_t p_base_mipmap, uint32_t p_mipmaps, uint32_t p_base_layer, uint32_t p_layers, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; + virtual Error texture_resolve_multisample(RID p_from_texture, RID p_to_texture, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; /*********************/ /**** FRAMEBUFFER ****/ @@ -468,11 +511,11 @@ public: // This ID is warranted to be unique for the same formats, does not need to be freed virtual FramebufferFormatID framebuffer_format_create(const Vector<AttachmentFormat> &p_format) = 0; - virtual FramebufferFormatID framebuffer_format_create_empty(const Size2i &p_size) = 0; + virtual FramebufferFormatID framebuffer_format_create_empty(TextureSamples p_samples = TEXTURE_SAMPLES_1) = 0; virtual TextureSamples framebuffer_format_get_texture_samples(FramebufferFormatID p_format) = 0; virtual RID framebuffer_create(const Vector<RID> &p_texture_attachments, FramebufferFormatID p_format_check = INVALID_ID) = 0; - virtual RID framebuffer_create_empty(const Size2i &p_size, FramebufferFormatID p_format_check = INVALID_ID) = 0; + virtual RID framebuffer_create_empty(const Size2i &p_size, TextureSamples p_samples = TEXTURE_SAMPLES_1, FramebufferFormatID p_format_check = INVALID_ID) = 0; virtual FramebufferFormatID framebuffer_get_format(RID p_framebuffer) = 0; @@ -585,6 +628,8 @@ public: /**** SHADER ****/ /****************/ + const Capabilities *get_device_capabilities() const { return &device_capabilities; }; + virtual Vector<uint8_t> shader_compile_from_source(ShaderStage p_stage, const String &p_source_code, ShaderLanguage p_language = SHADER_LANGUAGE_GLSL, String *r_error = nullptr, bool p_allow_cache = true); static void shader_set_compile_function(ShaderCompileFunction p_function); @@ -649,7 +694,8 @@ public: virtual RID uniform_set_create(const Vector<Uniform> &p_uniforms, RID p_shader, uint32_t p_shader_set) = 0; virtual bool uniform_set_is_valid(RID p_uniform_set) = 0; - virtual Error buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const void *p_data, bool p_sync_with_draw = false) = 0; //this function can be used from any thread and it takes effect at the beginning of the frame, unless sync with draw is used, which is used to mix updates with draw calls + virtual Error buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const void *p_data, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; + virtual Error buffer_clear(RID p_buffer, uint32_t p_offset, uint32_t p_size, uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; virtual Vector<uint8_t> buffer_get_data(RID p_buffer) = 0; //this causes stall, only use to retrieve large buffers for saving /*************************/ @@ -930,7 +976,9 @@ public: /********************/ enum InitialAction { - INITIAL_ACTION_CLEAR, //start rendering and clear the framebuffer (supply params) + INITIAL_ACTION_CLEAR, //start rendering and clear the whole framebuffer (region or not) (supply params) + INITIAL_ACTION_CLEAR_REGION, //start rendering and clear the framebuffer in the specified region (supply params) + INITIAL_ACTION_CLEAR_REGION_CONTINUE, //countinue rendering and clear the framebuffer in the specified region (supply params) INITIAL_ACTION_KEEP, //start rendering, but keep attached color texture contents (depth will be cleared) INITIAL_ACTION_DROP, //start rendering, ignore what is there, just write above it INITIAL_ACTION_CONTINUE, //continue rendering (framebuffer must have been left in "continue" state as final action previously) @@ -962,7 +1010,7 @@ public: virtual void draw_list_enable_scissor(DrawListID p_list, const Rect2 &p_rect) = 0; virtual void draw_list_disable_scissor(DrawListID p_list) = 0; - virtual void draw_list_end() = 0; + virtual void draw_list_end(uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; /***********************/ /**** COMPUTE LISTS ****/ @@ -970,17 +1018,18 @@ public: typedef int64_t ComputeListID; - virtual ComputeListID compute_list_begin() = 0; + virtual ComputeListID compute_list_begin(bool p_allow_draw_overlap = false) = 0; virtual void compute_list_bind_compute_pipeline(ComputeListID p_list, RID p_compute_pipeline) = 0; virtual void compute_list_bind_uniform_set(ComputeListID p_list, RID p_uniform_set, uint32_t p_index) = 0; virtual void compute_list_set_push_constant(ComputeListID p_list, const void *p_data, uint32_t p_data_size) = 0; virtual void compute_list_dispatch(ComputeListID p_list, uint32_t p_x_groups, uint32_t p_y_groups, uint32_t p_z_groups) = 0; - virtual void compute_list_dispatch_threads(ComputeListID p_list, uint32_t p_x_threads, uint32_t p_y_threads, uint32_t p_z_threads, uint32_t p_x_local_group, uint32_t p_y_local_group, uint32_t p_z_local_group); + virtual void compute_list_dispatch_threads(ComputeListID p_list, uint32_t p_x_threads, uint32_t p_y_threads, uint32_t p_z_threads) = 0; virtual void compute_list_dispatch_indirect(ComputeListID p_list, RID p_buffer, uint32_t p_offset) = 0; virtual void compute_list_add_barrier(ComputeListID p_list) = 0; - virtual void compute_list_end() = 0; + virtual void compute_list_end(uint32_t p_post_barrier = BARRIER_MASK_ALL) = 0; + virtual void barrier(uint32_t p_from = BARRIER_MASK_ALL, uint32_t p_to = BARRIER_MASK_ALL) = 0; virtual void full_barrier() = 0; /***************/ @@ -993,7 +1042,7 @@ public: /**** Timing ****/ /****************/ - virtual void capture_timestamp(const String &p_name, bool p_sync_to_draw) = 0; + virtual void capture_timestamp(const String &p_name) = 0; virtual uint32_t get_captured_timestamps_count() const = 0; virtual uint64_t get_captured_timestamps_frame() const = 0; virtual uint64_t get_captured_timestamp_gpu_time(uint32_t p_index) const = 0; @@ -1058,6 +1107,16 @@ public: virtual RenderingDevice *create_local_device() = 0; + virtual void set_resource_name(RID p_id, const String p_name) = 0; + + virtual void draw_command_begin_label(String p_label_name, const Color p_color = Color(1, 1, 1, 1)) = 0; + virtual void draw_command_insert_label(String p_label_name, const Color p_color = Color(1, 1, 1, 1)) = 0; + virtual void draw_command_end_label() = 0; + + virtual String get_device_vendor_name() const = 0; + virtual String get_device_name() const = 0; + virtual String get_device_pipeline_cache_uuid() const = 0; + static RenderingDevice *get_singleton(); RenderingDevice(); @@ -1077,7 +1136,7 @@ protected: RID _uniform_set_create(const Array &p_uniforms, RID p_shader, uint32_t p_shader_set); - Error _buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const Vector<uint8_t> &p_data, bool p_sync_with_draw = false); + Error _buffer_update(RID p_buffer, uint32_t p_offset, uint32_t p_size, const Vector<uint8_t> &p_data, uint32_t p_post_barrier = BARRIER_MASK_ALL); RID _render_pipeline_create(RID p_shader, FramebufferFormatID p_framebuffer_format, VertexFormatID p_vertex_format, RenderPrimitive p_render_primitive, const Ref<RDPipelineRasterizationState> &p_rasterization_state, const Ref<RDPipelineMultisampleState> &p_multisample_state, const Ref<RDPipelineDepthStencilState> &p_depth_stencil_state, const Ref<RDPipelineColorBlendState> &p_blend_state, int p_dynamic_state_flags = 0); diff --git a/servers/rendering/rendering_server_default.cpp b/servers/rendering/rendering_server_default.cpp index 8c6e97a0af..c6fe6a07e0 100644 --- a/servers/rendering/rendering_server_default.cpp +++ b/servers/rendering/rendering_server_default.cpp @@ -64,7 +64,7 @@ void RenderingServerDefault::_draw_margins() { /* FREE */ -void RenderingServerDefault::free(RID p_rid) { +void RenderingServerDefault::_free(RID p_rid) { if (RSG::storage->free(p_rid)) { return; } @@ -91,7 +91,7 @@ void RenderingServerDefault::request_frame_drawn_callback(Object *p_where, const frame_drawn_callbacks.push_back(fdc); } -void RenderingServerDefault::draw(bool p_swap_buffers, double frame_step) { +void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { //needs to be done before changes is reset to 0, to not force the editor to redraw RS::get_singleton()->emit_signal("frame_pre_draw"); @@ -162,24 +162,66 @@ void RenderingServerDefault::draw(bool p_swap_buffers, double frame_step) { } frame_profile_frame = RSG::storage->get_captured_timestamps_frame(); + + if (print_gpu_profile) { + if (print_frame_profile_ticks_from == 0) { + print_frame_profile_ticks_from = OS::get_singleton()->get_ticks_usec(); + } + float total_time = 0.0; + + for (int i = 0; i < frame_profile.size() - 1; i++) { + String name = frame_profile[i].name; + if (name[0] == '<' || name[0] == '>') { + continue; + } + + float time = frame_profile[i + 1].gpu_msec - frame_profile[i].gpu_msec; + + if (name[0] != '<' && name[0] != '>') { + if (print_gpu_profile_task_time.has(name)) { + print_gpu_profile_task_time[name] += time; + } else { + print_gpu_profile_task_time[name] = time; + } + } + } + + if (frame_profile.size()) { + total_time = frame_profile[frame_profile.size() - 1].gpu_msec; + } + + uint64_t ticks_elapsed = OS::get_singleton()->get_ticks_usec() - print_frame_profile_ticks_from; + print_frame_profile_frame_count++; + if (ticks_elapsed > 1000000) { + print_line("GPU PROFILE (total " + rtos(total_time) + "ms): "); + + float print_threshold = 0.01; + for (OrderedHashMap<String, float>::Element E = print_gpu_profile_task_time.front(); E; E = E.next()) { + float time = E.value() / float(print_frame_profile_frame_count); + if (time > print_threshold) { + print_line("\t-" + E.key() + ": " + rtos(time) + "ms"); + } + } + print_gpu_profile_task_time.clear(); + print_frame_profile_ticks_from = OS::get_singleton()->get_ticks_usec(); + print_frame_profile_frame_count = 0; + } + } } float RenderingServerDefault::get_frame_setup_time_cpu() const { return frame_setup_time; } -void RenderingServerDefault::sync() { -} - bool RenderingServerDefault::has_changed() const { return changes > 0; } -void RenderingServerDefault::init() { +void RenderingServerDefault::_init() { RSG::rasterizer->initialize(); } -void RenderingServerDefault::finish() { +void RenderingServerDefault::_finish() { if (test_cube.is_valid()) { free(test_cube); } @@ -187,9 +229,35 @@ void RenderingServerDefault::finish() { RSG::rasterizer->finalize(); } +void RenderingServerDefault::init() { + if (create_thread) { + print_verbose("RenderingServerWrapMT: Creating render thread"); + DisplayServer::get_singleton()->release_rendering_thread(); + if (create_thread) { + thread.start(_thread_callback, this); + print_verbose("RenderingServerWrapMT: Starting render thread"); + } + while (!draw_thread_up.is_set()) { + OS::get_singleton()->delay_usec(1000); + } + print_verbose("RenderingServerWrapMT: Finished render thread"); + } else { + _init(); + } +} + +void RenderingServerDefault::finish() { + if (create_thread) { + command_queue.push(this, &RenderingServerDefault::_thread_exit); + thread.wait_to_finish(); + } else { + _finish(); + } +} + /* STATUS INFORMATION */ -int RenderingServerDefault::get_render_info(RenderInfo p_info) { +uint64_t RenderingServerDefault::get_render_info(RenderInfo p_info) { return RSG::storage->get_render_info(p_info); } @@ -232,6 +300,11 @@ void RenderingServerDefault::sdfgi_set_debug_probe_select(const Vector3 &p_posit RSG::scene->sdfgi_set_debug_probe_select(p_position, p_dir); } +void RenderingServerDefault::set_print_gpu_profile(bool p_enable) { + RSG::storage->capturing_timestamps = p_enable; + print_gpu_profile = p_enable; +} + RID RenderingServerDefault::get_test_cube() { if (!test_cube.is_valid()) { test_cube = _make_test_cube(); @@ -247,10 +320,6 @@ void RenderingServerDefault::set_debug_generate_wireframes(bool p_generate) { RSG::storage->set_debug_generate_wireframes(p_generate); } -void RenderingServerDefault::call_set_use_vsync(bool p_enable) { - DisplayServer::get_singleton()->_set_use_vsync(p_enable); -} - bool RenderingServerDefault::is_low_end() const { // FIXME: Commented out when rebasing vulkan branch on master, // causes a crash, it seems rasterizer is not initialized yet the @@ -259,7 +328,74 @@ bool RenderingServerDefault::is_low_end() const { return false; } -RenderingServerDefault::RenderingServerDefault() { +void RenderingServerDefault::_thread_exit() { + exit.set(); +} + +void RenderingServerDefault::_thread_draw(bool p_swap_buffers, double frame_step) { + if (!draw_pending.decrement()) { + _draw(p_swap_buffers, frame_step); + } +} + +void RenderingServerDefault::_thread_flush() { + draw_pending.decrement(); +} + +void RenderingServerDefault::_thread_callback(void *_instance) { + RenderingServerDefault *vsmt = reinterpret_cast<RenderingServerDefault *>(_instance); + + vsmt->_thread_loop(); +} + +void RenderingServerDefault::_thread_loop() { + server_thread = Thread::get_caller_id(); + + DisplayServer::get_singleton()->make_rendering_thread(); + + _init(); + + draw_thread_up.set(); + while (!exit.is_set()) { + // flush commands one by one, until exit is requested + command_queue.wait_and_flush_one(); + } + + command_queue.flush_all(); // flush all + + _finish(); +} + +/* EVENT QUEUING */ + +void RenderingServerDefault::sync() { + if (create_thread) { + draw_pending.increment(); + command_queue.push_and_sync(this, &RenderingServerDefault::_thread_flush); + } else { + command_queue.flush_all(); //flush all pending from other threads + } +} + +void RenderingServerDefault::draw(bool p_swap_buffers, double frame_step) { + if (create_thread) { + draw_pending.increment(); + command_queue.push(this, &RenderingServerDefault::_thread_draw, p_swap_buffers, frame_step); + } else { + _draw(p_swap_buffers, frame_step); + } +} + +RenderingServerDefault::RenderingServerDefault(bool p_create_thread) : + command_queue(p_create_thread) { + create_thread = p_create_thread; + + if (!p_create_thread) { + server_thread = Thread::get_caller_id(); + } else { + server_thread = 0; + } + RSG::canvas = memnew(RendererCanvasCull); RSG::viewport = memnew(RendererViewport); RendererSceneCull *sr = memnew(RendererSceneCull); diff --git a/servers/rendering/rendering_server_default.h b/servers/rendering/rendering_server_default.h index 3ccb5c4bbc..e82d5cc3f8 100644 --- a/servers/rendering/rendering_server_default.h +++ b/servers/rendering/rendering_server_default.h @@ -32,12 +32,15 @@ #define RENDERING_SERVER_DEFAULT_H #include "core/math/octree.h" +#include "core/templates/command_queue_mt.h" +#include "core/templates/ordered_hash_map.h" #include "renderer_canvas_cull.h" #include "renderer_scene_cull.h" #include "renderer_viewport.h" #include "rendering_server_globals.h" #include "servers/rendering/renderer_compositor.h" #include "servers/rendering_server.h" +#include "servers/server_wrap_mt_common.h" class RenderingServerDefault : public RenderingServer { enum { @@ -74,6 +77,37 @@ class RenderingServerDefault : public RenderingServer { float frame_setup_time = 0; + //for printing + bool print_gpu_profile = false; + OrderedHashMap<String, float> print_gpu_profile_task_time; + uint64_t print_frame_profile_ticks_from = 0; + uint32_t print_frame_profile_frame_count = 0; + + mutable CommandQueueMT command_queue; + + static void _thread_callback(void *_instance); + void _thread_loop(); + + Thread::ID server_thread; + SafeFlag exit; + Thread thread; + SafeFlag draw_thread_up; + bool create_thread; + + SafeNumeric<uint64_t> draw_pending; + void _thread_draw(bool p_swap_buffers, double frame_step); + void _thread_flush(); + + void _thread_exit(); + + Mutex alloc_mutex; + + void _draw(bool p_swap_buffers, double frame_step); + void _init(); + void _finish(); + + void _free(RID p_rid); + public: //if editor is redrawing when it shouldn't, enable this and put a breakpoint in _changes_changed() //#define DEBUG_CHANGES @@ -90,800 +124,813 @@ public: #else _FORCE_INLINE_ static void redraw_request() { changes++; } +#endif -#define DISPLAY_CHANGED \ - changes++; +#define WRITE_ACTION redraw_request(); + +#ifdef DEBUG_SYNC +#define SYNC_DEBUG print_line("sync on: " + String(__FUNCTION__)); +#else +#define SYNC_DEBUG #endif -#define BIND0R(m_r, m_name) \ - m_r m_name() { return BINDBASE->m_name(); } -#define BIND0RC(m_r, m_name) \ - m_r m_name() const { return BINDBASE->m_name(); } -#define BIND1R(m_r, m_name, m_type1) \ - m_r m_name(m_type1 arg1) { return BINDBASE->m_name(arg1); } -#define BIND1RC(m_r, m_name, m_type1) \ - m_r m_name(m_type1 arg1) const { return BINDBASE->m_name(arg1); } -#define BIND2R(m_r, m_name, m_type1, m_type2) \ - m_r m_name(m_type1 arg1, m_type2 arg2) { return BINDBASE->m_name(arg1, arg2); } -#define BIND2RC(m_r, m_name, m_type1, m_type2) \ - m_r m_name(m_type1 arg1, m_type2 arg2) const { return BINDBASE->m_name(arg1, arg2); } -#define BIND3R(m_r, m_name, m_type1, m_type2, m_type3) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3) { return BINDBASE->m_name(arg1, arg2, arg3); } -#define BIND3RC(m_r, m_name, m_type1, m_type2, m_type3) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3) const { return BINDBASE->m_name(arg1, arg2, arg3); } -#define BIND4R(m_r, m_name, m_type1, m_type2, m_type3, m_type4) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4) { return BINDBASE->m_name(arg1, arg2, arg3, arg4); } -#define BIND4RC(m_r, m_name, m_type1, m_type2, m_type3, m_type4) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4) const { return BINDBASE->m_name(arg1, arg2, arg3, arg4); } -#define BIND5R(m_r, m_name, m_type1, m_type2, m_type3, m_type4, m_type5) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5) { return BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5); } -#define BIND5RC(m_r, m_name, m_type1, m_type2, m_type3, m_type4, m_type5) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5) const { return BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5); } -#define BIND6R(m_r, m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6) { return BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6); } -#define BIND6RC(m_r, m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6) \ - m_r m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6) const { return BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6); } - -#define BIND0(m_name) \ - void m_name() { DISPLAY_CHANGED BINDBASE->m_name(); } -#define BIND1(m_name, m_type1) \ - void m_name(m_type1 arg1) { DISPLAY_CHANGED BINDBASE->m_name(arg1); } -#define BIND1C(m_name, m_type1) \ - void m_name(m_type1 arg1) const { DISPLAY_CHANGED BINDBASE->m_name(arg1); } -#define BIND2(m_name, m_type1, m_type2) \ - void m_name(m_type1 arg1, m_type2 arg2) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2); } -#define BIND2C(m_name, m_type1, m_type2) \ - void m_name(m_type1 arg1, m_type2 arg2) const { BINDBASE->m_name(arg1, arg2); } -#define BIND3(m_name, m_type1, m_type2, m_type3) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3); } -#define BIND4(m_name, m_type1, m_type2, m_type3, m_type4) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4); } -#define BIND5(m_name, m_type1, m_type2, m_type3, m_type4, m_type5) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5); } -#define BIND6(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6); } -#define BIND7(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7); } -#define BIND8(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8); } -#define BIND9(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9); } -#define BIND10(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10); } -#define BIND11(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10, m_type11) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10, m_type11 arg11) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11); } -#define BIND12(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10, m_type11, m_type12) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10, m_type11 arg11, m_type12 arg12) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12); } -#define BIND13(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10, m_type11, m_type12, m_type13) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10, m_type11 arg11, m_type12 arg12, m_type13 arg13) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13); } -#define BIND14(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10, m_type11, m_type12, m_type13, m_type14) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10, m_type11 arg11, m_type12 arg12, m_type13 arg13, m_type14 arg14) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14); } -#define BIND15(m_name, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6, m_type7, m_type8, m_type9, m_type10, m_type11, m_type12, m_type13, m_type14, m_type15) \ - void m_name(m_type1 arg1, m_type2 arg2, m_type3 arg3, m_type4 arg4, m_type5 arg5, m_type6 arg6, m_type7 arg7, m_type8 arg8, m_type9 arg9, m_type10 arg10, m_type11 arg11, m_type12 arg12, m_type13 arg13, m_type14 arg14, m_type15 arg15) { DISPLAY_CHANGED BINDBASE->m_name(arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15); } +#include "servers/server_wrap_mt_common.h" //from now on, calls forwarded to this singleton -#define BINDBASE RSG::storage +#define ServerName RendererStorage +#define server_name RSG::storage /* TEXTURE API */ +#define FUNCRIDTEX0(m_type) \ + virtual RID m_type##_create() override { \ + RID ret = RSG::storage->texture_allocate(); \ + if (Thread::get_caller_id() == server_thread || RSG::storage->can_create_resources_async()) { \ + RSG::storage->m_type##_initialize(ret); \ + } else { \ + command_queue.push(RSG::storage, &RendererStorage::m_type##_initialize, ret); \ + } \ + return ret; \ + } + +#define FUNCRIDTEX1(m_type, m_type1) \ + virtual RID m_type##_create(m_type1 p1) override { \ + RID ret = RSG::storage->texture_allocate(); \ + if (Thread::get_caller_id() == server_thread || RSG::storage->can_create_resources_async()) { \ + RSG::storage->m_type##_initialize(ret, p1); \ + } else { \ + command_queue.push(RSG::storage, &RendererStorage::m_type##_initialize, ret, p1); \ + } \ + return ret; \ + } + +#define FUNCRIDTEX2(m_type, m_type1, m_type2) \ + virtual RID m_type##_create(m_type1 p1, m_type2 p2) override { \ + RID ret = RSG::storage->texture_allocate(); \ + if (Thread::get_caller_id() == server_thread || RSG::storage->can_create_resources_async()) { \ + RSG::storage->m_type##_initialize(ret, p1, p2); \ + } else { \ + command_queue.push(RSG::storage, &RendererStorage::m_type##_initialize, ret, p1, p2); \ + } \ + return ret; \ + } + +#define FUNCRIDTEX6(m_type, m_type1, m_type2, m_type3, m_type4, m_type5, m_type6) \ + virtual RID m_type##_create(m_type1 p1, m_type2 p2, m_type3 p3, m_type4 p4, m_type5 p5, m_type6 p6) override { \ + RID ret = RSG::storage->texture_allocate(); \ + if (Thread::get_caller_id() == server_thread || RSG::storage->can_create_resources_async()) { \ + RSG::storage->m_type##_initialize(ret, p1, p2, p3, p4, p5, p6); \ + } else { \ + command_queue.push(RSG::storage, &RendererStorage::m_type##_initialize, ret, p1, p2, p3, p4, p5, p6); \ + } \ + return ret; \ + } + //these go pass-through, as they can be called from any thread - BIND1R(RID, texture_2d_create, const Ref<Image> &) - BIND2R(RID, texture_2d_layered_create, const Vector<Ref<Image>> &, TextureLayeredType) - BIND6R(RID, texture_3d_create, Image::Format, int, int, int, bool, const Vector<Ref<Image>> &) - BIND1R(RID, texture_proxy_create, RID) + FUNCRIDTEX1(texture_2d, const Ref<Image> &) + FUNCRIDTEX2(texture_2d_layered, const Vector<Ref<Image>> &, TextureLayeredType) + FUNCRIDTEX6(texture_3d, Image::Format, int, int, int, bool, const Vector<Ref<Image>> &) + FUNCRIDTEX1(texture_proxy, RID) //goes pass-through - BIND3(texture_2d_update_immediate, RID, const Ref<Image> &, int) + FUNC3(texture_2d_update_immediate, RID, const Ref<Image> &, int) //these go through command queue if they are in another thread - BIND3(texture_2d_update, RID, const Ref<Image> &, int) - BIND2(texture_3d_update, RID, const Vector<Ref<Image>> &) - BIND2(texture_proxy_update, RID, RID) + FUNC3(texture_2d_update, RID, const Ref<Image> &, int) + FUNC2(texture_3d_update, RID, const Vector<Ref<Image>> &) + FUNC2(texture_proxy_update, RID, RID) //these also go pass-through - BIND0R(RID, texture_2d_placeholder_create) - BIND1R(RID, texture_2d_layered_placeholder_create, TextureLayeredType) - BIND0R(RID, texture_3d_placeholder_create) + FUNCRIDTEX0(texture_2d_placeholder) + FUNCRIDTEX1(texture_2d_layered_placeholder, TextureLayeredType) + FUNCRIDTEX0(texture_3d_placeholder) - BIND1RC(Ref<Image>, texture_2d_get, RID) - BIND2RC(Ref<Image>, texture_2d_layer_get, RID, int) - BIND1RC(Vector<Ref<Image>>, texture_3d_get, RID) + FUNC1RC(Ref<Image>, texture_2d_get, RID) + FUNC2RC(Ref<Image>, texture_2d_layer_get, RID, int) + FUNC1RC(Vector<Ref<Image>>, texture_3d_get, RID) - BIND2(texture_replace, RID, RID) + FUNC2(texture_replace, RID, RID) - BIND3(texture_set_size_override, RID, int, int) + FUNC3(texture_set_size_override, RID, int, int) // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 - BIND2(texture_bind, RID, uint32_t) + FUNC2(texture_bind, RID, uint32_t) #endif - BIND3(texture_set_detect_3d_callback, RID, TextureDetectCallback, void *) - BIND3(texture_set_detect_normal_callback, RID, TextureDetectCallback, void *) - BIND3(texture_set_detect_roughness_callback, RID, TextureDetectRoughnessCallback, void *) + FUNC3(texture_set_detect_3d_callback, RID, TextureDetectCallback, void *) + FUNC3(texture_set_detect_normal_callback, RID, TextureDetectCallback, void *) + FUNC3(texture_set_detect_roughness_callback, RID, TextureDetectRoughnessCallback, void *) - BIND2(texture_set_path, RID, const String &) - BIND1RC(String, texture_get_path, RID) - BIND1(texture_debug_usage, List<TextureInfo> *) + FUNC2(texture_set_path, RID, const String &) + FUNC1RC(String, texture_get_path, RID) + FUNC1(texture_debug_usage, List<TextureInfo> *) - BIND2(texture_set_force_redraw_if_visible, RID, bool) + FUNC2(texture_set_force_redraw_if_visible, RID, bool) /* SHADER API */ - BIND0R(RID, shader_create) + FUNCRIDSPLIT(shader) - BIND2(shader_set_code, RID, const String &) - BIND1RC(String, shader_get_code, RID) + FUNC2(shader_set_code, RID, const String &) + FUNC1RC(String, shader_get_code, RID) - BIND2C(shader_get_param_list, RID, List<PropertyInfo> *) + FUNC2SC(shader_get_param_list, RID, List<PropertyInfo> *) - BIND3(shader_set_default_texture_param, RID, const StringName &, RID) - BIND2RC(RID, shader_get_default_texture_param, RID, const StringName &) - BIND2RC(Variant, shader_get_param_default, RID, const StringName &) + FUNC3(shader_set_default_texture_param, RID, const StringName &, RID) + FUNC2RC(RID, shader_get_default_texture_param, RID, const StringName &) + FUNC2RC(Variant, shader_get_param_default, RID, const StringName &) + + FUNC1RC(ShaderNativeSourceCode, shader_get_native_source_code, RID) /* COMMON MATERIAL API */ - BIND0R(RID, material_create) + FUNCRIDSPLIT(material) - BIND2(material_set_shader, RID, RID) + FUNC2(material_set_shader, RID, RID) - BIND3(material_set_param, RID, const StringName &, const Variant &) - BIND2RC(Variant, material_get_param, RID, const StringName &) + FUNC3(material_set_param, RID, const StringName &, const Variant &) + FUNC2RC(Variant, material_get_param, RID, const StringName &) - BIND2(material_set_render_priority, RID, int) - BIND2(material_set_next_pass, RID, RID) + FUNC2(material_set_render_priority, RID, int) + FUNC2(material_set_next_pass, RID, RID) /* MESH API */ - virtual RID mesh_create_from_surfaces(const Vector<SurfaceData> &p_surfaces, int p_blend_shape_count = 0) { - RID mesh = mesh_create(); - mesh_set_blend_shape_count(mesh, p_blend_shape_count); - for (int i = 0; i < p_surfaces.size(); i++) { - mesh_add_surface(mesh, p_surfaces[i]); + virtual RID mesh_create_from_surfaces(const Vector<SurfaceData> &p_surfaces, int p_blend_shape_count = 0) override { + RID mesh = RSG::storage->mesh_allocate(); + + if (Thread::get_caller_id() == server_thread || RSG::storage->can_create_resources_async()) { + if (Thread::get_caller_id() == server_thread) { + command_queue.flush_if_pending(); + } + RSG::storage->mesh_initialize(mesh); + RSG::storage->mesh_set_blend_shape_count(mesh, p_blend_shape_count); + for (int i = 0; i < p_surfaces.size(); i++) { + RSG::storage->mesh_add_surface(mesh, p_surfaces[i]); + } + } else { + command_queue.push(RSG::storage, &RendererStorage::mesh_initialize, mesh); + command_queue.push(RSG::storage, &RendererStorage::mesh_set_blend_shape_count, mesh, p_blend_shape_count); + for (int i = 0; i < p_surfaces.size(); i++) { + RSG::storage->mesh_add_surface(mesh, p_surfaces[i]); + command_queue.push(RSG::storage, &RendererStorage::mesh_add_surface, mesh, p_surfaces[i]); + } } + return mesh; } - BIND2(mesh_set_blend_shape_count, RID, int) + FUNC2(mesh_set_blend_shape_count, RID, int) - BIND0R(RID, mesh_create) + FUNCRIDSPLIT(mesh) - BIND2(mesh_add_surface, RID, const SurfaceData &) + FUNC2(mesh_add_surface, RID, const SurfaceData &) - BIND1RC(int, mesh_get_blend_shape_count, RID) + FUNC1RC(int, mesh_get_blend_shape_count, RID) - BIND2(mesh_set_blend_shape_mode, RID, BlendShapeMode) - BIND1RC(BlendShapeMode, mesh_get_blend_shape_mode, RID) + FUNC2(mesh_set_blend_shape_mode, RID, BlendShapeMode) + FUNC1RC(BlendShapeMode, mesh_get_blend_shape_mode, RID) - BIND4(mesh_surface_update_region, RID, int, int, const Vector<uint8_t> &) + FUNC4(mesh_surface_update_region, RID, int, int, const Vector<uint8_t> &) - BIND3(mesh_surface_set_material, RID, int, RID) - BIND2RC(RID, mesh_surface_get_material, RID, int) + FUNC3(mesh_surface_set_material, RID, int, RID) + FUNC2RC(RID, mesh_surface_get_material, RID, int) - BIND2RC(SurfaceData, mesh_get_surface, RID, int) + FUNC2RC(SurfaceData, mesh_get_surface, RID, int) - BIND1RC(int, mesh_get_surface_count, RID) + FUNC1RC(int, mesh_get_surface_count, RID) - BIND2(mesh_set_custom_aabb, RID, const AABB &) - BIND1RC(AABB, mesh_get_custom_aabb, RID) + FUNC2(mesh_set_custom_aabb, RID, const AABB &) + FUNC1RC(AABB, mesh_get_custom_aabb, RID) - BIND1(mesh_clear, RID) + FUNC2(mesh_set_shadow_mesh, RID, RID) + + FUNC1(mesh_clear, RID) /* MULTIMESH API */ - BIND0R(RID, multimesh_create) + FUNCRIDSPLIT(multimesh) - BIND5(multimesh_allocate, RID, int, MultimeshTransformFormat, bool, bool) - BIND1RC(int, multimesh_get_instance_count, RID) + FUNC5(multimesh_allocate_data, RID, int, MultimeshTransformFormat, bool, bool) + FUNC1RC(int, multimesh_get_instance_count, RID) - BIND2(multimesh_set_mesh, RID, RID) - BIND3(multimesh_instance_set_transform, RID, int, const Transform &) - BIND3(multimesh_instance_set_transform_2d, RID, int, const Transform2D &) - BIND3(multimesh_instance_set_color, RID, int, const Color &) - BIND3(multimesh_instance_set_custom_data, RID, int, const Color &) + FUNC2(multimesh_set_mesh, RID, RID) + FUNC3(multimesh_instance_set_transform, RID, int, const Transform &) + FUNC3(multimesh_instance_set_transform_2d, RID, int, const Transform2D &) + FUNC3(multimesh_instance_set_color, RID, int, const Color &) + FUNC3(multimesh_instance_set_custom_data, RID, int, const Color &) - BIND1RC(RID, multimesh_get_mesh, RID) - BIND1RC(AABB, multimesh_get_aabb, RID) + FUNC1RC(RID, multimesh_get_mesh, RID) + FUNC1RC(AABB, multimesh_get_aabb, RID) - BIND2RC(Transform, multimesh_instance_get_transform, RID, int) - BIND2RC(Transform2D, multimesh_instance_get_transform_2d, RID, int) - BIND2RC(Color, multimesh_instance_get_color, RID, int) - BIND2RC(Color, multimesh_instance_get_custom_data, RID, int) + FUNC2RC(Transform, multimesh_instance_get_transform, RID, int) + FUNC2RC(Transform2D, multimesh_instance_get_transform_2d, RID, int) + FUNC2RC(Color, multimesh_instance_get_color, RID, int) + FUNC2RC(Color, multimesh_instance_get_custom_data, RID, int) - BIND2(multimesh_set_buffer, RID, const Vector<float> &) - BIND1RC(Vector<float>, multimesh_get_buffer, RID) + FUNC2(multimesh_set_buffer, RID, const Vector<float> &) + FUNC1RC(Vector<float>, multimesh_get_buffer, RID) - BIND2(multimesh_set_visible_instances, RID, int) - BIND1RC(int, multimesh_get_visible_instances, RID) + FUNC2(multimesh_set_visible_instances, RID, int) + FUNC1RC(int, multimesh_get_visible_instances, RID) /* IMMEDIATE API */ - BIND0R(RID, immediate_create) - BIND3(immediate_begin, RID, PrimitiveType, RID) - BIND2(immediate_vertex, RID, const Vector3 &) - BIND2(immediate_normal, RID, const Vector3 &) - BIND2(immediate_tangent, RID, const Plane &) - BIND2(immediate_color, RID, const Color &) - BIND2(immediate_uv, RID, const Vector2 &) - BIND2(immediate_uv2, RID, const Vector2 &) - BIND1(immediate_end, RID) - BIND1(immediate_clear, RID) - BIND2(immediate_set_material, RID, RID) - BIND1RC(RID, immediate_get_material, RID) + FUNCRIDSPLIT(immediate) + FUNC3(immediate_begin, RID, PrimitiveType, RID) + FUNC2(immediate_vertex, RID, const Vector3 &) + FUNC2(immediate_normal, RID, const Vector3 &) + FUNC2(immediate_tangent, RID, const Plane &) + FUNC2(immediate_color, RID, const Color &) + FUNC2(immediate_uv, RID, const Vector2 &) + FUNC2(immediate_uv2, RID, const Vector2 &) + FUNC1(immediate_end, RID) + FUNC1(immediate_clear, RID) + FUNC2(immediate_set_material, RID, RID) + FUNC1RC(RID, immediate_get_material, RID) /* SKELETON API */ - BIND0R(RID, skeleton_create) - BIND3(skeleton_allocate, RID, int, bool) - BIND1RC(int, skeleton_get_bone_count, RID) - BIND3(skeleton_bone_set_transform, RID, int, const Transform &) - BIND2RC(Transform, skeleton_bone_get_transform, RID, int) - BIND3(skeleton_bone_set_transform_2d, RID, int, const Transform2D &) - BIND2RC(Transform2D, skeleton_bone_get_transform_2d, RID, int) - BIND2(skeleton_set_base_transform_2d, RID, const Transform2D &) + FUNCRIDSPLIT(skeleton) + FUNC3(skeleton_allocate_data, RID, int, bool) + FUNC1RC(int, skeleton_get_bone_count, RID) + FUNC3(skeleton_bone_set_transform, RID, int, const Transform &) + FUNC2RC(Transform, skeleton_bone_get_transform, RID, int) + FUNC3(skeleton_bone_set_transform_2d, RID, int, const Transform2D &) + FUNC2RC(Transform2D, skeleton_bone_get_transform_2d, RID, int) + FUNC2(skeleton_set_base_transform_2d, RID, const Transform2D &) /* Light API */ - BIND0R(RID, directional_light_create) - BIND0R(RID, omni_light_create) - BIND0R(RID, spot_light_create) + FUNCRIDSPLIT(directional_light) + FUNCRIDSPLIT(omni_light) + FUNCRIDSPLIT(spot_light) - BIND2(light_set_color, RID, const Color &) - BIND3(light_set_param, RID, LightParam, float) - BIND2(light_set_shadow, RID, bool) - BIND2(light_set_shadow_color, RID, const Color &) - BIND2(light_set_projector, RID, RID) - BIND2(light_set_negative, RID, bool) - BIND2(light_set_cull_mask, RID, uint32_t) - BIND2(light_set_reverse_cull_face_mode, RID, bool) - BIND2(light_set_bake_mode, RID, LightBakeMode) - BIND2(light_set_max_sdfgi_cascade, RID, uint32_t) + FUNC2(light_set_color, RID, const Color &) + FUNC3(light_set_param, RID, LightParam, float) + FUNC2(light_set_shadow, RID, bool) + FUNC2(light_set_shadow_color, RID, const Color &) + FUNC2(light_set_projector, RID, RID) + FUNC2(light_set_negative, RID, bool) + FUNC2(light_set_cull_mask, RID, uint32_t) + FUNC2(light_set_reverse_cull_face_mode, RID, bool) + FUNC2(light_set_bake_mode, RID, LightBakeMode) + FUNC2(light_set_max_sdfgi_cascade, RID, uint32_t) - BIND2(light_omni_set_shadow_mode, RID, LightOmniShadowMode) + FUNC2(light_omni_set_shadow_mode, RID, LightOmniShadowMode) - BIND2(light_directional_set_shadow_mode, RID, LightDirectionalShadowMode) - BIND2(light_directional_set_blend_splits, RID, bool) - BIND2(light_directional_set_sky_only, RID, bool) - BIND2(light_directional_set_shadow_depth_range_mode, RID, LightDirectionalShadowDepthRangeMode) + FUNC2(light_directional_set_shadow_mode, RID, LightDirectionalShadowMode) + FUNC2(light_directional_set_blend_splits, RID, bool) + FUNC2(light_directional_set_sky_only, RID, bool) + FUNC2(light_directional_set_shadow_depth_range_mode, RID, LightDirectionalShadowDepthRangeMode) /* PROBE API */ - BIND0R(RID, reflection_probe_create) - - BIND2(reflection_probe_set_update_mode, RID, ReflectionProbeUpdateMode) - BIND2(reflection_probe_set_intensity, RID, float) - BIND2(reflection_probe_set_ambient_color, RID, const Color &) - BIND2(reflection_probe_set_ambient_energy, RID, float) - BIND2(reflection_probe_set_ambient_mode, RID, ReflectionProbeAmbientMode) - BIND2(reflection_probe_set_max_distance, RID, float) - BIND2(reflection_probe_set_extents, RID, const Vector3 &) - BIND2(reflection_probe_set_origin_offset, RID, const Vector3 &) - BIND2(reflection_probe_set_as_interior, RID, bool) - BIND2(reflection_probe_set_enable_box_projection, RID, bool) - BIND2(reflection_probe_set_enable_shadows, RID, bool) - BIND2(reflection_probe_set_cull_mask, RID, uint32_t) - BIND2(reflection_probe_set_resolution, RID, int) - BIND2(reflection_probe_set_lod_threshold, RID, float) + FUNCRIDSPLIT(reflection_probe) + + FUNC2(reflection_probe_set_update_mode, RID, ReflectionProbeUpdateMode) + FUNC2(reflection_probe_set_intensity, RID, float) + FUNC2(reflection_probe_set_ambient_color, RID, const Color &) + FUNC2(reflection_probe_set_ambient_energy, RID, float) + FUNC2(reflection_probe_set_ambient_mode, RID, ReflectionProbeAmbientMode) + FUNC2(reflection_probe_set_max_distance, RID, float) + FUNC2(reflection_probe_set_extents, RID, const Vector3 &) + FUNC2(reflection_probe_set_origin_offset, RID, const Vector3 &) + FUNC2(reflection_probe_set_as_interior, RID, bool) + FUNC2(reflection_probe_set_enable_box_projection, RID, bool) + FUNC2(reflection_probe_set_enable_shadows, RID, bool) + FUNC2(reflection_probe_set_cull_mask, RID, uint32_t) + FUNC2(reflection_probe_set_resolution, RID, int) + FUNC2(reflection_probe_set_lod_threshold, RID, float) /* DECAL API */ - BIND0R(RID, decal_create) + FUNCRIDSPLIT(decal) - BIND2(decal_set_extents, RID, const Vector3 &) - BIND3(decal_set_texture, RID, DecalTexture, RID) - BIND2(decal_set_emission_energy, RID, float) - BIND2(decal_set_albedo_mix, RID, float) - BIND2(decal_set_modulate, RID, const Color &) - BIND2(decal_set_cull_mask, RID, uint32_t) - BIND4(decal_set_distance_fade, RID, bool, float, float) - BIND3(decal_set_fade, RID, float, float) - BIND2(decal_set_normal_fade, RID, float) + FUNC2(decal_set_extents, RID, const Vector3 &) + FUNC3(decal_set_texture, RID, DecalTexture, RID) + FUNC2(decal_set_emission_energy, RID, float) + FUNC2(decal_set_albedo_mix, RID, float) + FUNC2(decal_set_modulate, RID, const Color &) + FUNC2(decal_set_cull_mask, RID, uint32_t) + FUNC4(decal_set_distance_fade, RID, bool, float, float) + FUNC3(decal_set_fade, RID, float, float) + FUNC2(decal_set_normal_fade, RID, float) /* BAKED LIGHT API */ - BIND0R(RID, gi_probe_create) + FUNCRIDSPLIT(gi_probe) - BIND8(gi_probe_allocate, RID, const Transform &, const AABB &, const Vector3i &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<int> &) + FUNC8(gi_probe_allocate_data, RID, const Transform &, const AABB &, const Vector3i &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<int> &) - BIND1RC(AABB, gi_probe_get_bounds, RID) - BIND1RC(Vector3i, gi_probe_get_octree_size, RID) - BIND1RC(Vector<uint8_t>, gi_probe_get_octree_cells, RID) - BIND1RC(Vector<uint8_t>, gi_probe_get_data_cells, RID) - BIND1RC(Vector<uint8_t>, gi_probe_get_distance_field, RID) - BIND1RC(Vector<int>, gi_probe_get_level_counts, RID) - BIND1RC(Transform, gi_probe_get_to_cell_xform, RID) + FUNC1RC(AABB, gi_probe_get_bounds, RID) + FUNC1RC(Vector3i, gi_probe_get_octree_size, RID) + FUNC1RC(Vector<uint8_t>, gi_probe_get_octree_cells, RID) + FUNC1RC(Vector<uint8_t>, gi_probe_get_data_cells, RID) + FUNC1RC(Vector<uint8_t>, gi_probe_get_distance_field, RID) + FUNC1RC(Vector<int>, gi_probe_get_level_counts, RID) + FUNC1RC(Transform, gi_probe_get_to_cell_xform, RID) - BIND2(gi_probe_set_dynamic_range, RID, float) - BIND1RC(float, gi_probe_get_dynamic_range, RID) + FUNC2(gi_probe_set_dynamic_range, RID, float) + FUNC1RC(float, gi_probe_get_dynamic_range, RID) - BIND2(gi_probe_set_propagation, RID, float) - BIND1RC(float, gi_probe_get_propagation, RID) + FUNC2(gi_probe_set_propagation, RID, float) + FUNC1RC(float, gi_probe_get_propagation, RID) - BIND2(gi_probe_set_energy, RID, float) - BIND1RC(float, gi_probe_get_energy, RID) + FUNC2(gi_probe_set_energy, RID, float) + FUNC1RC(float, gi_probe_get_energy, RID) - BIND2(gi_probe_set_ao, RID, float) - BIND1RC(float, gi_probe_get_ao, RID) + FUNC2(gi_probe_set_ao, RID, float) + FUNC1RC(float, gi_probe_get_ao, RID) - BIND2(gi_probe_set_ao_size, RID, float) - BIND1RC(float, gi_probe_get_ao_size, RID) + FUNC2(gi_probe_set_ao_size, RID, float) + FUNC1RC(float, gi_probe_get_ao_size, RID) - BIND2(gi_probe_set_bias, RID, float) - BIND1RC(float, gi_probe_get_bias, RID) + FUNC2(gi_probe_set_bias, RID, float) + FUNC1RC(float, gi_probe_get_bias, RID) - BIND2(gi_probe_set_normal_bias, RID, float) - BIND1RC(float, gi_probe_get_normal_bias, RID) + FUNC2(gi_probe_set_normal_bias, RID, float) + FUNC1RC(float, gi_probe_get_normal_bias, RID) - BIND2(gi_probe_set_interior, RID, bool) - BIND1RC(bool, gi_probe_is_interior, RID) + FUNC2(gi_probe_set_interior, RID, bool) + FUNC1RC(bool, gi_probe_is_interior, RID) - BIND2(gi_probe_set_use_two_bounces, RID, bool) - BIND1RC(bool, gi_probe_is_using_two_bounces, RID) + FUNC2(gi_probe_set_use_two_bounces, RID, bool) + FUNC1RC(bool, gi_probe_is_using_two_bounces, RID) - BIND2(gi_probe_set_anisotropy_strength, RID, float) - BIND1RC(float, gi_probe_get_anisotropy_strength, RID) + FUNC2(gi_probe_set_anisotropy_strength, RID, float) + FUNC1RC(float, gi_probe_get_anisotropy_strength, RID) /* LIGHTMAP */ - BIND0R(RID, lightmap_create) + FUNCRIDSPLIT(lightmap) - BIND3(lightmap_set_textures, RID, RID, bool) - BIND2(lightmap_set_probe_bounds, RID, const AABB &) - BIND2(lightmap_set_probe_interior, RID, bool) - BIND5(lightmap_set_probe_capture_data, RID, const PackedVector3Array &, const PackedColorArray &, const PackedInt32Array &, const PackedInt32Array &) - BIND1RC(PackedVector3Array, lightmap_get_probe_capture_points, RID) - BIND1RC(PackedColorArray, lightmap_get_probe_capture_sh, RID) - BIND1RC(PackedInt32Array, lightmap_get_probe_capture_tetrahedra, RID) - BIND1RC(PackedInt32Array, lightmap_get_probe_capture_bsp_tree, RID) - BIND1(lightmap_set_probe_capture_update_speed, float) + FUNC3(lightmap_set_textures, RID, RID, bool) + FUNC2(lightmap_set_probe_bounds, RID, const AABB &) + FUNC2(lightmap_set_probe_interior, RID, bool) + FUNC5(lightmap_set_probe_capture_data, RID, const PackedVector3Array &, const PackedColorArray &, const PackedInt32Array &, const PackedInt32Array &) + FUNC1RC(PackedVector3Array, lightmap_get_probe_capture_points, RID) + FUNC1RC(PackedColorArray, lightmap_get_probe_capture_sh, RID) + FUNC1RC(PackedInt32Array, lightmap_get_probe_capture_tetrahedra, RID) + FUNC1RC(PackedInt32Array, lightmap_get_probe_capture_bsp_tree, RID) + FUNC1(lightmap_set_probe_capture_update_speed, float) /* PARTICLES */ - BIND0R(RID, particles_create) - - BIND2(particles_set_emitting, RID, bool) - BIND1R(bool, particles_get_emitting, RID) - BIND2(particles_set_amount, RID, int) - BIND2(particles_set_lifetime, RID, float) - BIND2(particles_set_one_shot, RID, bool) - BIND2(particles_set_pre_process_time, RID, float) - BIND2(particles_set_explosiveness_ratio, RID, float) - BIND2(particles_set_randomness_ratio, RID, float) - BIND2(particles_set_custom_aabb, RID, const AABB &) - BIND2(particles_set_speed_scale, RID, float) - BIND2(particles_set_use_local_coordinates, RID, bool) - BIND2(particles_set_process_material, RID, RID) - BIND2(particles_set_fixed_fps, RID, int) - BIND2(particles_set_fractional_delta, RID, bool) - BIND1R(bool, particles_is_inactive, RID) - BIND1(particles_request_process, RID) - BIND1(particles_restart, RID) - BIND6(particles_emit, RID, const Transform &, const Vector3 &, const Color &, const Color &, uint32_t) - BIND2(particles_set_subemitter, RID, RID) - BIND2(particles_set_collision_base_size, RID, float) - - BIND2(particles_set_draw_order, RID, RS::ParticlesDrawOrder) - - BIND2(particles_set_draw_passes, RID, int) - BIND3(particles_set_draw_pass_mesh, RID, int, RID) - - BIND1R(AABB, particles_get_current_aabb, RID) - BIND2(particles_set_emission_transform, RID, const Transform &) + FUNCRIDSPLIT(particles) + + FUNC2(particles_set_emitting, RID, bool) + FUNC1R(bool, particles_get_emitting, RID) + FUNC2(particles_set_amount, RID, int) + FUNC2(particles_set_lifetime, RID, float) + FUNC2(particles_set_one_shot, RID, bool) + FUNC2(particles_set_pre_process_time, RID, float) + FUNC2(particles_set_explosiveness_ratio, RID, float) + FUNC2(particles_set_randomness_ratio, RID, float) + FUNC2(particles_set_custom_aabb, RID, const AABB &) + FUNC2(particles_set_speed_scale, RID, float) + FUNC2(particles_set_use_local_coordinates, RID, bool) + FUNC2(particles_set_process_material, RID, RID) + FUNC2(particles_set_fixed_fps, RID, int) + FUNC2(particles_set_fractional_delta, RID, bool) + FUNC1R(bool, particles_is_inactive, RID) + FUNC1(particles_request_process, RID) + FUNC1(particles_restart, RID) + FUNC6(particles_emit, RID, const Transform &, const Vector3 &, const Color &, const Color &, uint32_t) + FUNC2(particles_set_subemitter, RID, RID) + FUNC2(particles_set_collision_base_size, RID, float) + + FUNC2(particles_set_draw_order, RID, RS::ParticlesDrawOrder) + + FUNC2(particles_set_draw_passes, RID, int) + FUNC3(particles_set_draw_pass_mesh, RID, int, RID) + + FUNC1R(AABB, particles_get_current_aabb, RID) + FUNC2(particles_set_emission_transform, RID, const Transform &) /* PARTICLES COLLISION */ - BIND0R(RID, particles_collision_create) - - BIND2(particles_collision_set_collision_type, RID, ParticlesCollisionType) - BIND2(particles_collision_set_cull_mask, RID, uint32_t) - BIND2(particles_collision_set_sphere_radius, RID, float) - BIND2(particles_collision_set_box_extents, RID, const Vector3 &) - BIND2(particles_collision_set_attractor_strength, RID, float) - BIND2(particles_collision_set_attractor_directionality, RID, float) - BIND2(particles_collision_set_attractor_attenuation, RID, float) - BIND2(particles_collision_set_field_texture, RID, RID) - BIND1(particles_collision_height_field_update, RID) - BIND2(particles_collision_set_height_field_resolution, RID, ParticlesCollisionHeightfieldResolution) - -#undef BINDBASE + FUNCRIDSPLIT(particles_collision) + + FUNC2(particles_collision_set_collision_type, RID, ParticlesCollisionType) + FUNC2(particles_collision_set_cull_mask, RID, uint32_t) + FUNC2(particles_collision_set_sphere_radius, RID, float) + FUNC2(particles_collision_set_box_extents, RID, const Vector3 &) + FUNC2(particles_collision_set_attractor_strength, RID, float) + FUNC2(particles_collision_set_attractor_directionality, RID, float) + FUNC2(particles_collision_set_attractor_attenuation, RID, float) + FUNC2(particles_collision_set_field_texture, RID, RID) + FUNC1(particles_collision_height_field_update, RID) + FUNC2(particles_collision_set_height_field_resolution, RID, ParticlesCollisionHeightfieldResolution) + +#undef server_name +#undef ServerName //from now on, calls forwarded to this singleton -#define BINDBASE RSG::scene +#define ServerName RendererScene +#define server_name RSG::scene /* CAMERA API */ - BIND0R(RID, camera_create) - BIND4(camera_set_perspective, RID, float, float, float) - BIND4(camera_set_orthogonal, RID, float, float, float) - BIND5(camera_set_frustum, RID, float, Vector2, float, float) - BIND2(camera_set_transform, RID, const Transform &) - BIND2(camera_set_cull_mask, RID, uint32_t) - BIND2(camera_set_environment, RID, RID) - BIND2(camera_set_camera_effects, RID, RID) - BIND2(camera_set_use_vertical_aspect, RID, bool) - -#undef BINDBASE + FUNCRIDSPLIT(camera) + FUNC4(camera_set_perspective, RID, float, float, float) + FUNC4(camera_set_orthogonal, RID, float, float, float) + FUNC5(camera_set_frustum, RID, float, Vector2, float, float) + FUNC2(camera_set_transform, RID, const Transform &) + FUNC2(camera_set_cull_mask, RID, uint32_t) + FUNC2(camera_set_environment, RID, RID) + FUNC2(camera_set_camera_effects, RID, RID) + FUNC2(camera_set_use_vertical_aspect, RID, bool) + +#undef server_name +#undef ServerName //from now on, calls forwarded to this singleton -#define BINDBASE RSG::viewport +#define ServerName RendererViewport +#define server_name RSG::viewport /* VIEWPORT TARGET API */ - BIND0R(RID, viewport_create) + FUNCRIDSPLIT(viewport) + + FUNC2(viewport_set_use_xr, RID, bool) + FUNC3(viewport_set_size, RID, int, int) - BIND2(viewport_set_use_xr, RID, bool) - BIND3(viewport_set_size, RID, int, int) + FUNC2(viewport_set_active, RID, bool) + FUNC2(viewport_set_parent_viewport, RID, RID) - BIND2(viewport_set_active, RID, bool) - BIND2(viewport_set_parent_viewport, RID, RID) + FUNC2(viewport_set_clear_mode, RID, ViewportClearMode) - BIND2(viewport_set_clear_mode, RID, ViewportClearMode) + FUNC3(viewport_attach_to_screen, RID, const Rect2 &, int) + FUNC2(viewport_set_render_direct_to_screen, RID, bool) - BIND3(viewport_attach_to_screen, RID, const Rect2 &, int) - BIND2(viewport_set_render_direct_to_screen, RID, bool) + FUNC2(viewport_set_update_mode, RID, ViewportUpdateMode) - BIND2(viewport_set_update_mode, RID, ViewportUpdateMode) - BIND2(viewport_set_vflip, RID, bool) + FUNC1RC(RID, viewport_get_texture, RID) - BIND1RC(RID, viewport_get_texture, RID) + FUNC2(viewport_set_hide_scenario, RID, bool) + FUNC2(viewport_set_hide_canvas, RID, bool) + FUNC2(viewport_set_disable_environment, RID, bool) - BIND2(viewport_set_hide_scenario, RID, bool) - BIND2(viewport_set_hide_canvas, RID, bool) - BIND2(viewport_set_disable_environment, RID, bool) + FUNC2(viewport_attach_camera, RID, RID) + FUNC2(viewport_set_scenario, RID, RID) + FUNC2(viewport_attach_canvas, RID, RID) - BIND2(viewport_attach_camera, RID, RID) - BIND2(viewport_set_scenario, RID, RID) - BIND2(viewport_attach_canvas, RID, RID) + FUNC2(viewport_remove_canvas, RID, RID) + FUNC3(viewport_set_canvas_transform, RID, RID, const Transform2D &) + FUNC2(viewport_set_transparent_background, RID, bool) + FUNC2(viewport_set_snap_2d_transforms_to_pixel, RID, bool) + FUNC2(viewport_set_snap_2d_vertices_to_pixel, RID, bool) - BIND2(viewport_remove_canvas, RID, RID) - BIND3(viewport_set_canvas_transform, RID, RID, const Transform2D &) - BIND2(viewport_set_transparent_background, RID, bool) - BIND2(viewport_set_snap_2d_transforms_to_pixel, RID, bool) - BIND2(viewport_set_snap_2d_vertices_to_pixel, RID, bool) + FUNC2(viewport_set_default_canvas_item_texture_filter, RID, CanvasItemTextureFilter) + FUNC2(viewport_set_default_canvas_item_texture_repeat, RID, CanvasItemTextureRepeat) - BIND2(viewport_set_default_canvas_item_texture_filter, RID, CanvasItemTextureFilter) - BIND2(viewport_set_default_canvas_item_texture_repeat, RID, CanvasItemTextureRepeat) + FUNC2(viewport_set_global_canvas_transform, RID, const Transform2D &) + FUNC4(viewport_set_canvas_stacking, RID, RID, int, int) + FUNC3(viewport_set_shadow_atlas_size, RID, int, bool) + FUNC3(viewport_set_sdf_oversize_and_scale, RID, ViewportSDFOversize, ViewportSDFScale) + FUNC3(viewport_set_shadow_atlas_quadrant_subdivision, RID, int, int) + FUNC2(viewport_set_msaa, RID, ViewportMSAA) + FUNC2(viewport_set_screen_space_aa, RID, ViewportScreenSpaceAA) + FUNC2(viewport_set_use_debanding, RID, bool) + FUNC2(viewport_set_lod_threshold, RID, float) - BIND2(viewport_set_global_canvas_transform, RID, const Transform2D &) - BIND4(viewport_set_canvas_stacking, RID, RID, int, int) - BIND2(viewport_set_shadow_atlas_size, RID, int) - BIND3(viewport_set_sdf_oversize_and_scale, RID, ViewportSDFOversize, ViewportSDFScale) - BIND3(viewport_set_shadow_atlas_quadrant_subdivision, RID, int, int) - BIND2(viewport_set_msaa, RID, ViewportMSAA) - BIND2(viewport_set_screen_space_aa, RID, ViewportScreenSpaceAA) - BIND2(viewport_set_use_debanding, RID, bool) - BIND2(viewport_set_lod_threshold, RID, float) + FUNC2R(int, viewport_get_render_info, RID, ViewportRenderInfo) + FUNC2(viewport_set_debug_draw, RID, ViewportDebugDraw) - BIND2R(int, viewport_get_render_info, RID, ViewportRenderInfo) - BIND2(viewport_set_debug_draw, RID, ViewportDebugDraw) + FUNC2(viewport_set_measure_render_time, RID, bool) + FUNC1RC(float, viewport_get_measured_render_time_cpu, RID) + FUNC1RC(float, viewport_get_measured_render_time_gpu, RID) - BIND2(viewport_set_measure_render_time, RID, bool) - BIND1RC(float, viewport_get_measured_render_time_cpu, RID) - BIND1RC(float, viewport_get_measured_render_time_gpu, RID) + FUNC1(call_set_use_vsync, bool) /* ENVIRONMENT API */ -#undef BINDBASE +#undef server_name +#undef ServerName //from now on, calls forwarded to this singleton -#define BINDBASE RSG::scene +#define ServerName RendererScene +#define server_name RSG::scene - BIND1(directional_shadow_atlas_set_size, int) - BIND1(gi_probe_set_quality, GIProbeQuality) + FUNC2(directional_shadow_atlas_set_size, int, bool) + FUNC1(gi_probe_set_quality, GIProbeQuality) /* SKY API */ - BIND0R(RID, sky_create) - BIND2(sky_set_radiance_size, RID, int) - BIND2(sky_set_mode, RID, SkyMode) - BIND2(sky_set_material, RID, RID) - BIND4R(Ref<Image>, sky_bake_panorama, RID, float, bool, const Size2i &) + FUNCRIDSPLIT(sky) + FUNC2(sky_set_radiance_size, RID, int) + FUNC2(sky_set_mode, RID, SkyMode) + FUNC2(sky_set_material, RID, RID) + FUNC4R(Ref<Image>, sky_bake_panorama, RID, float, bool, const Size2i &) - BIND0R(RID, environment_create) + FUNCRIDSPLIT(environment) - BIND2(environment_set_background, RID, EnvironmentBG) - BIND2(environment_set_sky, RID, RID) - BIND2(environment_set_sky_custom_fov, RID, float) - BIND2(environment_set_sky_orientation, RID, const Basis &) - BIND2(environment_set_bg_color, RID, const Color &) - BIND2(environment_set_bg_energy, RID, float) - BIND2(environment_set_canvas_max_layer, RID, int) - BIND7(environment_set_ambient_light, RID, const Color &, EnvironmentAmbientSource, float, float, EnvironmentReflectionSource, const Color &) + FUNC2(environment_set_background, RID, EnvironmentBG) + FUNC2(environment_set_sky, RID, RID) + FUNC2(environment_set_sky_custom_fov, RID, float) + FUNC2(environment_set_sky_orientation, RID, const Basis &) + FUNC2(environment_set_bg_color, RID, const Color &) + FUNC2(environment_set_bg_energy, RID, float) + FUNC2(environment_set_canvas_max_layer, RID, int) + FUNC7(environment_set_ambient_light, RID, const Color &, EnvironmentAmbientSource, float, float, EnvironmentReflectionSource, const Color &) // FIXME: Disabled during Vulkan refactoring, should be ported. #if 0 - BIND2(environment_set_camera_feed_id, RID, int) + FUNC2(environment_set_camera_feed_id, RID, int) #endif - BIND6(environment_set_ssr, RID, bool, int, float, float, float) - BIND1(environment_set_ssr_roughness_quality, EnvironmentSSRRoughnessQuality) + FUNC6(environment_set_ssr, RID, bool, int, float, float, float) + FUNC1(environment_set_ssr_roughness_quality, EnvironmentSSRRoughnessQuality) - BIND10(environment_set_ssao, RID, bool, float, float, float, float, float, float, float, float) - BIND6(environment_set_ssao_quality, EnvironmentSSAOQuality, bool, float, int, float, float) + FUNC10(environment_set_ssao, RID, bool, float, float, float, float, float, float, float, float) + FUNC6(environment_set_ssao_quality, EnvironmentSSAOQuality, bool, float, int, float, float) - BIND11(environment_set_glow, RID, bool, Vector<float>, float, float, float, float, EnvironmentGlowBlendMode, float, float, float) - BIND1(environment_glow_set_use_bicubic_upscale, bool) - BIND1(environment_glow_set_use_high_quality, bool) + FUNC11(environment_set_glow, RID, bool, Vector<float>, float, float, float, float, EnvironmentGlowBlendMode, float, float, float) + FUNC1(environment_glow_set_use_bicubic_upscale, bool) + FUNC1(environment_glow_set_use_high_quality, bool) - BIND9(environment_set_tonemap, RID, EnvironmentToneMapper, float, float, bool, float, float, float, float) + FUNC9(environment_set_tonemap, RID, EnvironmentToneMapper, float, float, bool, float, float, float, float) - BIND7(environment_set_adjustment, RID, bool, float, float, float, bool, RID) + FUNC7(environment_set_adjustment, RID, bool, float, float, float, bool, RID) - BIND9(environment_set_fog, RID, bool, const Color &, float, float, float, float, float, float) - BIND9(environment_set_volumetric_fog, RID, bool, float, const Color &, float, float, float, float, EnvVolumetricFogShadowFilter) + FUNC9(environment_set_fog, RID, bool, const Color &, float, float, float, float, float, float) + FUNC10(environment_set_volumetric_fog, RID, bool, float, const Color &, float, float, float, float, bool, float) - BIND2(environment_set_volumetric_fog_volume_size, int, int) - BIND1(environment_set_volumetric_fog_filter_active, bool) - BIND1(environment_set_volumetric_fog_directional_shadow_shrink_size, int) - BIND1(environment_set_volumetric_fog_positional_shadow_shrink_size, int) + FUNC2(environment_set_volumetric_fog_volume_size, int, int) + FUNC1(environment_set_volumetric_fog_filter_active, bool) - BIND11(environment_set_sdfgi, RID, bool, EnvironmentSDFGICascades, float, EnvironmentSDFGIYScale, bool, bool, bool, float, float, float) - BIND1(environment_set_sdfgi_ray_count, EnvironmentSDFGIRayCount) - BIND1(environment_set_sdfgi_frames_to_converge, EnvironmentSDFGIFramesToConverge) + FUNC11(environment_set_sdfgi, RID, bool, EnvironmentSDFGICascades, float, EnvironmentSDFGIYScale, bool, float, bool, float, float, float) + FUNC1(environment_set_sdfgi_ray_count, EnvironmentSDFGIRayCount) + FUNC1(environment_set_sdfgi_frames_to_converge, EnvironmentSDFGIFramesToConverge) + FUNC1(environment_set_sdfgi_frames_to_update_light, EnvironmentSDFGIFramesToUpdateLight) - BIND3R(Ref<Image>, environment_bake_panorama, RID, bool, const Size2i &) + FUNC3R(Ref<Image>, environment_bake_panorama, RID, bool, const Size2i &) - BIND3(screen_space_roughness_limiter_set_active, bool, float, float) - BIND1(sub_surface_scattering_set_quality, SubSurfaceScatteringQuality) - BIND2(sub_surface_scattering_set_scale, float, float) + FUNC3(screen_space_roughness_limiter_set_active, bool, float, float) + FUNC1(sub_surface_scattering_set_quality, SubSurfaceScatteringQuality) + FUNC2(sub_surface_scattering_set_scale, float, float) /* CAMERA EFFECTS */ - BIND0R(RID, camera_effects_create) + FUNCRIDSPLIT(camera_effects) - BIND2(camera_effects_set_dof_blur_quality, DOFBlurQuality, bool) - BIND1(camera_effects_set_dof_blur_bokeh_shape, DOFBokehShape) + FUNC2(camera_effects_set_dof_blur_quality, DOFBlurQuality, bool) + FUNC1(camera_effects_set_dof_blur_bokeh_shape, DOFBokehShape) - BIND8(camera_effects_set_dof_blur, RID, bool, float, float, bool, float, float, float) - BIND3(camera_effects_set_custom_exposure, RID, bool, float) + FUNC8(camera_effects_set_dof_blur, RID, bool, float, float, bool, float, float, float) + FUNC3(camera_effects_set_custom_exposure, RID, bool, float) - BIND1(shadows_quality_set, ShadowQuality); - BIND1(directional_shadow_quality_set, ShadowQuality); + FUNC1(shadows_quality_set, ShadowQuality); + FUNC1(directional_shadow_quality_set, ShadowQuality); /* SCENARIO API */ -#undef BINDBASE -#define BINDBASE RSG::scene +#undef server_name +#undef ServerName - BIND0R(RID, scenario_create) +#define ServerName RendererScene +#define server_name RSG::scene - BIND2(scenario_set_debug, RID, ScenarioDebugMode) - BIND2(scenario_set_environment, RID, RID) - BIND2(scenario_set_camera_effects, RID, RID) - BIND2(scenario_set_fallback_environment, RID, RID) + FUNCRIDSPLIT(scenario) + + FUNC2(scenario_set_debug, RID, ScenarioDebugMode) + FUNC2(scenario_set_environment, RID, RID) + FUNC2(scenario_set_camera_effects, RID, RID) + FUNC2(scenario_set_fallback_environment, RID, RID) /* INSTANCING API */ - BIND0R(RID, instance_create) + FUNCRIDSPLIT(instance) - BIND2(instance_set_base, RID, RID) - BIND2(instance_set_scenario, RID, RID) - BIND2(instance_set_layer_mask, RID, uint32_t) - BIND2(instance_set_transform, RID, const Transform &) - BIND2(instance_attach_object_instance_id, RID, ObjectID) - BIND3(instance_set_blend_shape_weight, RID, int, float) - BIND3(instance_set_surface_material, RID, int, RID) - BIND2(instance_set_visible, RID, bool) + FUNC2(instance_set_base, RID, RID) + FUNC2(instance_set_scenario, RID, RID) + FUNC2(instance_set_layer_mask, RID, uint32_t) + FUNC2(instance_set_transform, RID, const Transform &) + FUNC2(instance_attach_object_instance_id, RID, ObjectID) + FUNC3(instance_set_blend_shape_weight, RID, int, float) + FUNC3(instance_set_surface_material, RID, int, RID) + FUNC2(instance_set_visible, RID, bool) - BIND2(instance_set_custom_aabb, RID, AABB) + FUNC2(instance_set_custom_aabb, RID, AABB) - BIND2(instance_attach_skeleton, RID, RID) - BIND2(instance_set_exterior, RID, bool) + FUNC2(instance_attach_skeleton, RID, RID) + FUNC2(instance_set_exterior, RID, bool) - BIND2(instance_set_extra_visibility_margin, RID, real_t) + FUNC2(instance_set_extra_visibility_margin, RID, real_t) // don't use these in a game! - BIND2RC(Vector<ObjectID>, instances_cull_aabb, const AABB &, RID) - BIND3RC(Vector<ObjectID>, instances_cull_ray, const Vector3 &, const Vector3 &, RID) - BIND2RC(Vector<ObjectID>, instances_cull_convex, const Vector<Plane> &, RID) + FUNC2RC(Vector<ObjectID>, instances_cull_aabb, const AABB &, RID) + FUNC3RC(Vector<ObjectID>, instances_cull_ray, const Vector3 &, const Vector3 &, RID) + FUNC2RC(Vector<ObjectID>, instances_cull_convex, const Vector<Plane> &, RID) + + FUNC3(instance_geometry_set_flag, RID, InstanceFlags, bool) + FUNC2(instance_geometry_set_cast_shadows_setting, RID, ShadowCastingSetting) + FUNC2(instance_geometry_set_material_override, RID, RID) - BIND3(instance_geometry_set_flag, RID, InstanceFlags, bool) - BIND2(instance_geometry_set_cast_shadows_setting, RID, ShadowCastingSetting) - BIND2(instance_geometry_set_material_override, RID, RID) + FUNC5(instance_geometry_set_draw_range, RID, float, float, float, float) + FUNC2(instance_geometry_set_as_instance_lod, RID, RID) + FUNC4(instance_geometry_set_lightmap, RID, RID, const Rect2 &, int) + FUNC2(instance_geometry_set_lod_bias, RID, float) - BIND5(instance_geometry_set_draw_range, RID, float, float, float, float) - BIND2(instance_geometry_set_as_instance_lod, RID, RID) - BIND4(instance_geometry_set_lightmap, RID, RID, const Rect2 &, int) - BIND2(instance_geometry_set_lod_bias, RID, float) + FUNC3(instance_geometry_set_shader_parameter, RID, const StringName &, const Variant &) + FUNC2RC(Variant, instance_geometry_get_shader_parameter, RID, const StringName &) + FUNC2RC(Variant, instance_geometry_get_shader_parameter_default_value, RID, const StringName &) + FUNC2C(instance_geometry_get_shader_parameter_list, RID, List<PropertyInfo> *) - BIND3(instance_geometry_set_shader_parameter, RID, const StringName &, const Variant &) - BIND2RC(Variant, instance_geometry_get_shader_parameter, RID, const StringName &) - BIND2RC(Variant, instance_geometry_get_shader_parameter_default_value, RID, const StringName &) - BIND2C(instance_geometry_get_shader_parameter_list, RID, List<PropertyInfo> *) + FUNC3R(TypedArray<Image>, bake_render_uv2, RID, const Vector<RID> &, const Size2i &) - BIND3R(TypedArray<Image>, bake_render_uv2, RID, const Vector<RID> &, const Size2i &) + FUNC1(gi_set_use_half_resolution, bool) -#undef BINDBASE +#undef server_name +#undef ServerName //from now on, calls forwarded to this singleton -#define BINDBASE RSG::canvas +#define ServerName RendererCanvasCull +#define server_name RSG::canvas /* CANVAS (2D) */ - BIND0R(RID, canvas_create) - BIND3(canvas_set_item_mirroring, RID, RID, const Point2 &) - BIND2(canvas_set_modulate, RID, const Color &) - BIND3(canvas_set_parent, RID, RID, float) - BIND1(canvas_set_disable_scale, bool) - - BIND0R(RID, canvas_texture_create) - BIND3(canvas_texture_set_channel, RID, CanvasTextureChannel, RID) - BIND3(canvas_texture_set_shading_parameters, RID, const Color &, float) - - BIND2(canvas_texture_set_texture_filter, RID, CanvasItemTextureFilter) - BIND2(canvas_texture_set_texture_repeat, RID, CanvasItemTextureRepeat) - - BIND0R(RID, canvas_item_create) - BIND2(canvas_item_set_parent, RID, RID) - - BIND2(canvas_item_set_default_texture_filter, RID, CanvasItemTextureFilter) - BIND2(canvas_item_set_default_texture_repeat, RID, CanvasItemTextureRepeat) - - BIND2(canvas_item_set_visible, RID, bool) - BIND2(canvas_item_set_light_mask, RID, int) - - BIND2(canvas_item_set_update_when_visible, RID, bool) - - BIND2(canvas_item_set_transform, RID, const Transform2D &) - BIND2(canvas_item_set_clip, RID, bool) - BIND2(canvas_item_set_distance_field_mode, RID, bool) - BIND3(canvas_item_set_custom_rect, RID, bool, const Rect2 &) - BIND2(canvas_item_set_modulate, RID, const Color &) - BIND2(canvas_item_set_self_modulate, RID, const Color &) - - BIND2(canvas_item_set_draw_behind_parent, RID, bool) - - BIND5(canvas_item_add_line, RID, const Point2 &, const Point2 &, const Color &, float) - BIND5(canvas_item_add_polyline, RID, const Vector<Point2> &, const Vector<Color> &, float, bool) - BIND4(canvas_item_add_multiline, RID, const Vector<Point2> &, const Vector<Color> &, float) - BIND3(canvas_item_add_rect, RID, const Rect2 &, const Color &) - BIND4(canvas_item_add_circle, RID, const Point2 &, float, const Color &) - BIND6(canvas_item_add_texture_rect, RID, const Rect2 &, RID, bool, const Color &, bool) - BIND7(canvas_item_add_texture_rect_region, RID, const Rect2 &, RID, const Rect2 &, const Color &, bool, bool) - BIND10(canvas_item_add_nine_patch, RID, const Rect2 &, const Rect2 &, RID, const Vector2 &, const Vector2 &, NinePatchAxisMode, NinePatchAxisMode, bool, const Color &) - BIND6(canvas_item_add_primitive, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID, float) - BIND5(canvas_item_add_polygon, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID) - BIND9(canvas_item_add_triangle_array, RID, const Vector<int> &, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, const Vector<int> &, const Vector<float> &, RID, int) - BIND5(canvas_item_add_mesh, RID, const RID &, const Transform2D &, const Color &, RID) - BIND3(canvas_item_add_multimesh, RID, RID, RID) - BIND3(canvas_item_add_particles, RID, RID, RID) - BIND2(canvas_item_add_set_transform, RID, const Transform2D &) - BIND2(canvas_item_add_clip_ignore, RID, bool) - BIND2(canvas_item_set_sort_children_by_y, RID, bool) - BIND2(canvas_item_set_z_index, RID, int) - BIND2(canvas_item_set_z_as_relative_to_parent, RID, bool) - BIND3(canvas_item_set_copy_to_backbuffer, RID, bool, const Rect2 &) - BIND2(canvas_item_attach_skeleton, RID, RID) - - BIND1(canvas_item_clear, RID) - BIND2(canvas_item_set_draw_index, RID, int) - - BIND2(canvas_item_set_material, RID, RID) - - BIND2(canvas_item_set_use_parent_material, RID, bool) - - BIND6(canvas_item_set_canvas_group_mode, RID, CanvasGroupMode, float, bool, float, bool) - - BIND0R(RID, canvas_light_create) - - BIND2(canvas_light_set_mode, RID, CanvasLightMode) - - BIND2(canvas_light_attach_to_canvas, RID, RID) - BIND2(canvas_light_set_enabled, RID, bool) - BIND2(canvas_light_set_texture_scale, RID, float) - BIND2(canvas_light_set_transform, RID, const Transform2D &) - BIND2(canvas_light_set_texture, RID, RID) - BIND2(canvas_light_set_texture_offset, RID, const Vector2 &) - BIND2(canvas_light_set_color, RID, const Color &) - BIND2(canvas_light_set_height, RID, float) - BIND2(canvas_light_set_energy, RID, float) - BIND3(canvas_light_set_z_range, RID, int, int) - BIND3(canvas_light_set_layer_range, RID, int, int) - BIND2(canvas_light_set_item_cull_mask, RID, int) - BIND2(canvas_light_set_item_shadow_cull_mask, RID, int) - BIND2(canvas_light_set_directional_distance, RID, float) - - BIND2(canvas_light_set_blend_mode, RID, CanvasLightBlendMode) - - BIND2(canvas_light_set_shadow_enabled, RID, bool) - BIND2(canvas_light_set_shadow_filter, RID, CanvasLightShadowFilter) - BIND2(canvas_light_set_shadow_color, RID, const Color &) - BIND2(canvas_light_set_shadow_smooth, RID, float) - - BIND0R(RID, canvas_light_occluder_create) - BIND2(canvas_light_occluder_attach_to_canvas, RID, RID) - BIND2(canvas_light_occluder_set_enabled, RID, bool) - BIND2(canvas_light_occluder_set_polygon, RID, RID) - BIND2(canvas_light_occluder_set_as_sdf_collision, RID, bool) - BIND2(canvas_light_occluder_set_transform, RID, const Transform2D &) - BIND2(canvas_light_occluder_set_light_mask, RID, int) - - BIND0R(RID, canvas_occluder_polygon_create) - BIND3(canvas_occluder_polygon_set_shape, RID, const Vector<Vector2> &, bool) - - BIND2(canvas_occluder_polygon_set_cull_mode, RID, CanvasOccluderPolygonCullMode) - - BIND1(canvas_set_shadow_texture_size, int) + FUNCRIDSPLIT(canvas) + FUNC3(canvas_set_item_mirroring, RID, RID, const Point2 &) + FUNC2(canvas_set_modulate, RID, const Color &) + FUNC3(canvas_set_parent, RID, RID, float) + FUNC1(canvas_set_disable_scale, bool) + + FUNCRIDSPLIT(canvas_texture) + FUNC3(canvas_texture_set_channel, RID, CanvasTextureChannel, RID) + FUNC3(canvas_texture_set_shading_parameters, RID, const Color &, float) + + FUNC2(canvas_texture_set_texture_filter, RID, CanvasItemTextureFilter) + FUNC2(canvas_texture_set_texture_repeat, RID, CanvasItemTextureRepeat) + + FUNCRIDSPLIT(canvas_item) + FUNC2(canvas_item_set_parent, RID, RID) + + FUNC2(canvas_item_set_default_texture_filter, RID, CanvasItemTextureFilter) + FUNC2(canvas_item_set_default_texture_repeat, RID, CanvasItemTextureRepeat) + + FUNC2(canvas_item_set_visible, RID, bool) + FUNC2(canvas_item_set_light_mask, RID, int) + + FUNC2(canvas_item_set_update_when_visible, RID, bool) + + FUNC2(canvas_item_set_transform, RID, const Transform2D &) + FUNC2(canvas_item_set_clip, RID, bool) + FUNC2(canvas_item_set_distance_field_mode, RID, bool) + FUNC3(canvas_item_set_custom_rect, RID, bool, const Rect2 &) + FUNC2(canvas_item_set_modulate, RID, const Color &) + FUNC2(canvas_item_set_self_modulate, RID, const Color &) + + FUNC2(canvas_item_set_draw_behind_parent, RID, bool) + + FUNC5(canvas_item_add_line, RID, const Point2 &, const Point2 &, const Color &, float) + FUNC5(canvas_item_add_polyline, RID, const Vector<Point2> &, const Vector<Color> &, float, bool) + FUNC4(canvas_item_add_multiline, RID, const Vector<Point2> &, const Vector<Color> &, float) + FUNC3(canvas_item_add_rect, RID, const Rect2 &, const Color &) + FUNC4(canvas_item_add_circle, RID, const Point2 &, float, const Color &) + FUNC6(canvas_item_add_texture_rect, RID, const Rect2 &, RID, bool, const Color &, bool) + FUNC7(canvas_item_add_texture_rect_region, RID, const Rect2 &, RID, const Rect2 &, const Color &, bool, bool) + FUNC10(canvas_item_add_nine_patch, RID, const Rect2 &, const Rect2 &, RID, const Vector2 &, const Vector2 &, NinePatchAxisMode, NinePatchAxisMode, bool, const Color &) + FUNC6(canvas_item_add_primitive, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID, float) + FUNC5(canvas_item_add_polygon, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID) + FUNC9(canvas_item_add_triangle_array, RID, const Vector<int> &, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, const Vector<int> &, const Vector<float> &, RID, int) + FUNC5(canvas_item_add_mesh, RID, const RID &, const Transform2D &, const Color &, RID) + FUNC3(canvas_item_add_multimesh, RID, RID, RID) + FUNC3(canvas_item_add_particles, RID, RID, RID) + FUNC2(canvas_item_add_set_transform, RID, const Transform2D &) + FUNC2(canvas_item_add_clip_ignore, RID, bool) + FUNC2(canvas_item_set_sort_children_by_y, RID, bool) + FUNC2(canvas_item_set_z_index, RID, int) + FUNC2(canvas_item_set_z_as_relative_to_parent, RID, bool) + FUNC3(canvas_item_set_copy_to_backbuffer, RID, bool, const Rect2 &) + FUNC2(canvas_item_attach_skeleton, RID, RID) + + FUNC1(canvas_item_clear, RID) + FUNC2(canvas_item_set_draw_index, RID, int) + + FUNC2(canvas_item_set_material, RID, RID) + + FUNC2(canvas_item_set_use_parent_material, RID, bool) + + FUNC6(canvas_item_set_canvas_group_mode, RID, CanvasGroupMode, float, bool, float, bool) + + FUNCRIDSPLIT(canvas_light) + + FUNC2(canvas_light_set_mode, RID, CanvasLightMode) + + FUNC2(canvas_light_attach_to_canvas, RID, RID) + FUNC2(canvas_light_set_enabled, RID, bool) + FUNC2(canvas_light_set_texture_scale, RID, float) + FUNC2(canvas_light_set_transform, RID, const Transform2D &) + FUNC2(canvas_light_set_texture, RID, RID) + FUNC2(canvas_light_set_texture_offset, RID, const Vector2 &) + FUNC2(canvas_light_set_color, RID, const Color &) + FUNC2(canvas_light_set_height, RID, float) + FUNC2(canvas_light_set_energy, RID, float) + FUNC3(canvas_light_set_z_range, RID, int, int) + FUNC3(canvas_light_set_layer_range, RID, int, int) + FUNC2(canvas_light_set_item_cull_mask, RID, int) + FUNC2(canvas_light_set_item_shadow_cull_mask, RID, int) + FUNC2(canvas_light_set_directional_distance, RID, float) + + FUNC2(canvas_light_set_blend_mode, RID, CanvasLightBlendMode) + + FUNC2(canvas_light_set_shadow_enabled, RID, bool) + FUNC2(canvas_light_set_shadow_filter, RID, CanvasLightShadowFilter) + FUNC2(canvas_light_set_shadow_color, RID, const Color &) + FUNC2(canvas_light_set_shadow_smooth, RID, float) + + FUNCRIDSPLIT(canvas_light_occluder) + FUNC2(canvas_light_occluder_attach_to_canvas, RID, RID) + FUNC2(canvas_light_occluder_set_enabled, RID, bool) + FUNC2(canvas_light_occluder_set_polygon, RID, RID) + FUNC2(canvas_light_occluder_set_as_sdf_collision, RID, bool) + FUNC2(canvas_light_occluder_set_transform, RID, const Transform2D &) + FUNC2(canvas_light_occluder_set_light_mask, RID, int) + + FUNCRIDSPLIT(canvas_occluder_polygon) + FUNC3(canvas_occluder_polygon_set_shape, RID, const Vector<Vector2> &, bool) + + FUNC2(canvas_occluder_polygon_set_cull_mode, RID, CanvasOccluderPolygonCullMode) + + FUNC1(canvas_set_shadow_texture_size, int) /* GLOBAL VARIABLES */ -#undef BINDBASE +#undef server_name +#undef ServerName //from now on, calls forwarded to this singleton -#define BINDBASE RSG::storage +#define ServerName RendererStorage +#define server_name RSG::storage - BIND3(global_variable_add, const StringName &, GlobalVariableType, const Variant &) - BIND1(global_variable_remove, const StringName &) - BIND0RC(Vector<StringName>, global_variable_get_list) - BIND2(global_variable_set, const StringName &, const Variant &) - BIND2(global_variable_set_override, const StringName &, const Variant &) - BIND1RC(GlobalVariableType, global_variable_get_type, const StringName &) - BIND1RC(Variant, global_variable_get, const StringName &) + FUNC3(global_variable_add, const StringName &, GlobalVariableType, const Variant &) + FUNC1(global_variable_remove, const StringName &) + FUNC0RC(Vector<StringName>, global_variable_get_list) + FUNC2(global_variable_set, const StringName &, const Variant &) + FUNC2(global_variable_set_override, const StringName &, const Variant &) + FUNC1RC(GlobalVariableType, global_variable_get_type, const StringName &) + FUNC1RC(Variant, global_variable_get, const StringName &) - BIND1(global_variables_load_settings, bool) - BIND0(global_variables_clear) + FUNC1(global_variables_load_settings, bool) + FUNC0(global_variables_clear) + +#undef server_name +#undef ServerName +#undef WRITE_ACTION +#undef SYNC_DEBUG /* BLACK BARS */ - virtual void black_bars_set_margins(int p_left, int p_top, int p_right, int p_bottom); - virtual void black_bars_set_images(RID p_left, RID p_top, RID p_right, RID p_bottom); + virtual void black_bars_set_margins(int p_left, int p_top, int p_right, int p_bottom) override; + virtual void black_bars_set_images(RID p_left, RID p_top, RID p_right, RID p_bottom) override; /* FREE */ - virtual void free(RID p_rid); ///< free RIDs associated with the visual server + virtual void free(RID p_rid) override { + if (Thread::get_caller_id() == server_thread) { + command_queue.flush_if_pending(); + _free(p_rid); + } else { + command_queue.push(this, &RenderingServerDefault::_free, p_rid); + } + } /* EVENT QUEUING */ - virtual void request_frame_drawn_callback(Object *p_where, const StringName &p_method, const Variant &p_userdata); + virtual void request_frame_drawn_callback(Object *p_where, const StringName &p_method, const Variant &p_userdata) override; - virtual void draw(bool p_swap_buffers, double frame_step); - virtual void sync(); - virtual bool has_changed() const; - virtual void init(); - virtual void finish(); + virtual void draw(bool p_swap_buffers, double frame_step) override; + virtual void sync() override; + virtual bool has_changed() const override; + virtual void init() override; + virtual void finish() override; /* STATUS INFORMATION */ - virtual int get_render_info(RenderInfo p_info); - virtual String get_video_adapter_name() const; - virtual String get_video_adapter_vendor() const; + virtual uint64_t get_render_info(RenderInfo p_info) override; + virtual String get_video_adapter_name() const override; + virtual String get_video_adapter_vendor() const override; - virtual void set_frame_profiling_enabled(bool p_enable); - virtual Vector<FrameProfileArea> get_frame_profile(); - virtual uint64_t get_frame_profile_frame(); + virtual void set_frame_profiling_enabled(bool p_enable) override; + virtual Vector<FrameProfileArea> get_frame_profile() override; + virtual uint64_t get_frame_profile_frame() override; - virtual RID get_test_cube(); + virtual RID get_test_cube() override; /* TESTING */ - virtual float get_frame_setup_time_cpu() const; + virtual float get_frame_setup_time_cpu() const override; - virtual void set_boot_image(const Ref<Image> &p_image, const Color &p_color, bool p_scale, bool p_use_filter = true); - virtual void set_default_clear_color(const Color &p_color); + virtual void set_boot_image(const Ref<Image> &p_image, const Color &p_color, bool p_scale, bool p_use_filter = true) override; + virtual void set_default_clear_color(const Color &p_color) override; - virtual bool has_feature(Features p_feature) const; + virtual bool has_feature(Features p_feature) const override; - virtual bool has_os_feature(const String &p_feature) const; - virtual void set_debug_generate_wireframes(bool p_generate); + virtual bool has_os_feature(const String &p_feature) const override; + virtual void set_debug_generate_wireframes(bool p_generate) override; - virtual void call_set_use_vsync(bool p_enable); + virtual bool is_low_end() const override; - virtual bool is_low_end() const; + virtual void sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir) override; - virtual void sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir); + virtual void set_print_gpu_profile(bool p_enable) override; - RenderingServerDefault(); + RenderingServerDefault(bool p_create_thread = false); ~RenderingServerDefault(); - -#undef DISPLAY_CHANGED - -#undef BIND0R -#undef BIND1RC -#undef BIND2RC -#undef BIND3RC -#undef BIND4RC - -#undef BIND1 -#undef BIND2 -#undef BIND3 -#undef BIND4 -#undef BIND5 -#undef BIND6 -#undef BIND7 -#undef BIND8 -#undef BIND9 -#undef BIND10 }; #endif diff --git a/servers/rendering/rendering_server_wrap_mt.cpp b/servers/rendering/rendering_server_wrap_mt.cpp deleted file mode 100644 index 3572c4dc78..0000000000 --- a/servers/rendering/rendering_server_wrap_mt.cpp +++ /dev/null @@ -1,178 +0,0 @@ -/*************************************************************************/ -/* rendering_server_wrap_mt.cpp */ -/*************************************************************************/ -/* This file is part of: */ -/* GODOT ENGINE */ -/* https://godotengine.org */ -/*************************************************************************/ -/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ -/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ -/* */ -/* Permission is hereby granted, free of charge, to any person obtaining */ -/* a copy of this software and associated documentation files (the */ -/* "Software"), to deal in the Software without restriction, including */ -/* without limitation the rights to use, copy, modify, merge, publish, */ -/* distribute, sublicense, and/or sell copies of the Software, and to */ -/* permit persons to whom the Software is furnished to do so, subject to */ -/* the following conditions: */ -/* */ -/* The above copyright notice and this permission notice shall be */ -/* included in all copies or substantial portions of the Software. */ -/* */ -/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ -/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ -/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ -/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ -/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ -/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ -/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ -/*************************************************************************/ - -#include "rendering_server_wrap_mt.h" -#include "core/config/project_settings.h" -#include "core/os/os.h" -#include "servers/display_server.h" - -void RenderingServerWrapMT::thread_exit() { - exit = true; -} - -void RenderingServerWrapMT::thread_draw(bool p_swap_buffers, double frame_step) { - if (!atomic_decrement(&draw_pending)) { - rendering_server->draw(p_swap_buffers, frame_step); - } -} - -void RenderingServerWrapMT::thread_flush() { - atomic_decrement(&draw_pending); -} - -void RenderingServerWrapMT::_thread_callback(void *_instance) { - RenderingServerWrapMT *vsmt = reinterpret_cast<RenderingServerWrapMT *>(_instance); - - vsmt->thread_loop(); -} - -void RenderingServerWrapMT::thread_loop() { - server_thread = Thread::get_caller_id(); - - DisplayServer::get_singleton()->make_rendering_thread(); - - rendering_server->init(); - - exit = false; - draw_thread_up = true; - while (!exit) { - // flush commands one by one, until exit is requested - command_queue.wait_and_flush_one(); - } - - command_queue.flush_all(); // flush all - - rendering_server->finish(); -} - -/* EVENT QUEUING */ - -void RenderingServerWrapMT::sync() { - if (create_thread) { - atomic_increment(&draw_pending); - command_queue.push_and_sync(this, &RenderingServerWrapMT::thread_flush); - } else { - command_queue.flush_all(); //flush all pending from other threads - } -} - -void RenderingServerWrapMT::draw(bool p_swap_buffers, double frame_step) { - if (create_thread) { - atomic_increment(&draw_pending); - command_queue.push(this, &RenderingServerWrapMT::thread_draw, p_swap_buffers, frame_step); - } else { - rendering_server->draw(p_swap_buffers, frame_step); - } -} - -void RenderingServerWrapMT::init() { - if (create_thread) { - print_verbose("RenderingServerWrapMT: Creating render thread"); - DisplayServer::get_singleton()->release_rendering_thread(); - if (create_thread) { - thread = Thread::create(_thread_callback, this); - print_verbose("RenderingServerWrapMT: Starting render thread"); - } - while (!draw_thread_up) { - OS::get_singleton()->delay_usec(1000); - } - print_verbose("RenderingServerWrapMT: Finished render thread"); - } else { - rendering_server->init(); - } -} - -void RenderingServerWrapMT::finish() { - sky_free_cached_ids(); - shader_free_cached_ids(); - material_free_cached_ids(); - mesh_free_cached_ids(); - multimesh_free_cached_ids(); - immediate_free_cached_ids(); - skeleton_free_cached_ids(); - directional_light_free_cached_ids(); - omni_light_free_cached_ids(); - spot_light_free_cached_ids(); - reflection_probe_free_cached_ids(); - gi_probe_free_cached_ids(); - lightmap_free_cached_ids(); - particles_free_cached_ids(); - particles_collision_free_cached_ids(); - camera_free_cached_ids(); - viewport_free_cached_ids(); - environment_free_cached_ids(); - camera_effects_free_cached_ids(); - scenario_free_cached_ids(); - instance_free_cached_ids(); - canvas_free_cached_ids(); - canvas_item_free_cached_ids(); - canvas_light_occluder_free_cached_ids(); - canvas_occluder_polygon_free_cached_ids(); - - if (thread) { - command_queue.push(this, &RenderingServerWrapMT::thread_exit); - Thread::wait_to_finish(thread); - memdelete(thread); - - thread = nullptr; - } else { - rendering_server->finish(); - } -} - -void RenderingServerWrapMT::set_use_vsync_callback(bool p_enable) { - singleton_mt->call_set_use_vsync(p_enable); -} - -RenderingServerWrapMT *RenderingServerWrapMT::singleton_mt = nullptr; - -RenderingServerWrapMT::RenderingServerWrapMT(RenderingServer *p_contained, bool p_create_thread) : - command_queue(p_create_thread) { - singleton_mt = this; - DisplayServer::switch_vsync_function = set_use_vsync_callback; //as this goes to another thread, make sure it goes properly - - rendering_server = p_contained; - create_thread = p_create_thread; - thread = nullptr; - draw_pending = 0; - draw_thread_up = false; - pool_max_size = GLOBAL_GET("memory/limits/multithreaded_server/rid_pool_prealloc"); - - if (!p_create_thread) { - server_thread = Thread::get_caller_id(); - } else { - server_thread = 0; - } -} - -RenderingServerWrapMT::~RenderingServerWrapMT() { - memdelete(rendering_server); - //finish(); -} diff --git a/servers/rendering/rendering_server_wrap_mt.h b/servers/rendering/rendering_server_wrap_mt.h deleted file mode 100644 index 29ee846f31..0000000000 --- a/servers/rendering/rendering_server_wrap_mt.h +++ /dev/null @@ -1,800 +0,0 @@ -/*************************************************************************/ -/* rendering_server_wrap_mt.h */ -/*************************************************************************/ -/* This file is part of: */ -/* GODOT ENGINE */ -/* https://godotengine.org */ -/*************************************************************************/ -/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ -/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ -/* */ -/* Permission is hereby granted, free of charge, to any person obtaining */ -/* a copy of this software and associated documentation files (the */ -/* "Software"), to deal in the Software without restriction, including */ -/* without limitation the rights to use, copy, modify, merge, publish, */ -/* distribute, sublicense, and/or sell copies of the Software, and to */ -/* permit persons to whom the Software is furnished to do so, subject to */ -/* the following conditions: */ -/* */ -/* The above copyright notice and this permission notice shall be */ -/* included in all copies or substantial portions of the Software. */ -/* */ -/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ -/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ -/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ -/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ -/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ -/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ -/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ -/*************************************************************************/ - -#ifndef RENDERING_SERVER_WRAP_MT_H -#define RENDERING_SERVER_WRAP_MT_H - -#include "core/os/thread.h" -#include "core/templates/command_queue_mt.h" -#include "servers/rendering_server.h" - -class RenderingServerWrapMT : public RenderingServer { - // the real visual server - mutable RenderingServer *rendering_server; - - mutable CommandQueueMT command_queue; - - static void _thread_callback(void *_instance); - void thread_loop(); - - Thread::ID server_thread; - volatile bool exit; - Thread *thread; - volatile bool draw_thread_up; - bool create_thread; - - uint64_t draw_pending; - void thread_draw(bool p_swap_buffers, double frame_step); - void thread_flush(); - - void thread_exit(); - - Mutex alloc_mutex; - - int pool_max_size; - - //#define DEBUG_SYNC - - static RenderingServerWrapMT *singleton_mt; - -#ifdef DEBUG_SYNC -#define SYNC_DEBUG print_line("sync on: " + String(__FUNCTION__)); -#else -#define SYNC_DEBUG -#endif - -public: -#define ServerName RenderingServer -#define ServerNameWrapMT RenderingServerWrapMT -#define server_name rendering_server -#include "servers/server_wrap_mt_common.h" - - //these go pass-through, as they can be called from any thread - virtual RID texture_2d_create(const Ref<Image> &p_image) { return rendering_server->texture_2d_create(p_image); } - virtual RID texture_2d_layered_create(const Vector<Ref<Image>> &p_layers, TextureLayeredType p_layered_type) { return rendering_server->texture_2d_layered_create(p_layers, p_layered_type); } - virtual RID texture_3d_create(Image::Format p_format, int p_width, int p_height, int p_depth, bool p_mipmaps, const Vector<Ref<Image>> &p_data) { return rendering_server->texture_3d_create(p_format, p_width, p_height, p_depth, p_mipmaps, p_data); } - virtual RID texture_proxy_create(RID p_base) { return rendering_server->texture_proxy_create(p_base); } - - //goes pass-through - virtual void texture_2d_update_immediate(RID p_texture, const Ref<Image> &p_image, int p_layer = 0) { rendering_server->texture_2d_update_immediate(p_texture, p_image, p_layer); } - //these go through command queue if they are in another thread - FUNC3(texture_2d_update, RID, const Ref<Image> &, int) - FUNC2(texture_3d_update, RID, const Vector<Ref<Image>> &) - FUNC2(texture_proxy_update, RID, RID) - - //these also go pass-through - virtual RID texture_2d_placeholder_create() { return rendering_server->texture_2d_placeholder_create(); } - virtual RID texture_2d_layered_placeholder_create(TextureLayeredType p_type) { return rendering_server->texture_2d_layered_placeholder_create(p_type); } - virtual RID texture_3d_placeholder_create() { return rendering_server->texture_3d_placeholder_create(); } - - FUNC1RC(Ref<Image>, texture_2d_get, RID) - FUNC2RC(Ref<Image>, texture_2d_layer_get, RID, int) - FUNC1RC(Vector<Ref<Image>>, texture_3d_get, RID) - - FUNC2(texture_replace, RID, RID) - - FUNC3(texture_set_size_override, RID, int, int) -// FIXME: Disabled during Vulkan refactoring, should be ported. -#if 0 - FUNC2(texture_bind, RID, uint32_t) -#endif - - FUNC3(texture_set_detect_3d_callback, RID, TextureDetectCallback, void *) - FUNC3(texture_set_detect_normal_callback, RID, TextureDetectCallback, void *) - FUNC3(texture_set_detect_roughness_callback, RID, TextureDetectRoughnessCallback, void *) - - FUNC2(texture_set_path, RID, const String &) - FUNC1RC(String, texture_get_path, RID) - FUNC1S(texture_debug_usage, List<TextureInfo> *) - - FUNC2(texture_set_force_redraw_if_visible, RID, bool) - - /* SHADER API */ - - FUNCRID(shader) - - FUNC2(shader_set_code, RID, const String &) - FUNC1RC(String, shader_get_code, RID) - - FUNC2SC(shader_get_param_list, RID, List<PropertyInfo> *) - - FUNC3(shader_set_default_texture_param, RID, const StringName &, RID) - FUNC2RC(RID, shader_get_default_texture_param, RID, const StringName &) - FUNC2RC(Variant, shader_get_param_default, RID, const StringName &) - - /* COMMON MATERIAL API */ - - FUNCRID(material) - - FUNC2(material_set_shader, RID, RID) - - FUNC3(material_set_param, RID, const StringName &, const Variant &) - FUNC2RC(Variant, material_get_param, RID, const StringName &) - - FUNC2(material_set_render_priority, RID, int) - FUNC2(material_set_next_pass, RID, RID) - - /* MESH API */ - - virtual RID mesh_create_from_surfaces(const Vector<SurfaceData> &p_surfaces, int p_blend_shape_count = 0) { - return rendering_server->mesh_create_from_surfaces(p_surfaces, p_blend_shape_count); - } - - FUNC2(mesh_set_blend_shape_count, RID, int) - - FUNCRID(mesh) - - FUNC2(mesh_add_surface, RID, const SurfaceData &) - - FUNC1RC(int, mesh_get_blend_shape_count, RID) - - FUNC2(mesh_set_blend_shape_mode, RID, BlendShapeMode) - FUNC1RC(BlendShapeMode, mesh_get_blend_shape_mode, RID) - - FUNC4(mesh_surface_update_region, RID, int, int, const Vector<uint8_t> &) - - FUNC3(mesh_surface_set_material, RID, int, RID) - FUNC2RC(RID, mesh_surface_get_material, RID, int) - - FUNC2RC(SurfaceData, mesh_get_surface, RID, int) - - FUNC1RC(int, mesh_get_surface_count, RID) - - FUNC2(mesh_set_custom_aabb, RID, const AABB &) - FUNC1RC(AABB, mesh_get_custom_aabb, RID) - - FUNC1(mesh_clear, RID) - - /* MULTIMESH API */ - - FUNCRID(multimesh) - - FUNC5(multimesh_allocate, RID, int, MultimeshTransformFormat, bool, bool) - FUNC1RC(int, multimesh_get_instance_count, RID) - - FUNC2(multimesh_set_mesh, RID, RID) - FUNC3(multimesh_instance_set_transform, RID, int, const Transform &) - FUNC3(multimesh_instance_set_transform_2d, RID, int, const Transform2D &) - FUNC3(multimesh_instance_set_color, RID, int, const Color &) - FUNC3(multimesh_instance_set_custom_data, RID, int, const Color &) - - FUNC1RC(RID, multimesh_get_mesh, RID) - FUNC1RC(AABB, multimesh_get_aabb, RID) - - FUNC2RC(Transform, multimesh_instance_get_transform, RID, int) - FUNC2RC(Transform2D, multimesh_instance_get_transform_2d, RID, int) - FUNC2RC(Color, multimesh_instance_get_color, RID, int) - FUNC2RC(Color, multimesh_instance_get_custom_data, RID, int) - - FUNC2(multimesh_set_buffer, RID, const Vector<float> &) - FUNC1RC(Vector<float>, multimesh_get_buffer, RID) - - FUNC2(multimesh_set_visible_instances, RID, int) - FUNC1RC(int, multimesh_get_visible_instances, RID) - - /* IMMEDIATE API */ - - FUNCRID(immediate) - FUNC3(immediate_begin, RID, PrimitiveType, RID) - FUNC2(immediate_vertex, RID, const Vector3 &) - FUNC2(immediate_normal, RID, const Vector3 &) - FUNC2(immediate_tangent, RID, const Plane &) - FUNC2(immediate_color, RID, const Color &) - FUNC2(immediate_uv, RID, const Vector2 &) - FUNC2(immediate_uv2, RID, const Vector2 &) - FUNC1(immediate_end, RID) - FUNC1(immediate_clear, RID) - FUNC2(immediate_set_material, RID, RID) - FUNC1RC(RID, immediate_get_material, RID) - - /* SKELETON API */ - - FUNCRID(skeleton) - FUNC3(skeleton_allocate, RID, int, bool) - FUNC1RC(int, skeleton_get_bone_count, RID) - FUNC3(skeleton_bone_set_transform, RID, int, const Transform &) - FUNC2RC(Transform, skeleton_bone_get_transform, RID, int) - FUNC3(skeleton_bone_set_transform_2d, RID, int, const Transform2D &) - FUNC2RC(Transform2D, skeleton_bone_get_transform_2d, RID, int) - FUNC2(skeleton_set_base_transform_2d, RID, const Transform2D &) - - /* Light API */ - - FUNCRID(directional_light) - FUNCRID(omni_light) - FUNCRID(spot_light) - - FUNC2(light_set_color, RID, const Color &) - FUNC3(light_set_param, RID, LightParam, float) - FUNC2(light_set_shadow, RID, bool) - FUNC2(light_set_shadow_color, RID, const Color &) - FUNC2(light_set_projector, RID, RID) - FUNC2(light_set_negative, RID, bool) - FUNC2(light_set_cull_mask, RID, uint32_t) - FUNC2(light_set_reverse_cull_face_mode, RID, bool) - FUNC2(light_set_bake_mode, RID, LightBakeMode) - FUNC2(light_set_max_sdfgi_cascade, RID, uint32_t) - - FUNC2(light_omni_set_shadow_mode, RID, LightOmniShadowMode) - - FUNC2(light_directional_set_shadow_mode, RID, LightDirectionalShadowMode) - FUNC2(light_directional_set_blend_splits, RID, bool) - FUNC2(light_directional_set_sky_only, RID, bool) - FUNC2(light_directional_set_shadow_depth_range_mode, RID, LightDirectionalShadowDepthRangeMode) - - /* PROBE API */ - - FUNCRID(reflection_probe) - - FUNC2(reflection_probe_set_update_mode, RID, ReflectionProbeUpdateMode) - FUNC2(reflection_probe_set_intensity, RID, float) - FUNC2(reflection_probe_set_ambient_color, RID, const Color &) - FUNC2(reflection_probe_set_ambient_energy, RID, float) - FUNC2(reflection_probe_set_ambient_mode, RID, ReflectionProbeAmbientMode) - FUNC2(reflection_probe_set_max_distance, RID, float) - FUNC2(reflection_probe_set_extents, RID, const Vector3 &) - FUNC2(reflection_probe_set_origin_offset, RID, const Vector3 &) - FUNC2(reflection_probe_set_as_interior, RID, bool) - FUNC2(reflection_probe_set_enable_box_projection, RID, bool) - FUNC2(reflection_probe_set_enable_shadows, RID, bool) - FUNC2(reflection_probe_set_cull_mask, RID, uint32_t) - FUNC2(reflection_probe_set_resolution, RID, int) - FUNC2(reflection_probe_set_lod_threshold, RID, float) - - /* DECAL API */ - - FUNCRID(decal) - - FUNC2(decal_set_extents, RID, const Vector3 &) - FUNC3(decal_set_texture, RID, DecalTexture, RID) - FUNC2(decal_set_emission_energy, RID, float) - FUNC2(decal_set_albedo_mix, RID, float) - FUNC2(decal_set_modulate, RID, const Color &) - FUNC2(decal_set_cull_mask, RID, uint32_t) - FUNC4(decal_set_distance_fade, RID, bool, float, float) - FUNC3(decal_set_fade, RID, float, float) - FUNC2(decal_set_normal_fade, RID, float) - - /* BAKED LIGHT API */ - - FUNCRID(gi_probe) - - FUNC8(gi_probe_allocate, RID, const Transform &, const AABB &, const Vector3i &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<uint8_t> &, const Vector<int> &) - - FUNC1RC(AABB, gi_probe_get_bounds, RID) - FUNC1RC(Vector3i, gi_probe_get_octree_size, RID) - FUNC1RC(Vector<uint8_t>, gi_probe_get_octree_cells, RID) - FUNC1RC(Vector<uint8_t>, gi_probe_get_data_cells, RID) - FUNC1RC(Vector<uint8_t>, gi_probe_get_distance_field, RID) - FUNC1RC(Vector<int>, gi_probe_get_level_counts, RID) - FUNC1RC(Transform, gi_probe_get_to_cell_xform, RID) - - FUNC2(gi_probe_set_dynamic_range, RID, float) - FUNC1RC(float, gi_probe_get_dynamic_range, RID) - - FUNC2(gi_probe_set_propagation, RID, float) - FUNC1RC(float, gi_probe_get_propagation, RID) - - FUNC2(gi_probe_set_energy, RID, float) - FUNC1RC(float, gi_probe_get_energy, RID) - - FUNC2(gi_probe_set_ao, RID, float) - FUNC1RC(float, gi_probe_get_ao, RID) - - FUNC2(gi_probe_set_ao_size, RID, float) - FUNC1RC(float, gi_probe_get_ao_size, RID) - - FUNC2(gi_probe_set_bias, RID, float) - FUNC1RC(float, gi_probe_get_bias, RID) - - FUNC2(gi_probe_set_normal_bias, RID, float) - FUNC1RC(float, gi_probe_get_normal_bias, RID) - - FUNC2(gi_probe_set_interior, RID, bool) - FUNC1RC(bool, gi_probe_is_interior, RID) - - FUNC2(gi_probe_set_use_two_bounces, RID, bool) - FUNC1RC(bool, gi_probe_is_using_two_bounces, RID) - - FUNC2(gi_probe_set_anisotropy_strength, RID, float) - FUNC1RC(float, gi_probe_get_anisotropy_strength, RID) - - FUNC1(gi_probe_set_quality, GIProbeQuality) - - /* LIGHTMAP CAPTURE */ - - FUNCRID(lightmap) - FUNC3(lightmap_set_textures, RID, RID, bool) - FUNC2(lightmap_set_probe_bounds, RID, const AABB &) - FUNC2(lightmap_set_probe_interior, RID, bool) - FUNC5(lightmap_set_probe_capture_data, RID, const PackedVector3Array &, const PackedColorArray &, const PackedInt32Array &, const PackedInt32Array &) - FUNC1RC(PackedVector3Array, lightmap_get_probe_capture_points, RID) - FUNC1RC(PackedColorArray, lightmap_get_probe_capture_sh, RID) - FUNC1RC(PackedInt32Array, lightmap_get_probe_capture_tetrahedra, RID) - FUNC1RC(PackedInt32Array, lightmap_get_probe_capture_bsp_tree, RID) - - FUNC1(lightmap_set_probe_capture_update_speed, float) - - /* PARTICLES */ - - FUNCRID(particles) - - FUNC2(particles_set_emitting, RID, bool) - FUNC1R(bool, particles_get_emitting, RID) - FUNC2(particles_set_amount, RID, int) - FUNC2(particles_set_lifetime, RID, float) - FUNC2(particles_set_one_shot, RID, bool) - FUNC2(particles_set_pre_process_time, RID, float) - FUNC2(particles_set_explosiveness_ratio, RID, float) - FUNC2(particles_set_randomness_ratio, RID, float) - FUNC2(particles_set_custom_aabb, RID, const AABB &) - FUNC2(particles_set_speed_scale, RID, float) - FUNC2(particles_set_use_local_coordinates, RID, bool) - FUNC2(particles_set_process_material, RID, RID) - FUNC2(particles_set_fixed_fps, RID, int) - FUNC2(particles_set_fractional_delta, RID, bool) - FUNC2(particles_set_collision_base_size, RID, float) - - FUNC1R(bool, particles_is_inactive, RID) - FUNC1(particles_request_process, RID) - FUNC1(particles_restart, RID) - - FUNC6(particles_emit, RID, const Transform &, const Vector3 &, const Color &, const Color &, uint32_t) - - FUNC2(particles_set_draw_order, RID, RS::ParticlesDrawOrder) - - FUNC2(particles_set_draw_passes, RID, int) - FUNC3(particles_set_draw_pass_mesh, RID, int, RID) - FUNC2(particles_set_emission_transform, RID, const Transform &) - FUNC2(particles_set_subemitter, RID, RID) - - FUNC1R(AABB, particles_get_current_aabb, RID) - - /* PARTICLES COLLISION */ - - FUNCRID(particles_collision) - - FUNC2(particles_collision_set_collision_type, RID, ParticlesCollisionType) - FUNC2(particles_collision_set_cull_mask, RID, uint32_t) - FUNC2(particles_collision_set_sphere_radius, RID, float) - FUNC2(particles_collision_set_box_extents, RID, const Vector3 &) - FUNC2(particles_collision_set_attractor_strength, RID, float) - FUNC2(particles_collision_set_attractor_directionality, RID, float) - FUNC2(particles_collision_set_attractor_attenuation, RID, float) - FUNC2(particles_collision_set_field_texture, RID, RID) - FUNC1(particles_collision_height_field_update, RID) - FUNC2(particles_collision_set_height_field_resolution, RID, ParticlesCollisionHeightfieldResolution) - - /* CAMERA API */ - - FUNCRID(camera) - FUNC4(camera_set_perspective, RID, float, float, float) - FUNC4(camera_set_orthogonal, RID, float, float, float) - FUNC5(camera_set_frustum, RID, float, Vector2, float, float) - FUNC2(camera_set_transform, RID, const Transform &) - FUNC2(camera_set_cull_mask, RID, uint32_t) - FUNC2(camera_set_environment, RID, RID) - FUNC2(camera_set_camera_effects, RID, RID) - FUNC2(camera_set_use_vertical_aspect, RID, bool) - - /* VIEWPORT TARGET API */ - - FUNCRID(viewport) - - FUNC2(viewport_set_use_xr, RID, bool) - - FUNC3(viewport_set_size, RID, int, int) - - FUNC2(viewport_set_active, RID, bool) - FUNC2(viewport_set_parent_viewport, RID, RID) - - FUNC2(viewport_set_clear_mode, RID, ViewportClearMode) - - FUNC3(viewport_attach_to_screen, RID, const Rect2 &, DisplayServer::WindowID) - FUNC2(viewport_set_render_direct_to_screen, RID, bool) - - FUNC2(viewport_set_update_mode, RID, ViewportUpdateMode) - - FUNC1RC(RID, viewport_get_texture, RID) - - FUNC2(viewport_set_hide_scenario, RID, bool) - FUNC2(viewport_set_hide_canvas, RID, bool) - FUNC2(viewport_set_disable_environment, RID, bool) - - FUNC2(viewport_attach_camera, RID, RID) - FUNC2(viewport_set_scenario, RID, RID) - FUNC2(viewport_attach_canvas, RID, RID) - - FUNC2(viewport_remove_canvas, RID, RID) - FUNC3(viewport_set_canvas_transform, RID, RID, const Transform2D &) - FUNC2(viewport_set_transparent_background, RID, bool) - FUNC2(viewport_set_snap_2d_transforms_to_pixel, RID, bool) - FUNC2(viewport_set_snap_2d_vertices_to_pixel, RID, bool) - - FUNC2(viewport_set_default_canvas_item_texture_filter, RID, CanvasItemTextureFilter) - FUNC2(viewport_set_default_canvas_item_texture_repeat, RID, CanvasItemTextureRepeat) - - FUNC2(viewport_set_global_canvas_transform, RID, const Transform2D &) - FUNC4(viewport_set_canvas_stacking, RID, RID, int, int) - FUNC2(viewport_set_shadow_atlas_size, RID, int) - FUNC3(viewport_set_sdf_oversize_and_scale, RID, ViewportSDFOversize, ViewportSDFScale) - - FUNC3(viewport_set_shadow_atlas_quadrant_subdivision, RID, int, int) - FUNC2(viewport_set_msaa, RID, ViewportMSAA) - FUNC2(viewport_set_screen_space_aa, RID, ViewportScreenSpaceAA) - FUNC2(viewport_set_use_debanding, RID, bool) - - FUNC2(viewport_set_lod_threshold, RID, float) - - //this passes directly to avoid stalling, but it's pretty dangerous, so don't call after freeing a viewport - virtual int viewport_get_render_info(RID p_viewport, ViewportRenderInfo p_info) { - return rendering_server->viewport_get_render_info(p_viewport, p_info); - } - - FUNC2(viewport_set_debug_draw, RID, ViewportDebugDraw) - - FUNC2(viewport_set_measure_render_time, RID, bool) - virtual float viewport_get_measured_render_time_cpu(RID p_viewport) const { - return rendering_server->viewport_get_measured_render_time_cpu(p_viewport); - } - virtual float viewport_get_measured_render_time_gpu(RID p_viewport) const { - return rendering_server->viewport_get_measured_render_time_gpu(p_viewport); - } - - FUNC1(directional_shadow_atlas_set_size, int) - - /* SKY API */ - - FUNCRID(sky) - FUNC2(sky_set_radiance_size, RID, int) - FUNC2(sky_set_mode, RID, SkyMode) - FUNC2(sky_set_material, RID, RID) - FUNC4R(Ref<Image>, sky_bake_panorama, RID, float, bool, const Size2i &) - - /* ENVIRONMENT API */ - - FUNCRID(environment) - - FUNC2(environment_set_background, RID, EnvironmentBG) - FUNC2(environment_set_sky, RID, RID) - FUNC2(environment_set_sky_custom_fov, RID, float) - FUNC2(environment_set_sky_orientation, RID, const Basis &) - FUNC2(environment_set_bg_color, RID, const Color &) - FUNC2(environment_set_bg_energy, RID, float) - FUNC2(environment_set_canvas_max_layer, RID, int) - FUNC7(environment_set_ambient_light, RID, const Color &, EnvironmentAmbientSource, float, float, EnvironmentReflectionSource, const Color &) - -// FIXME: Disabled during Vulkan refactoring, should be ported. -#if 0 - FUNC2(environment_set_camera_feed_id, RID, int) -#endif - FUNC6(environment_set_ssr, RID, bool, int, float, float, float) - FUNC1(environment_set_ssr_roughness_quality, EnvironmentSSRRoughnessQuality) - - FUNC10(environment_set_ssao, RID, bool, float, float, float, float, float, float, float, float) - - FUNC6(environment_set_ssao_quality, EnvironmentSSAOQuality, bool, float, int, float, float) - - FUNC11(environment_set_sdfgi, RID, bool, EnvironmentSDFGICascades, float, EnvironmentSDFGIYScale, bool, bool, bool, float, float, float) - FUNC1(environment_set_sdfgi_ray_count, EnvironmentSDFGIRayCount) - FUNC1(environment_set_sdfgi_frames_to_converge, EnvironmentSDFGIFramesToConverge) - - FUNC11(environment_set_glow, RID, bool, Vector<float>, float, float, float, float, EnvironmentGlowBlendMode, float, float, float) - FUNC1(environment_glow_set_use_bicubic_upscale, bool) - FUNC1(environment_glow_set_use_high_quality, bool) - - FUNC9(environment_set_tonemap, RID, EnvironmentToneMapper, float, float, bool, float, float, float, float) - - FUNC7(environment_set_adjustment, RID, bool, float, float, float, bool, RID) - - FUNC9(environment_set_fog, RID, bool, const Color &, float, float, float, float, float, float) - - FUNC9(environment_set_volumetric_fog, RID, bool, float, const Color &, float, float, float, float, EnvVolumetricFogShadowFilter) - - FUNC2(environment_set_volumetric_fog_volume_size, int, int) - FUNC1(environment_set_volumetric_fog_filter_active, bool) - FUNC1(environment_set_volumetric_fog_directional_shadow_shrink_size, int) - FUNC1(environment_set_volumetric_fog_positional_shadow_shrink_size, int) - - FUNC3R(Ref<Image>, environment_bake_panorama, RID, bool, const Size2i &) - - FUNC3(screen_space_roughness_limiter_set_active, bool, float, float) - FUNC1(sub_surface_scattering_set_quality, SubSurfaceScatteringQuality) - FUNC2(sub_surface_scattering_set_scale, float, float) - - FUNCRID(camera_effects) - - FUNC2(camera_effects_set_dof_blur_quality, DOFBlurQuality, bool) - FUNC1(camera_effects_set_dof_blur_bokeh_shape, DOFBokehShape) - - FUNC8(camera_effects_set_dof_blur, RID, bool, float, float, bool, float, float, float) - FUNC3(camera_effects_set_custom_exposure, RID, bool, float) - - FUNC1(shadows_quality_set, ShadowQuality); - FUNC1(directional_shadow_quality_set, ShadowQuality); - - FUNCRID(scenario) - - FUNC2(scenario_set_debug, RID, ScenarioDebugMode) - FUNC2(scenario_set_environment, RID, RID) - FUNC2(scenario_set_camera_effects, RID, RID) - FUNC2(scenario_set_fallback_environment, RID, RID) - - /* INSTANCING API */ - FUNCRID(instance) - - FUNC2(instance_set_base, RID, RID) - FUNC2(instance_set_scenario, RID, RID) - FUNC2(instance_set_layer_mask, RID, uint32_t) - FUNC2(instance_set_transform, RID, const Transform &) - FUNC2(instance_attach_object_instance_id, RID, ObjectID) - FUNC3(instance_set_blend_shape_weight, RID, int, float) - FUNC3(instance_set_surface_material, RID, int, RID) - FUNC2(instance_set_visible, RID, bool) - - FUNC2(instance_set_custom_aabb, RID, AABB) - - FUNC2(instance_attach_skeleton, RID, RID) - FUNC2(instance_set_exterior, RID, bool) - - FUNC2(instance_set_extra_visibility_margin, RID, real_t) - - // don't use these in a game! - FUNC2RC(Vector<ObjectID>, instances_cull_aabb, const AABB &, RID) - FUNC3RC(Vector<ObjectID>, instances_cull_ray, const Vector3 &, const Vector3 &, RID) - FUNC2RC(Vector<ObjectID>, instances_cull_convex, const Vector<Plane> &, RID) - - FUNC3(instance_geometry_set_flag, RID, InstanceFlags, bool) - FUNC2(instance_geometry_set_cast_shadows_setting, RID, ShadowCastingSetting) - FUNC2(instance_geometry_set_material_override, RID, RID) - - FUNC5(instance_geometry_set_draw_range, RID, float, float, float, float) - FUNC2(instance_geometry_set_as_instance_lod, RID, RID) - FUNC4(instance_geometry_set_lightmap, RID, RID, const Rect2 &, int) - FUNC2(instance_geometry_set_lod_bias, RID, float) - - FUNC3(instance_geometry_set_shader_parameter, RID, const StringName &, const Variant &) - FUNC2RC(Variant, instance_geometry_get_shader_parameter, RID, const StringName &) - FUNC2RC(Variant, instance_geometry_get_shader_parameter_default_value, RID, const StringName &) - FUNC2SC(instance_geometry_get_shader_parameter_list, RID, List<PropertyInfo> *) - - /* BAKE */ - - FUNC3R(TypedArray<Image>, bake_render_uv2, RID, const Vector<RID> &, const Size2i &) - - /* CANVAS (2D) */ - - FUNCRID(canvas) - FUNC3(canvas_set_item_mirroring, RID, RID, const Point2 &) - FUNC2(canvas_set_modulate, RID, const Color &) - FUNC3(canvas_set_parent, RID, RID, float) - FUNC1(canvas_set_disable_scale, bool) - - FUNCRID(canvas_texture) - FUNC3(canvas_texture_set_channel, RID, CanvasTextureChannel, RID) - FUNC3(canvas_texture_set_shading_parameters, RID, const Color &, float) - - FUNC2(canvas_texture_set_texture_filter, RID, CanvasItemTextureFilter) - FUNC2(canvas_texture_set_texture_repeat, RID, CanvasItemTextureRepeat) - - FUNCRID(canvas_item) - FUNC2(canvas_item_set_parent, RID, RID) - - FUNC2(canvas_item_set_default_texture_filter, RID, CanvasItemTextureFilter) - FUNC2(canvas_item_set_default_texture_repeat, RID, CanvasItemTextureRepeat) - - FUNC2(canvas_item_set_visible, RID, bool) - FUNC2(canvas_item_set_light_mask, RID, int) - - FUNC2(canvas_item_set_update_when_visible, RID, bool) - - FUNC2(canvas_item_set_transform, RID, const Transform2D &) - FUNC2(canvas_item_set_clip, RID, bool) - FUNC2(canvas_item_set_distance_field_mode, RID, bool) - FUNC3(canvas_item_set_custom_rect, RID, bool, const Rect2 &) - FUNC2(canvas_item_set_modulate, RID, const Color &) - FUNC2(canvas_item_set_self_modulate, RID, const Color &) - - FUNC2(canvas_item_set_draw_behind_parent, RID, bool) - - FUNC5(canvas_item_add_line, RID, const Point2 &, const Point2 &, const Color &, float) - FUNC5(canvas_item_add_polyline, RID, const Vector<Point2> &, const Vector<Color> &, float, bool) - FUNC4(canvas_item_add_multiline, RID, const Vector<Point2> &, const Vector<Color> &, float) - FUNC3(canvas_item_add_rect, RID, const Rect2 &, const Color &) - FUNC4(canvas_item_add_circle, RID, const Point2 &, float, const Color &) - FUNC6(canvas_item_add_texture_rect, RID, const Rect2 &, RID, bool, const Color &, bool) - FUNC7(canvas_item_add_texture_rect_region, RID, const Rect2 &, RID, const Rect2 &, const Color &, bool, bool) - FUNC10(canvas_item_add_nine_patch, RID, const Rect2 &, const Rect2 &, RID, const Vector2 &, const Vector2 &, NinePatchAxisMode, NinePatchAxisMode, bool, const Color &) - FUNC6(canvas_item_add_primitive, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID, float) - FUNC5(canvas_item_add_polygon, RID, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, RID) - FUNC9(canvas_item_add_triangle_array, RID, const Vector<int> &, const Vector<Point2> &, const Vector<Color> &, const Vector<Point2> &, const Vector<int> &, const Vector<float> &, RID, int) - FUNC5(canvas_item_add_mesh, RID, const RID &, const Transform2D &, const Color &, RID) - FUNC3(canvas_item_add_multimesh, RID, RID, RID) - FUNC3(canvas_item_add_particles, RID, RID, RID) - FUNC2(canvas_item_add_set_transform, RID, const Transform2D &) - FUNC2(canvas_item_add_clip_ignore, RID, bool) - FUNC2(canvas_item_set_sort_children_by_y, RID, bool) - FUNC2(canvas_item_set_z_index, RID, int) - FUNC2(canvas_item_set_z_as_relative_to_parent, RID, bool) - FUNC3(canvas_item_set_copy_to_backbuffer, RID, bool, const Rect2 &) - FUNC2(canvas_item_attach_skeleton, RID, RID) - - FUNC1(canvas_item_clear, RID) - FUNC2(canvas_item_set_draw_index, RID, int) - - FUNC2(canvas_item_set_material, RID, RID) - - FUNC2(canvas_item_set_use_parent_material, RID, bool) - - FUNC6(canvas_item_set_canvas_group_mode, RID, CanvasGroupMode, float, bool, float, bool) - - FUNC0R(RID, canvas_light_create) - - FUNC2(canvas_light_set_mode, RID, CanvasLightMode) - - FUNC2(canvas_light_attach_to_canvas, RID, RID) - FUNC2(canvas_light_set_enabled, RID, bool) - FUNC2(canvas_light_set_texture_scale, RID, float) - FUNC2(canvas_light_set_transform, RID, const Transform2D &) - FUNC2(canvas_light_set_texture, RID, RID) - FUNC2(canvas_light_set_texture_offset, RID, const Vector2 &) - FUNC2(canvas_light_set_color, RID, const Color &) - FUNC2(canvas_light_set_height, RID, float) - FUNC2(canvas_light_set_energy, RID, float) - FUNC3(canvas_light_set_z_range, RID, int, int) - FUNC3(canvas_light_set_layer_range, RID, int, int) - FUNC2(canvas_light_set_item_cull_mask, RID, int) - FUNC2(canvas_light_set_item_shadow_cull_mask, RID, int) - FUNC2(canvas_light_set_directional_distance, RID, float) - - FUNC2(canvas_light_set_blend_mode, RID, CanvasLightBlendMode) - - FUNC2(canvas_light_set_shadow_enabled, RID, bool) - FUNC2(canvas_light_set_shadow_filter, RID, CanvasLightShadowFilter) - FUNC2(canvas_light_set_shadow_color, RID, const Color &) - FUNC2(canvas_light_set_shadow_smooth, RID, float) - - FUNCRID(canvas_light_occluder) - FUNC2(canvas_light_occluder_attach_to_canvas, RID, RID) - FUNC2(canvas_light_occluder_set_enabled, RID, bool) - FUNC2(canvas_light_occluder_set_polygon, RID, RID) - FUNC2(canvas_light_occluder_set_as_sdf_collision, RID, bool) - FUNC2(canvas_light_occluder_set_transform, RID, const Transform2D &) - FUNC2(canvas_light_occluder_set_light_mask, RID, int) - - FUNCRID(canvas_occluder_polygon) - FUNC3(canvas_occluder_polygon_set_shape, RID, const Vector<Vector2> &, bool) - - FUNC2(canvas_occluder_polygon_set_cull_mode, RID, CanvasOccluderPolygonCullMode) - - FUNC1(canvas_set_shadow_texture_size, int) - - /* GLOBAL VARIABLES */ - - FUNC3(global_variable_add, const StringName &, GlobalVariableType, const Variant &) - FUNC1(global_variable_remove, const StringName &) - FUNC0RC(Vector<StringName>, global_variable_get_list) - FUNC2(global_variable_set, const StringName &, const Variant &) - FUNC2(global_variable_set_override, const StringName &, const Variant &) - FUNC1RC(GlobalVariableType, global_variable_get_type, const StringName &) - FUNC1RC(Variant, global_variable_get, const StringName &) - FUNC1(global_variables_load_settings, bool) - FUNC0(global_variables_clear) - - /* BLACK BARS */ - - FUNC4(black_bars_set_margins, int, int, int, int) - FUNC4(black_bars_set_images, RID, RID, RID, RID) - - /* FREE */ - - FUNC1(free, RID) - - /* EVENT QUEUING */ - - FUNC3(request_frame_drawn_callback, Object *, const StringName &, const Variant &) - - virtual void init(); - virtual void finish(); - virtual void draw(bool p_swap_buffers, double frame_step); - virtual void sync(); - FUNC0RC(bool, has_changed) - - /* RENDER INFO */ - - //this passes directly to avoid stalling - virtual int get_render_info(RenderInfo p_info) { - return rendering_server->get_render_info(p_info); - } - - virtual String get_video_adapter_name() const { - return rendering_server->get_video_adapter_name(); - } - - virtual String get_video_adapter_vendor() const { - return rendering_server->get_video_adapter_vendor(); - } - - FUNC4(set_boot_image, const Ref<Image> &, const Color &, bool, bool) - FUNC1(set_default_clear_color, const Color &) - - FUNC0R(RID, get_test_cube) - - FUNC1(set_debug_generate_wireframes, bool) - - virtual bool has_feature(Features p_feature) const { - return rendering_server->has_feature(p_feature); - } - virtual bool has_os_feature(const String &p_feature) const { - return rendering_server->has_os_feature(p_feature); - } - - FUNC1(call_set_use_vsync, bool) - - static void set_use_vsync_callback(bool p_enable); - - virtual bool is_low_end() const { - return rendering_server->is_low_end(); - } - - virtual uint64_t get_frame_profile_frame() { - return rendering_server->get_frame_profile_frame(); - } - - virtual void set_frame_profiling_enabled(bool p_enabled) { - rendering_server->set_frame_profiling_enabled(p_enabled); - } - - virtual Vector<FrameProfileArea> get_frame_profile() { - return rendering_server->get_frame_profile(); - } - - virtual float get_frame_setup_time_cpu() const { - return rendering_server->get_frame_setup_time_cpu(); - } - - virtual void sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir) { - rendering_server->sdfgi_set_debug_probe_select(p_position, p_dir); - } - - RenderingServerWrapMT(RenderingServer *p_contained, bool p_create_thread); - ~RenderingServerWrapMT(); - -#undef ServerName -#undef ServerNameWrapMT -#undef server_name -}; - -#ifdef DEBUG_SYNC -#undef DEBUG_SYNC -#endif -#undef SYNC_DEBUG - -#endif diff --git a/servers/rendering/shader_language.cpp b/servers/rendering/shader_language.cpp index 0cb9220bb3..4ae0eda232 100644 --- a/servers/rendering/shader_language.cpp +++ b/servers/rendering/shader_language.cpp @@ -558,13 +558,13 @@ ShaderLanguage::Token ShaderLanguage::_get_token() { return _make_token(TK_ERROR, "Invalid numeric constant"); } hexa_found = true; - } else if (GETCHAR(i) == 'e') { - if (hexa_found || exponent_found || float_suffix_found) { + } else if (GETCHAR(i) == 'e' && !hexa_found) { + if (exponent_found || float_suffix_found) { return _make_token(TK_ERROR, "Invalid numeric constant"); } exponent_found = true; - } else if (GETCHAR(i) == 'f') { - if (hexa_found || exponent_found) { + } else if (GETCHAR(i) == 'f' && !hexa_found) { + if (exponent_found) { return _make_token(TK_ERROR, "Invalid numeric constant"); } float_suffix_found = true; @@ -643,7 +643,7 @@ ShaderLanguage::Token ShaderLanguage::_get_token() { } if (hexa_found) { - tk.constant = (double)str.hex_to_int(true); + tk.constant = (double)str.hex_to_int(); } else { tk.constant = str.to_float(); } @@ -3102,6 +3102,72 @@ bool ShaderLanguage::_is_operator_assign(Operator p_op) const { return false; } +bool ShaderLanguage::_validate_varying_assign(ShaderNode::Varying &p_varying, String *r_message) { + if (current_function == String("light")) { + *r_message = RTR("Varying may not be assigned in the 'light' function."); + return false; + } + switch (p_varying.stage) { + case ShaderNode::Varying::STAGE_UNKNOWN: // first assign + if (current_function == String("vertex")) { + p_varying.stage = ShaderNode::Varying::STAGE_VERTEX; + } else if (current_function == String("fragment")) { + p_varying.stage = ShaderNode::Varying::STAGE_FRAGMENT; + } + break; + case ShaderNode::Varying::STAGE_VERTEX: + if (current_function == String("fragment")) { + *r_message = RTR("Varyings which assigned in 'vertex' function may not be reassigned in 'fragment' or 'light'."); + return false; + } + break; + case ShaderNode::Varying::STAGE_FRAGMENT: + if (current_function == String("vertex")) { + *r_message = RTR("Varyings which assigned in 'fragment' function may not be reassigned in 'vertex' or 'light'."); + return false; + } + break; + default: + break; + } + return true; +} + +bool ShaderLanguage::_validate_varying_using(ShaderNode::Varying &p_varying, String *r_message) { + switch (p_varying.stage) { + case ShaderNode::Varying::STAGE_UNKNOWN: + *r_message = RTR("Varying must be assigned before using!"); + return false; + case ShaderNode::Varying::STAGE_VERTEX: + if (current_function == String("fragment")) { + p_varying.stage = ShaderNode::Varying::STAGE_VERTEX_TO_FRAGMENT; + } else if (current_function == String("light")) { + p_varying.stage = ShaderNode::Varying::STAGE_VERTEX_TO_LIGHT; + } + break; + case ShaderNode::Varying::STAGE_FRAGMENT: + if (current_function == String("light")) { + p_varying.stage = ShaderNode::Varying::STAGE_FRAGMENT_TO_LIGHT; + } + break; + case ShaderNode::Varying::STAGE_VERTEX_TO_FRAGMENT: + if (current_function == String("light")) { + *r_message = RTR("Varying must only be used in two different stages, which can be 'vertex' 'fragment' and 'light'"); + return false; + } + break; + case ShaderNode::Varying::STAGE_VERTEX_TO_LIGHT: + if (current_function == String("fragment")) { + *r_message = RTR("Varying must only be used in two different stages, which can be 'vertex' 'fragment' and 'light'"); + return false; + } + break; + default: + break; + } + return true; +} + bool ShaderLanguage::_validate_assign(Node *p_node, const FunctionInfo &p_function_info, String *r_message) { if (p_node->type == Node::TYPE_OPERATOR) { OperatorNode *op = static_cast<OperatorNode *>(p_node); @@ -3142,13 +3208,6 @@ bool ShaderLanguage::_validate_assign(Node *p_node, const FunctionInfo &p_functi return false; } - if (shader->varyings.has(var->name) && current_function != String("vertex")) { - if (r_message) { - *r_message = RTR("Varyings can only be assigned in vertex function."); - } - return false; - } - if (shader->constants.has(var->name) || var->is_const) { if (r_message) { *r_message = RTR("Constants cannot be modified."); @@ -3169,13 +3228,6 @@ bool ShaderLanguage::_validate_assign(Node *p_node, const FunctionInfo &p_functi return false; } - if (shader->varyings.has(arr->name) && current_function != String("vertex")) { - if (r_message) { - *r_message = RTR("Varyings can only be assigned in vertex function."); - } - return false; - } - return true; } @@ -3761,6 +3813,23 @@ ShaderLanguage::Node *ShaderLanguage::_parse_expression(BlockNode *p_block, cons _set_error("Unknown identifier in expression: " + String(identifier)); return nullptr; } + if (ident_type == IDENTIFIER_VARYING) { + TkPos prev_pos = _get_tkpos(); + Token next_token = _get_token(); + _set_tkpos(prev_pos); + String error; + if (next_token.type == TK_OP_ASSIGN) { + if (!_validate_varying_assign(shader->varyings[identifier], &error)) { + _set_error(error); + return nullptr; + } + } else { + if (!_validate_varying_using(shader->varyings[identifier], &error)) { + _set_error(error); + return nullptr; + } + } + } last_const = is_const; if (ident_type == IDENTIFIER_FUNCTION) { @@ -3786,10 +3855,6 @@ ShaderLanguage::Node *ShaderLanguage::_parse_expression(BlockNode *p_block, cons _set_error("Constants cannot be modified."); return nullptr; } - if (shader->varyings.has(identifier) && current_function != String("vertex")) { - _set_error("Varyings can only be assigned in vertex function."); - return nullptr; - } assign_expression = _parse_array_constructor(p_block, p_function_info, data_type, struct_name, array_size); if (!assign_expression) { return nullptr; @@ -5796,6 +5861,8 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun return ERR_BUG; } + String return_struct_name = String(b->parent_function->return_struct_name); + ControlFlowNode *flow = alloc_node<ControlFlowNode>(); flow->flow_op = FLOW_OP_RETURN; @@ -5804,7 +5871,7 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun if (tk.type == TK_SEMICOLON) { //all is good if (b->parent_function->return_type != TYPE_VOID) { - _set_error("Expected return with expression of type '" + get_datatype_name(b->parent_function->return_type) + "'"); + _set_error("Expected return with an expression of type '" + (return_struct_name != "" ? return_struct_name : get_datatype_name(b->parent_function->return_type)) + "'"); return ERR_PARSE_ERROR; } } else { @@ -5814,8 +5881,8 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun return ERR_PARSE_ERROR; } - if (b->parent_function->return_type != expr->get_datatype()) { - _set_error("Expected return expression of type '" + get_datatype_name(b->parent_function->return_type) + "'"); + if (b->parent_function->return_type != expr->get_datatype() || return_struct_name != expr->get_datatype_name()) { + _set_error("Expected return with an expression of type '" + (return_struct_name != "" ? return_struct_name : get_datatype_name(b->parent_function->return_type)) + "'"); return ERR_PARSE_ERROR; } @@ -5859,15 +5926,15 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun pos = _get_tkpos(); tk = _get_token(); if (tk.type != TK_SEMICOLON) { - //all is good _set_error("Expected ';' after discard"); + return ERR_PARSE_ERROR; } p_block->statements.push_back(flow); } else if (tk.type == TK_CF_BREAK) { if (!p_can_break) { - //all is good - _set_error("Breaking is not allowed here"); + _set_error("'break' is not allowed outside of a loop or 'switch' statement"); + return ERR_PARSE_ERROR; } ControlFlowNode *flow = alloc_node<ControlFlowNode>(); @@ -5876,8 +5943,8 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun pos = _get_tkpos(); tk = _get_token(); if (tk.type != TK_SEMICOLON) { - //all is good _set_error("Expected ';' after break"); + return ERR_PARSE_ERROR; } p_block->statements.push_back(flow); @@ -5892,8 +5959,8 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun } else if (tk.type == TK_CF_CONTINUE) { if (!p_can_continue) { - //all is good - _set_error("Continuing is not allowed here"); + _set_error("'continue' is not allowed outside of a loop"); + return ERR_PARSE_ERROR; } ControlFlowNode *flow = alloc_node<ControlFlowNode>(); @@ -5904,6 +5971,7 @@ Error ShaderLanguage::_parse_block(BlockNode *p_block, const FunctionInfo &p_fun if (tk.type != TK_SEMICOLON) { //all is good _set_error("Expected ';' after continue"); + return ERR_PARSE_ERROR; } p_block->statements.push_back(flow); @@ -6068,6 +6136,10 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct tk = _get_token(); if (tk.type == TK_IDENTIFIER) { st.name = tk.text; + if (shader->structs.has(st.name)) { + _set_error("Redefinition of '" + String(st.name) + "'"); + return ERR_PARSE_ERROR; + } tk = _get_token(); if (tk.type != TK_CURLY_BRACKET_OPEN) { _set_error("Expected '{'"); @@ -6260,6 +6332,7 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct return ERR_PARSE_ERROR; } + TkPos name_pos = _get_tkpos(); name = tk.text; if (_find_identifier(nullptr, false, FunctionInfo(), name)) { @@ -6305,7 +6378,9 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct } uniform2.texture_order = -1; - uniform2.order = uniforms++; + if (uniform_scope != ShaderNode::Uniform::SCOPE_INSTANCE) { + uniform2.order = uniforms++; + } } uniform2.type = type; uniform2.scope = uniform_scope; @@ -6539,11 +6614,12 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct _set_error("Expected ';'"); return ERR_PARSE_ERROR; } - } else { + } else { // varying ShaderNode::Varying varying; varying.type = type; varying.precision = precision; varying.interpolation = interpolation; + varying.tkpos = name_pos; tk = _get_token(); if (tk.type != TK_SEMICOLON && tk.type != TK_BRACKET_OPEN) { @@ -6749,7 +6825,7 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct } else { _set_tkpos(pos2); - Node *n = _parse_and_reduce_expression(NULL, FunctionInfo()); + Node *n = _parse_and_reduce_expression(nullptr, FunctionInfo()); if (!n || n->type != Node::TYPE_CONSTANT || n->get_datatype() != TYPE_INT) { _set_error("Expected single integer constant > 0"); return ERR_PARSE_ERROR; @@ -6830,7 +6906,7 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct if (tk.type == TK_PARENTHESIS_OPEN || curly) { // initialization while (true) { - Node *n = _parse_and_reduce_expression(NULL, FunctionInfo()); + Node *n = _parse_and_reduce_expression(nullptr, FunctionInfo()); if (!n) { return ERR_PARSE_ERROR; } @@ -6856,10 +6932,11 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct decl.initializer.push_back(n); break; } else { - if (curly) + if (curly) { _set_error("Expected '}' or ','"); - else + } else { _set_error("Expected ')' or ','"); + } return ERR_PARSE_ERROR; } } @@ -6885,9 +6962,10 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct constant.initializer = static_cast<ConstantNode *>(expr); } else { //variable created with assignment! must parse an expression - Node *expr = _parse_and_reduce_expression(NULL, FunctionInfo()); - if (!expr) + Node *expr = _parse_and_reduce_expression(nullptr, FunctionInfo()); + if (!expr) { return ERR_PARSE_ERROR; + } if (expr->type == Node::TYPE_OPERATOR && ((OperatorNode *)expr)->op == OP_CALL) { _set_error("Expected constant expression after '='"); return ERR_PARSE_ERROR; @@ -7156,6 +7234,14 @@ Error ShaderLanguage::_parse_shader(const Map<StringName, FunctionInfo> &p_funct tk = _get_token(); } + for (Map<StringName, ShaderNode::Varying>::Element *E = shader->varyings.front(); E; E = E->next()) { + if (E->get().stage == ShaderNode::Varying::STAGE_VERTEX || E->get().stage == ShaderNode::Varying::STAGE_FRAGMENT) { + _set_tkpos(E->get().tkpos); + _set_error(RTR("Varying must only be used in two different stages, which can be 'vertex' 'fragment' and 'light'")); + return ERR_PARSE_ERROR; + } + } + return OK; } diff --git a/servers/rendering/shader_language.h b/servers/rendering/shader_language.h index 27767378f9..14594b039c 100644 --- a/servers/rendering/shader_language.h +++ b/servers/rendering/shader_language.h @@ -41,6 +41,11 @@ class ShaderLanguage { public: + struct TkPos { + int char_idx; + int tk_line; + }; + enum TokenType { TK_EMPTY, TK_IDENTIFIER, @@ -598,10 +603,21 @@ public: }; struct Varying { + enum Stage { + STAGE_UNKNOWN, + STAGE_VERTEX, // transition stage to STAGE_VERTEX_TO_FRAGMENT or STAGE_VERTEX_TO_LIGHT, emits error if they are not used + STAGE_FRAGMENT, // transition stage to STAGE_FRAGMENT_TO_LIGHT, emits error if it's not used + STAGE_VERTEX_TO_FRAGMENT, + STAGE_VERTEX_TO_LIGHT, + STAGE_FRAGMENT_TO_LIGHT, + }; + + Stage stage = STAGE_UNKNOWN; DataType type = TYPE_VOID; DataInterpolation interpolation = INTERPOLATION_FLAT; DataPrecision precision = PRECISION_DEFAULT; int array_size = 0; + TkPos tkpos; Varying() {} }; @@ -780,11 +796,6 @@ private: StringName current_function; bool last_const = false; - struct TkPos { - int char_idx; - int tk_line; - }; - TkPos _get_tkpos() { TkPos tkp; tkp.char_idx = char_idx; @@ -864,6 +875,8 @@ private: bool _parse_function_arguments(BlockNode *p_block, const FunctionInfo &p_function_info, OperatorNode *p_func, int *r_complete_arg = nullptr); bool _propagate_function_call_sampler_uniform_settings(StringName p_name, int p_argument, TextureFilter p_filter, TextureRepeat p_repeat); bool _propagate_function_call_sampler_builtin_reference(StringName p_name, int p_argument, const StringName &p_builtin); + bool _validate_varying_assign(ShaderNode::Varying &p_varying, String *r_message); + bool _validate_varying_using(ShaderNode::Varying &p_varying, String *r_message); Node *_parse_expression(BlockNode *p_block, const FunctionInfo &p_function_info); Node *_parse_array_constructor(BlockNode *p_block, const FunctionInfo &p_function_info, DataType p_type, const StringName &p_struct_name, int p_array_size); diff --git a/servers/rendering/shader_types.cpp b/servers/rendering/shader_types.cpp index c1fa4a8ca7..e99b8504bb 100644 --- a/servers/rendering/shader_types.cpp +++ b/servers/rendering/shader_types.cpp @@ -347,7 +347,7 @@ ShaderTypes::ShaderTypes() { emit_vertex_func.arguments.push_back(ShaderLanguage::StageFunctionInfo::Argument("custom", ShaderLanguage::TYPE_VEC4)); emit_vertex_func.arguments.push_back(ShaderLanguage::StageFunctionInfo::Argument("flags", ShaderLanguage::TYPE_UINT)); emit_vertex_func.return_type = ShaderLanguage::TYPE_BOOL; //whether it could emit - shader_modes[RS::SHADER_PARTICLES].functions["compute"].stage_functions["emit_particle"] = emit_vertex_func; + shader_modes[RS::SHADER_PARTICLES].functions["compute"].stage_functions["emit_subparticle"] = emit_vertex_func; } shader_modes[RS::SHADER_PARTICLES].modes.push_back("collision_use_scale"); diff --git a/servers/rendering_server.cpp b/servers/rendering_server.cpp index b87171dc5e..809343114c 100644 --- a/servers/rendering_server.cpp +++ b/servers/rendering_server.cpp @@ -242,22 +242,24 @@ RID RenderingServer::_make_test_cube() { RID RenderingServer::make_sphere_mesh(int p_lats, int p_lons, float p_radius) { Vector<Vector3> vertices; Vector<Vector3> normals; + const double lat_step = Math_TAU / p_lats; + const double lon_step = Math_TAU / p_lons; for (int i = 1; i <= p_lats; i++) { - double lat0 = Math_PI * (-0.5 + (double)(i - 1) / p_lats); + double lat0 = lat_step * (i - 1) - Math_TAU / 4; double z0 = Math::sin(lat0); double zr0 = Math::cos(lat0); - double lat1 = Math_PI * (-0.5 + (double)i / p_lats); + double lat1 = lat_step * i - Math_TAU / 4; double z1 = Math::sin(lat1); double zr1 = Math::cos(lat1); for (int j = p_lons; j >= 1; j--) { - double lng0 = 2 * Math_PI * (double)(j - 1) / p_lons; + double lng0 = lon_step * (j - 1); double x0 = Math::cos(lng0); double y0 = Math::sin(lng0); - double lng1 = 2 * Math_PI * (double)(j) / p_lons; + double lng1 = lon_step * j; double x1 = Math::cos(lng1); double y1 = Math::sin(lng1); @@ -1487,7 +1489,7 @@ void RenderingServer::_bind_methods() { ClassDB::bind_method(D_METHOD("mesh_clear", "mesh"), &RenderingServer::mesh_clear); ClassDB::bind_method(D_METHOD("multimesh_create"), &RenderingServer::multimesh_create); - ClassDB::bind_method(D_METHOD("multimesh_allocate", "multimesh", "instances", "transform_format", "color_format", "custom_data_format"), &RenderingServer::multimesh_allocate, DEFVAL(false), DEFVAL(false)); + ClassDB::bind_method(D_METHOD("multimesh_allocate_data", "multimesh", "instances", "transform_format", "color_format", "custom_data_format"), &RenderingServer::multimesh_allocate_data, DEFVAL(false), DEFVAL(false)); ClassDB::bind_method(D_METHOD("multimesh_get_instance_count", "multimesh"), &RenderingServer::multimesh_get_instance_count); ClassDB::bind_method(D_METHOD("multimesh_set_mesh", "multimesh", "mesh"), &RenderingServer::multimesh_set_mesh); ClassDB::bind_method(D_METHOD("multimesh_instance_set_transform", "multimesh", "index", "transform"), &RenderingServer::multimesh_instance_set_transform); @@ -1521,7 +1523,7 @@ void RenderingServer::_bind_methods() { #endif ClassDB::bind_method(D_METHOD("skeleton_create"), &RenderingServer::skeleton_create); - ClassDB::bind_method(D_METHOD("skeleton_allocate", "skeleton", "bones", "is_2d_skeleton"), &RenderingServer::skeleton_allocate, DEFVAL(false)); + ClassDB::bind_method(D_METHOD("skeleton_allocate_data", "skeleton", "bones", "is_2d_skeleton"), &RenderingServer::skeleton_allocate_data, DEFVAL(false)); ClassDB::bind_method(D_METHOD("skeleton_get_bone_count", "skeleton"), &RenderingServer::skeleton_get_bone_count); ClassDB::bind_method(D_METHOD("skeleton_bone_set_transform", "skeleton", "bone", "transform"), &RenderingServer::skeleton_bone_set_transform); ClassDB::bind_method(D_METHOD("skeleton_bone_get_transform", "skeleton", "bone"), &RenderingServer::skeleton_bone_get_transform); @@ -1661,7 +1663,7 @@ void RenderingServer::_bind_methods() { ClassDB::bind_method(D_METHOD("viewport_set_transparent_background", "viewport", "enabled"), &RenderingServer::viewport_set_transparent_background); ClassDB::bind_method(D_METHOD("viewport_set_global_canvas_transform", "viewport", "transform"), &RenderingServer::viewport_set_global_canvas_transform); ClassDB::bind_method(D_METHOD("viewport_set_canvas_stacking", "viewport", "canvas", "layer", "sublayer"), &RenderingServer::viewport_set_canvas_stacking); - ClassDB::bind_method(D_METHOD("viewport_set_shadow_atlas_size", "viewport", "size"), &RenderingServer::viewport_set_shadow_atlas_size); + ClassDB::bind_method(D_METHOD("viewport_set_shadow_atlas_size", "viewport", "size", "use_16_bits"), &RenderingServer::viewport_set_shadow_atlas_size, DEFVAL(false)); ClassDB::bind_method(D_METHOD("viewport_set_shadow_atlas_quadrant_subdivision", "viewport", "quadrant", "subdivision"), &RenderingServer::viewport_set_shadow_atlas_quadrant_subdivision); ClassDB::bind_method(D_METHOD("viewport_set_msaa", "viewport", "msaa"), &RenderingServer::viewport_set_msaa); ClassDB::bind_method(D_METHOD("viewport_set_use_debanding", "viewport", "enable"), &RenderingServer::viewport_set_use_debanding); @@ -2257,138 +2259,130 @@ RenderingServer::RenderingServer() { thread_pool = memnew(RendererThreadPool); singleton = this; - GLOBAL_DEF_RST("rendering/vram_compression/import_bptc", false); - GLOBAL_DEF_RST("rendering/vram_compression/import_s3tc", true); - GLOBAL_DEF_RST("rendering/vram_compression/import_etc", false); - GLOBAL_DEF_RST("rendering/vram_compression/import_etc2", true); - GLOBAL_DEF_RST("rendering/vram_compression/import_pvrtc", false); + GLOBAL_DEF_RST("rendering/textures/vram_compression/import_bptc", false); + GLOBAL_DEF_RST("rendering/textures/vram_compression/import_s3tc", true); + GLOBAL_DEF_RST("rendering/textures/vram_compression/import_etc", false); + GLOBAL_DEF_RST("rendering/textures/vram_compression/import_etc2", true); + GLOBAL_DEF_RST("rendering/textures/vram_compression/import_pvrtc", false); GLOBAL_DEF("rendering/limits/time/time_rollover_secs", 3600); ProjectSettings::get_singleton()->set_custom_property_info("rendering/limits/time/time_rollover_secs", PropertyInfo(Variant::FLOAT, "rendering/limits/time/time_rollover_secs", PROPERTY_HINT_RANGE, "0,10000,1,or_greater")); - GLOBAL_DEF("rendering/quality/directional_shadow/size", 4096); - GLOBAL_DEF("rendering/quality/directional_shadow/size.mobile", 2048); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/directional_shadow/size", PropertyInfo(Variant::INT, "rendering/quality/directional_shadow/size", PROPERTY_HINT_RANGE, "256,16384")); - GLOBAL_DEF("rendering/quality/directional_shadow/soft_shadow_quality", 2); - GLOBAL_DEF("rendering/quality/directional_shadow/soft_shadow_quality.mobile", 0); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/directional_shadow/soft_shadow_quality", PropertyInfo(Variant::INT, "rendering/quality/directional_shadow/soft_shadow_quality", PROPERTY_HINT_ENUM, "Hard (Fastest),Soft Low (Fast),Soft Medium (Average),Soft High (Slow),Soft Ultra (Slowest)")); - - GLOBAL_DEF("rendering/quality/shadows/soft_shadow_quality", 2); - GLOBAL_DEF("rendering/quality/shadows/soft_shadow_quality.mobile", 0); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadows/soft_shadow_quality", PropertyInfo(Variant::INT, "rendering/quality/shadows/soft_shadow_quality", PROPERTY_HINT_ENUM, "Hard (Fastest),Soft Low (Fast),Soft Medium (Average),Soft High (Slow),Soft Ultra (Slowest)")); - - GLOBAL_DEF("rendering/quality/2d_shadow_atlas/size", 2048); - - GLOBAL_DEF("rendering/quality/rd_renderer/use_low_end_renderer", false); - GLOBAL_DEF("rendering/quality/rd_renderer/use_low_end_renderer.mobile", true); - - GLOBAL_DEF("rendering/quality/shadow_atlas/size", 4096); - GLOBAL_DEF("rendering/quality/shadow_atlas/size.mobile", 2048); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadow_atlas/size", PropertyInfo(Variant::INT, "rendering/quality/shadow_atlas/size", PROPERTY_HINT_RANGE, "256,16384")); - GLOBAL_DEF("rendering/quality/shadow_atlas/quadrant_0_subdiv", 1); - GLOBAL_DEF("rendering/quality/shadow_atlas/quadrant_1_subdiv", 2); - GLOBAL_DEF("rendering/quality/shadow_atlas/quadrant_2_subdiv", 3); - GLOBAL_DEF("rendering/quality/shadow_atlas/quadrant_3_subdiv", 4); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadow_atlas/quadrant_0_subdiv", PropertyInfo(Variant::INT, "rendering/quality/shadow_atlas/quadrant_0_subdiv", PROPERTY_HINT_ENUM, "Disabled,1 Shadow,4 Shadows,16 Shadows,64 Shadows,256 Shadows,1024 Shadows")); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadow_atlas/quadrant_1_subdiv", PropertyInfo(Variant::INT, "rendering/quality/shadow_atlas/quadrant_1_subdiv", PROPERTY_HINT_ENUM, "Disabled,1 Shadow,4 Shadows,16 Shadows,64 Shadows,256 Shadows,1024 Shadows")); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadow_atlas/quadrant_2_subdiv", PropertyInfo(Variant::INT, "rendering/quality/shadow_atlas/quadrant_2_subdiv", PROPERTY_HINT_ENUM, "Disabled,1 Shadow,4 Shadows,16 Shadows,64 Shadows,256 Shadows,1024 Shadows")); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/shadow_atlas/quadrant_3_subdiv", PropertyInfo(Variant::INT, "rendering/quality/shadow_atlas/quadrant_3_subdiv", PROPERTY_HINT_ENUM, "Disabled,1 Shadow,4 Shadows,16 Shadows,64 Shadows,256 Shadows,1024 Shadows")); - - GLOBAL_DEF("rendering/quality/reflections/roughness_layers", 8); - GLOBAL_DEF("rendering/quality/reflections/texture_array_reflections", true); - GLOBAL_DEF("rendering/quality/reflections/texture_array_reflections.mobile", false); - GLOBAL_DEF("rendering/quality/reflections/ggx_samples", 1024); - GLOBAL_DEF("rendering/quality/reflections/ggx_samples.mobile", 128); - GLOBAL_DEF("rendering/quality/reflections/fast_filter_high_quality", false); - GLOBAL_DEF("rendering/quality/reflection_atlas/reflection_size", 256); - GLOBAL_DEF("rendering/quality/reflection_atlas/reflection_size.mobile", 128); - GLOBAL_DEF("rendering/quality/reflection_atlas/reflection_count", 64); - - GLOBAL_DEF("rendering/quality/gi_probes/anisotropic", false); - GLOBAL_DEF("rendering/quality/gi_probes/quality", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/gi_probes/quality", PropertyInfo(Variant::INT, "rendering/quality/gi_probes/quality", PROPERTY_HINT_ENUM, "Low (4 Cones - Fast),High (6 Cones - Slow)")); - - GLOBAL_DEF("rendering/quality/shading/force_vertex_shading", false); - GLOBAL_DEF("rendering/quality/shading/force_vertex_shading.mobile", true); - GLOBAL_DEF("rendering/quality/shading/force_lambert_over_burley", false); - GLOBAL_DEF("rendering/quality/shading/force_lambert_over_burley.mobile", true); - GLOBAL_DEF("rendering/quality/shading/force_blinn_over_ggx", false); - GLOBAL_DEF("rendering/quality/shading/force_blinn_over_ggx.mobile", true); - - GLOBAL_DEF("rendering/quality/depth_prepass/enable", true); - GLOBAL_DEF("rendering/quality/depth_prepass/disable_for_vendors", "PowerVR,Mali,Adreno,Apple"); - - GLOBAL_DEF("rendering/quality/texture_filters/use_nearest_mipmap_filter", false); - GLOBAL_DEF("rendering/quality/texture_filters/anisotropic_filtering_level", 2); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/texture_filters/anisotropic_filtering_level", PropertyInfo(Variant::INT, "rendering/quality/texture_filters/anisotropic_filtering_level", PROPERTY_HINT_ENUM, "Disabled (Fastest),2x (Faster),4x (Fast),8x (Average),16x (Slow)")); - - GLOBAL_DEF("rendering/quality/depth_of_field/depth_of_field_bokeh_shape", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/depth_of_field/depth_of_field_bokeh_shape", PropertyInfo(Variant::INT, "rendering/quality/depth_of_field/depth_of_field_bokeh_shape", PROPERTY_HINT_ENUM, "Box (Fast),Hexagon (Average),Circle (Slow)")); - GLOBAL_DEF("rendering/quality/depth_of_field/depth_of_field_bokeh_quality", 2); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/depth_of_field/depth_of_field_bokeh_quality", PropertyInfo(Variant::INT, "rendering/quality/depth_of_field/depth_of_field_bokeh_quality", PROPERTY_HINT_ENUM, "Very Low (Fastest),Low (Fast),Medium (Average),High (Slow)")); - GLOBAL_DEF("rendering/quality/depth_of_field/depth_of_field_use_jitter", false); - - GLOBAL_DEF("rendering/quality/ssao/quality", 2); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/ssao/quality", PropertyInfo(Variant::INT, "rendering/quality/ssao/quality", PROPERTY_HINT_ENUM, "Very Low (Fast),Low (Fast),Medium (Average),High (Slow),Ultra (Custom)")); - GLOBAL_DEF("rendering/quality/ssao/half_size", false); - GLOBAL_DEF("rendering/quality/ssao/half_size.mobile", true); - GLOBAL_DEF("rendering/quality/ssao/adaptive_target", 0.5); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/ssao/adaptive_target", PropertyInfo(Variant::FLOAT, "rendering/quality/ssao/adaptive_target", PROPERTY_HINT_RANGE, "0.0,1.0,0.01")); - GLOBAL_DEF("rendering/quality/ssao/blur_passes", 2); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/ssao/blur_passes", PropertyInfo(Variant::INT, "rendering/quality/ssao/blur_passes", PROPERTY_HINT_RANGE, "0,6")); - GLOBAL_DEF("rendering/quality/ssao/fadeout_from", 50.0); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/ssao/fadeout_from", PropertyInfo(Variant::FLOAT, "rendering/quality/ssao/fadeout_from", PROPERTY_HINT_RANGE, "0.0,512,0.1,or_greater")); - GLOBAL_DEF("rendering/quality/ssao/fadeout_to", 300.0); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/ssao/fadeout_to", PropertyInfo(Variant::FLOAT, "rendering/quality/ssao/fadeout_to", PROPERTY_HINT_RANGE, "64,65536,0.1,or_greater")); - - GLOBAL_DEF("rendering/quality/screen_filters/screen_space_roughness_limiter_enabled", true); - GLOBAL_DEF("rendering/quality/screen_filters/screen_space_roughness_limiter_amount", 0.25); - GLOBAL_DEF("rendering/quality/screen_filters/screen_space_roughness_limiter_limit", 0.18); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/screen_filters/screen_space_roughness_limiter_amount", PropertyInfo(Variant::FLOAT, "rendering/quality/screen_filters/screen_space_roughness_limiter_amount", PROPERTY_HINT_RANGE, "0.01,4.0,0.01")); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/screen_filters/screen_space_roughness_limiter_limit", PropertyInfo(Variant::FLOAT, "rendering/quality/screen_filters/screen_space_roughness_limiter_limit", PROPERTY_HINT_RANGE, "0.01,1.0,0.01")); - - GLOBAL_DEF("rendering/quality/glow/upscale_mode", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/glow/upscale_mode", PropertyInfo(Variant::INT, "rendering/quality/glow/upscale_mode", PROPERTY_HINT_ENUM, "Linear (Fast),Bicubic (Slow)")); - GLOBAL_DEF("rendering/quality/glow/upscale_mode.mobile", 0); - GLOBAL_DEF("rendering/quality/glow/use_high_quality", false); - - GLOBAL_DEF("rendering/quality/screen_space_reflection/roughness_quality", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/screen_space_reflection/roughness_quality", PropertyInfo(Variant::INT, "rendering/quality/screen_space_reflection/roughness_quality", PROPERTY_HINT_ENUM, "Disabled (Fastest),Low (Fast),Medium (Average),High (Slow)")); - - GLOBAL_DEF("rendering/quality/subsurface_scattering/subsurface_scattering_quality", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/subsurface_scattering/subsurface_scattering_quality", PropertyInfo(Variant::INT, "rendering/quality/subsurface_scattering/subsurface_scattering_quality", PROPERTY_HINT_ENUM, "Disabled (Fastest),Low (Fast),Medium (Average),High (Slow)")); - GLOBAL_DEF("rendering/quality/subsurface_scattering/subsurface_scattering_scale", 0.05); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/subsurface_scattering/subsurface_scattering_scale", PropertyInfo(Variant::FLOAT, "rendering/quality/subsurface_scattering/subsurface_scattering_scale", PROPERTY_HINT_RANGE, "0.001,1,0.001")); - GLOBAL_DEF("rendering/quality/subsurface_scattering/subsurface_scattering_depth_scale", 0.01); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/quality/subsurface_scattering/subsurface_scattering_depth_scale", PropertyInfo(Variant::FLOAT, "rendering/quality/subsurface_scattering/subsurface_scattering_depth_scale", PROPERTY_HINT_RANGE, "0.001,1,0.001")); - - GLOBAL_DEF("rendering/high_end/global_shader_variables_buffer_size", 65536); - - GLOBAL_DEF("rendering/lightmapper/probe_capture_update_speed", 15); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/lightmapper/probe_capture_update_speed", PropertyInfo(Variant::FLOAT, "rendering/lightmapper/probe_capture_update_speed", PROPERTY_HINT_RANGE, "0.001,256,0.001")); - - GLOBAL_DEF("rendering/sdfgi/probe_ray_count", 2); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/sdfgi/probe_ray_count", PropertyInfo(Variant::INT, "rendering/sdfgi/probe_ray_count", PROPERTY_HINT_ENUM, "8 (Fastest),16,32,64,96,128 (Slowest)")); - GLOBAL_DEF("rendering/sdfgi/frames_to_converge", 1); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/sdfgi/frames_to_converge", PropertyInfo(Variant::INT, "rendering/sdfgi/frames_to_converge", PROPERTY_HINT_ENUM, "5 (Less Latency but Lower Quality),10,15,20,25,30 (More Latency but Higher Quality)")); - - GLOBAL_DEF("rendering/volumetric_fog/volume_size", 64); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/volumetric_fog/volume_size", PropertyInfo(Variant::INT, "rendering/volumetric_fog/volume_size", PROPERTY_HINT_RANGE, "16,512,1")); - GLOBAL_DEF("rendering/volumetric_fog/volume_depth", 128); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/volumetric_fog/volume_depth", PropertyInfo(Variant::INT, "rendering/volumetric_fog/volume_depth", PROPERTY_HINT_RANGE, "16,512,1")); - GLOBAL_DEF("rendering/volumetric_fog/use_filter", 0); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/volumetric_fog/use_filter", PropertyInfo(Variant::INT, "rendering/volumetric_fog/use_filter", PROPERTY_HINT_ENUM, "No (Faster),Yes (Higher Quality)")); - GLOBAL_DEF("rendering/volumetric_fog/directional_shadow_shrink", 512); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/volumetric_fog/directional_shadow_shrink", PropertyInfo(Variant::INT, "rendering/volumetric_fog/directional_shadow_shrink", PROPERTY_HINT_RANGE, "32,2048,1")); - GLOBAL_DEF("rendering/volumetric_fog/positional_shadow_shrink", 512); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/volumetric_fog/positional_shadow_shrink", PropertyInfo(Variant::INT, "rendering/volumetric_fog/positional_shadow_shrink", PROPERTY_HINT_RANGE, "32,2048,1")); - - GLOBAL_DEF("rendering/spatial_indexer/update_iterations_per_frame", 10); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/spatial_indexer/update_iterations_per_frame", PropertyInfo(Variant::INT, "rendering/spatial_indexer/update_iterations_per_frame", PROPERTY_HINT_RANGE, "0,1024,1")); - GLOBAL_DEF("rendering/spatial_indexer/threaded_cull_minimum_instances", 1000); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/spatial_indexer/threaded_cull_minimum_instances", PropertyInfo(Variant::INT, "rendering/spatial_indexer/threaded_cull_minimum_instances", PROPERTY_HINT_RANGE, "32,65536,1")); - GLOBAL_DEF("rendering/forward_renderer/threaded_render_minimum_instances", 500); - ProjectSettings::get_singleton()->set_custom_property_info("rendering/forward_renderer/threaded_render_minimum_instances", PropertyInfo(Variant::INT, "rendering/forward_renderer/threaded_render_minimum_instances", PROPERTY_HINT_RANGE, "32,65536,1")); + GLOBAL_DEF("rendering/shadows/directional_shadow/size", 4096); + GLOBAL_DEF("rendering/shadows/directional_shadow/size.mobile", 2048); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/shadows/directional_shadow/size", PropertyInfo(Variant::INT, "rendering/shadows/directional_shadow/size", PROPERTY_HINT_RANGE, "256,16384")); + GLOBAL_DEF("rendering/shadows/directional_shadow/soft_shadow_quality", 2); + GLOBAL_DEF("rendering/shadows/directional_shadow/soft_shadow_quality.mobile", 0); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/shadows/directional_shadow/soft_shadow_quality", PropertyInfo(Variant::INT, "rendering/shadows/directional_shadow/soft_shadow_quality", PROPERTY_HINT_ENUM, "Hard (Fastest),Soft Low (Fast),Soft Medium (Average),Soft High (Slow),Soft Ultra (Slowest)")); + GLOBAL_DEF("rendering/shadows/directional_shadow/16_bits", true); + + GLOBAL_DEF("rendering/shadows/shadows/soft_shadow_quality", 2); + GLOBAL_DEF("rendering/shadows/shadows/soft_shadow_quality.mobile", 0); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/shadows/shadows/soft_shadow_quality", PropertyInfo(Variant::INT, "rendering/shadows/shadows/soft_shadow_quality", PROPERTY_HINT_ENUM, "Hard (Fastest),Soft Low (Fast),Soft Medium (Average),Soft High (Slow),Soft Ultra (Slowest)")); + + GLOBAL_DEF("rendering/2d/shadow_atlas/size", 2048); + + GLOBAL_DEF("rendering/driver/rd_renderer/use_low_end_renderer", false); + GLOBAL_DEF("rendering/driver/rd_renderer/use_low_end_renderer.mobile", true); + + GLOBAL_DEF("rendering/reflections/sky_reflections/roughness_layers", 8); + GLOBAL_DEF("rendering/reflections/sky_reflections/texture_array_reflections", true); + GLOBAL_DEF("rendering/reflections/sky_reflections/texture_array_reflections.mobile", false); + GLOBAL_DEF("rendering/reflections/sky_reflections/ggx_samples", 1024); + GLOBAL_DEF("rendering/reflections/sky_reflections/ggx_samples.mobile", 128); + GLOBAL_DEF("rendering/reflections/sky_reflections/fast_filter_high_quality", false); + GLOBAL_DEF("rendering/reflections/reflection_atlas/reflection_size", 256); + GLOBAL_DEF("rendering/reflections/reflection_atlas/reflection_size.mobile", 128); + GLOBAL_DEF("rendering/reflections/reflection_atlas/reflection_count", 64); + + GLOBAL_DEF("rendering/global_illumination/gi/use_half_resolution", false); + + GLOBAL_DEF("rendering/global_illumination/gi_probes/anisotropic", false); + GLOBAL_DEF("rendering/global_illumination/gi_probes/quality", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/global_illumination/gi_probes/quality", PropertyInfo(Variant::INT, "rendering/global_illumination/gi_probes/quality", PROPERTY_HINT_ENUM, "Low (4 Cones - Fast),High (6 Cones - Slow)")); + + GLOBAL_DEF("rendering/shading/overrides/force_vertex_shading", false); + GLOBAL_DEF("rendering/shading/overrides/force_vertex_shading.mobile", true); + GLOBAL_DEF("rendering/shading/overrides/force_lambert_over_burley", false); + GLOBAL_DEF("rendering/shading/overrides/force_lambert_over_burley.mobile", true); + GLOBAL_DEF("rendering/shading/overrides/force_blinn_over_ggx", false); + GLOBAL_DEF("rendering/shading/overrides/force_blinn_over_ggx.mobile", true); + + GLOBAL_DEF("rendering/driver/depth_prepass/enable", true); + GLOBAL_DEF("rendering/driver/depth_prepass/disable_for_vendors", "PowerVR,Mali,Adreno,Apple"); + + GLOBAL_DEF("rendering/textures/default_filters/use_nearest_mipmap_filter", false); + GLOBAL_DEF("rendering/textures/default_filters/anisotropic_filtering_level", 2); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/textures/default_filters/anisotropic_filtering_level", PropertyInfo(Variant::INT, "rendering/textures/default_filters/anisotropic_filtering_level", PROPERTY_HINT_ENUM, "Disabled (Fastest),2x (Faster),4x (Fast),8x (Average),16x (Slow)")); + + GLOBAL_DEF("rendering/camera/depth_of_field/depth_of_field_bokeh_shape", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/camera/depth_of_field/depth_of_field_bokeh_shape", PropertyInfo(Variant::INT, "rendering/camera/depth_of_field/depth_of_field_bokeh_shape", PROPERTY_HINT_ENUM, "Box (Fast),Hexagon (Average),Circle (Slow)")); + GLOBAL_DEF("rendering/camera/depth_of_field/depth_of_field_bokeh_quality", 2); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/camera/depth_of_field/depth_of_field_bokeh_quality", PropertyInfo(Variant::INT, "rendering/camera/depth_of_field/depth_of_field_bokeh_quality", PROPERTY_HINT_ENUM, "Very Low (Fastest),Low (Fast),Medium (Average),High (Slow)")); + GLOBAL_DEF("rendering/camera/depth_of_field/depth_of_field_use_jitter", false); + + GLOBAL_DEF("rendering/environment/ssao/quality", 2); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/ssao/quality", PropertyInfo(Variant::INT, "rendering/environment/ssao/quality", PROPERTY_HINT_ENUM, "Very Low (Fast),Low (Fast),Medium (Average),High (Slow),Ultra (Custom)")); + GLOBAL_DEF("rendering/environment/ssao/half_size", false); + GLOBAL_DEF("rendering/environment/ssao/half_size.mobile", true); + GLOBAL_DEF("rendering/environment/ssao/adaptive_target", 0.5); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/ssao/adaptive_target", PropertyInfo(Variant::FLOAT, "rendering/environment/ssao/adaptive_target", PROPERTY_HINT_RANGE, "0.0,1.0,0.01")); + GLOBAL_DEF("rendering/environment/ssao/blur_passes", 2); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/ssao/blur_passes", PropertyInfo(Variant::INT, "rendering/environment/ssao/blur_passes", PROPERTY_HINT_RANGE, "0,6")); + GLOBAL_DEF("rendering/environment/ssao/fadeout_from", 50.0); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/ssao/fadeout_from", PropertyInfo(Variant::FLOAT, "rendering/environment/ssao/fadeout_from", PROPERTY_HINT_RANGE, "0.0,512,0.1,or_greater")); + GLOBAL_DEF("rendering/environment/ssao/fadeout_to", 300.0); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/ssao/fadeout_to", PropertyInfo(Variant::FLOAT, "rendering/environment/ssao/fadeout_to", PROPERTY_HINT_RANGE, "64,65536,0.1,or_greater")); + + GLOBAL_DEF("rendering/anti_aliasing/screen_space_roughness_limiter/enabled", true); + GLOBAL_DEF("rendering/anti_aliasing/screen_space_roughness_limiter/amount", 0.25); + GLOBAL_DEF("rendering/anti_aliasing/screen_space_roughness_limiter/limit", 0.18); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/anti_aliasing/screen_space_roughness_limiter/amount", PropertyInfo(Variant::FLOAT, "rendering/anti_aliasing/screen_space_roughness_limiter/amount", PROPERTY_HINT_RANGE, "0.01,4.0,0.01")); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/anti_aliasing/screen_space_roughness_limiter/limit", PropertyInfo(Variant::FLOAT, "rendering/anti_aliasing/screen_space_roughness_limiter/limit", PROPERTY_HINT_RANGE, "0.01,1.0,0.01")); + + GLOBAL_DEF("rendering/environment/glow/upscale_mode", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/glow/upscale_mode", PropertyInfo(Variant::INT, "rendering/environment/glow/upscale_mode", PROPERTY_HINT_ENUM, "Linear (Fast),Bicubic (Slow)")); + GLOBAL_DEF("rendering/environment/glow/upscale_mode.mobile", 0); + GLOBAL_DEF("rendering/environment/glow/use_high_quality", false); + + GLOBAL_DEF("rendering/environment/screen_space_reflection/roughness_quality", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/screen_space_reflection/roughness_quality", PropertyInfo(Variant::INT, "rendering/environment/screen_space_reflection/roughness_quality", PROPERTY_HINT_ENUM, "Disabled (Fastest),Low (Fast),Medium (Average),High (Slow)")); + + GLOBAL_DEF("rendering/environment/subsurface_scattering/subsurface_scattering_quality", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/subsurface_scattering/subsurface_scattering_quality", PropertyInfo(Variant::INT, "rendering/environment/subsurface_scattering/subsurface_scattering_quality", PROPERTY_HINT_ENUM, "Disabled (Fastest),Low (Fast),Medium (Average),High (Slow)")); + GLOBAL_DEF("rendering/environment/subsurface_scattering/subsurface_scattering_scale", 0.05); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/subsurface_scattering/subsurface_scattering_scale", PropertyInfo(Variant::FLOAT, "rendering/environment/subsurface_scattering/subsurface_scattering_scale", PROPERTY_HINT_RANGE, "0.001,1,0.001")); + GLOBAL_DEF("rendering/environment/subsurface_scattering/subsurface_scattering_depth_scale", 0.01); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/subsurface_scattering/subsurface_scattering_depth_scale", PropertyInfo(Variant::FLOAT, "rendering/environment/subsurface_scattering/subsurface_scattering_depth_scale", PROPERTY_HINT_RANGE, "0.001,1,0.001")); + + GLOBAL_DEF("rendering/limits/global_shader_variables/buffer_size", 65536); + + GLOBAL_DEF("rendering/lightmapping/probe_capture/update_speed", 15); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/lightmapping/probe_capture/update_speed", PropertyInfo(Variant::FLOAT, "rendering/lightmapping/probe_capture/update_speed", PROPERTY_HINT_RANGE, "0.001,256,0.001")); + + GLOBAL_DEF("rendering/global_illumination/sdfgi/probe_ray_count", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/global_illumination/sdfgi/probe_ray_count", PropertyInfo(Variant::INT, "rendering/global_illumination/sdfgi/probe_ray_count", PROPERTY_HINT_ENUM, "8 (Fastest),16,32,64,96,128 (Slowest)")); + GLOBAL_DEF("rendering/global_illumination/sdfgi/frames_to_converge", 4); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/global_illumination/sdfgi/frames_to_converge", PropertyInfo(Variant::INT, "rendering/global_illumination/sdfgi/frames_to_converge", PROPERTY_HINT_ENUM, "5 (Less Latency but Lower Quality),10,15,20,25,30 (More Latency but Higher Quality)")); + GLOBAL_DEF("rendering/global_illumination/sdfgi/frames_to_update_lights", 2); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/global_illumination/sdfgi/frames_to_update_lights", PropertyInfo(Variant::INT, "rendering/global_illumination/sdfgi/frames_to_update_lights", PROPERTY_HINT_ENUM, "1 (Slower),2,4,8,16 (Faster)")); + + GLOBAL_DEF("rendering/environment/volumetric_fog/volume_size", 64); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/volumetric_fog/volume_size", PropertyInfo(Variant::INT, "rendering/environment/volumetric_fog/volume_size", PROPERTY_HINT_RANGE, "16,512,1")); + GLOBAL_DEF("rendering/environment/volumetric_fog/volume_depth", 128); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/volumetric_fog/volume_depth", PropertyInfo(Variant::INT, "rendering/environment/volumetric_fog/volume_depth", PROPERTY_HINT_RANGE, "16,512,1")); + GLOBAL_DEF("rendering/environment/volumetric_fog/use_filter", 1); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/environment/volumetric_fog/use_filter", PropertyInfo(Variant::INT, "rendering/environment/volumetric_fog/use_filter", PROPERTY_HINT_ENUM, "No (Faster),Yes (Higher Quality)")); + + GLOBAL_DEF("rendering/limits/spatial_indexer/update_iterations_per_frame", 10); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/limits/spatial_indexer/update_iterations_per_frame", PropertyInfo(Variant::INT, "rendering/limits/spatial_indexer/update_iterations_per_frame", PROPERTY_HINT_RANGE, "0,1024,1")); + GLOBAL_DEF("rendering/limits/spatial_indexer/threaded_cull_minimum_instances", 1000); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/limits/spatial_indexer/threaded_cull_minimum_instances", PropertyInfo(Variant::INT, "rendering/limits/spatial_indexer/threaded_cull_minimum_instances", PROPERTY_HINT_RANGE, "32,65536,1")); + GLOBAL_DEF("rendering/limits/forward_renderer/threaded_render_minimum_instances", 500); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/limits/forward_renderer/threaded_render_minimum_instances", PropertyInfo(Variant::INT, "rendering/limits/forward_renderer/threaded_render_minimum_instances", PROPERTY_HINT_RANGE, "32,65536,1")); + + GLOBAL_DEF("rendering/limits/cluster_builder/max_clustered_elements", 512); + ProjectSettings::get_singleton()->set_custom_property_info("rendering/limits/cluster_builder/max_clustered_elements", PropertyInfo(Variant::FLOAT, "rendering/limits/cluster_builder/max_clustered_elements", PROPERTY_HINT_RANGE, "32,8192,1")); } RenderingServer::~RenderingServer() { diff --git a/servers/rendering_server.h b/servers/rendering_server.h index 5481079694..6a8bb83ec1 100644 --- a/servers/rendering_server.h +++ b/servers/rendering_server.h @@ -179,6 +179,19 @@ public: virtual void shader_set_default_texture_param(RID p_shader, const StringName &p_name, RID p_texture) = 0; virtual RID shader_get_default_texture_param(RID p_shader, const StringName &p_name) const = 0; + struct ShaderNativeSourceCode { + struct Version { + struct Stage { + String name; + String code; + }; + Vector<Stage> stages; + }; + Vector<Version> versions; + }; + + virtual ShaderNativeSourceCode shader_get_native_source_code(RID p_shader) const = 0; + /* COMMON MATERIAL API */ enum { @@ -341,6 +354,8 @@ public: virtual void mesh_set_custom_aabb(RID p_mesh, const AABB &p_aabb) = 0; virtual AABB mesh_get_custom_aabb(RID p_mesh) const = 0; + virtual void mesh_set_shadow_mesh(RID p_mesh, RID p_shadow_mesh) = 0; + virtual void mesh_clear(RID p_mesh) = 0; /* MULTIMESH API */ @@ -352,7 +367,7 @@ public: MULTIMESH_TRANSFORM_3D, }; - virtual void multimesh_allocate(RID p_multimesh, int p_instances, MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false) = 0; + virtual void multimesh_allocate_data(RID p_multimesh, int p_instances, MultimeshTransformFormat p_transform_format, bool p_use_colors = false, bool p_use_custom_data = false) = 0; virtual int multimesh_get_instance_count(RID p_multimesh) const = 0; virtual void multimesh_set_mesh(RID p_multimesh, RID p_mesh) = 0; @@ -394,7 +409,7 @@ public: /* SKELETON API */ virtual RID skeleton_create() = 0; - virtual void skeleton_allocate(RID p_skeleton, int p_bones, bool p_2d_skeleton = false) = 0; + virtual void skeleton_allocate_data(RID p_skeleton, int p_bones, bool p_2d_skeleton = false) = 0; virtual int skeleton_get_bone_count(RID p_skeleton) const = 0; virtual void skeleton_bone_set_transform(RID p_skeleton, int p_bone, const Transform &p_transform) = 0; virtual Transform skeleton_bone_get_transform(RID p_skeleton, int p_bone) const = 0; @@ -537,7 +552,7 @@ public: virtual RID gi_probe_create() = 0; - virtual void gi_probe_allocate(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) = 0; + virtual void gi_probe_allocate_data(RID p_gi_probe, const Transform &p_to_cell_xform, const AABB &p_aabb, const Vector3i &p_octree_size, const Vector<uint8_t> &p_octree_cells, const Vector<uint8_t> &p_data_cells, const Vector<uint8_t> &p_distance_field, const Vector<int> &p_level_counts) = 0; virtual AABB gi_probe_get_bounds(RID p_gi_probe) const = 0; virtual Vector3i gi_probe_get_octree_size(RID p_gi_probe) const = 0; @@ -785,7 +800,7 @@ public: virtual void viewport_set_sdf_oversize_and_scale(RID p_viewport, ViewportSDFOversize p_oversize, ViewportSDFScale p_scale) = 0; - virtual void viewport_set_shadow_atlas_size(RID p_viewport, int p_size) = 0; + virtual void viewport_set_shadow_atlas_size(RID p_viewport, int p_size, bool p_16_bits = false) = 0; virtual void viewport_set_shadow_atlas_quadrant_subdivision(RID p_viewport, int p_quadrant, int p_subdiv) = 0; enum ViewportMSAA { @@ -843,7 +858,10 @@ public: VIEWPORT_DEBUG_DRAW_SDFGI_PROBES, VIEWPORT_DEBUG_DRAW_GI_BUFFER, VIEWPORT_DEBUG_DRAW_DISABLE_LOD, - + VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS, + VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS, + VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS, + VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES, }; virtual void viewport_set_debug_draw(RID p_viewport, ViewportDebugDraw p_draw) = 0; @@ -852,7 +870,7 @@ public: virtual float viewport_get_measured_render_time_cpu(RID p_viewport) const = 0; virtual float viewport_get_measured_render_time_gpu(RID p_viewport) const = 0; - virtual void directional_shadow_atlas_set_size(int p_size) = 0; + virtual void directional_shadow_atlas_set_size(int p_size, bool p_16_bits = false) = 0; /* SKY API */ @@ -967,9 +985,10 @@ public: ENV_SDFGI_Y_SCALE_50_PERCENT }; - virtual void environment_set_sdfgi(RID p_env, bool p_enable, EnvironmentSDFGICascades p_cascades, float p_min_cell_size, EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, bool p_use_multibounce, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; + virtual void environment_set_sdfgi(RID p_env, bool p_enable, EnvironmentSDFGICascades p_cascades, float p_min_cell_size, EnvironmentSDFGIYScale p_y_scale, bool p_use_occlusion, float p_bounce_feedback, bool p_read_sky, float p_energy, float p_normal_bias, float p_probe_bias) = 0; enum EnvironmentSDFGIRayCount { + ENV_SDFGI_RAY_COUNT_4, ENV_SDFGI_RAY_COUNT_8, ENV_SDFGI_RAY_COUNT_16, ENV_SDFGI_RAY_COUNT_32, @@ -993,20 +1012,22 @@ public: virtual void environment_set_sdfgi_frames_to_converge(EnvironmentSDFGIFramesToConverge p_frames) = 0; - virtual void environment_set_fog(RID p_env, bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_aerial_perspective) = 0; - - enum EnvVolumetricFogShadowFilter { - ENV_VOLUMETRIC_FOG_SHADOW_FILTER_DISABLED, - ENV_VOLUMETRIC_FOG_SHADOW_FILTER_LOW, - ENV_VOLUMETRIC_FOG_SHADOW_FILTER_MEDIUM, - ENV_VOLUMETRIC_FOG_SHADOW_FILTER_HIGH, + enum EnvironmentSDFGIFramesToUpdateLight { + ENV_SDFGI_UPDATE_LIGHT_IN_1_FRAME, + ENV_SDFGI_UPDATE_LIGHT_IN_2_FRAMES, + ENV_SDFGI_UPDATE_LIGHT_IN_4_FRAMES, + ENV_SDFGI_UPDATE_LIGHT_IN_8_FRAMES, + ENV_SDFGI_UPDATE_LIGHT_IN_16_FRAMES, + ENV_SDFGI_UPDATE_LIGHT_MAX, }; - virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, EnvVolumetricFogShadowFilter p_shadow_filter) = 0; + virtual void environment_set_sdfgi_frames_to_update_light(EnvironmentSDFGIFramesToUpdateLight p_update) = 0; + + virtual void environment_set_fog(RID p_env, bool p_enable, const Color &p_light_color, float p_light_energy, float p_sun_scatter, float p_density, float p_height, float p_height_density, float p_aerial_perspective) = 0; + + virtual void environment_set_volumetric_fog(RID p_env, bool p_enable, float p_density, const Color &p_light, float p_light_energy, float p_length, float p_detail_spread, float p_gi_inject, bool p_temporal_reprojection, float p_temporal_reprojection_amount) = 0; virtual void environment_set_volumetric_fog_volume_size(int p_size, int p_depth) = 0; virtual void environment_set_volumetric_fog_filter_active(bool p_enable) = 0; - virtual void environment_set_volumetric_fog_directional_shadow_shrink_size(int p_shrink_size) = 0; - virtual void environment_set_volumetric_fog_positional_shadow_shrink_size(int p_shrink_size) = 0; virtual Ref<Image> environment_bake_panorama(RID p_env, bool p_bake_irradiance, const Size2i &p_size) = 0; @@ -1398,7 +1419,7 @@ public: INFO_VERTEX_MEM_USED, }; - virtual int get_render_info(RenderInfo p_info) = 0; + virtual uint64_t get_render_info(RenderInfo p_info) = 0; virtual String get_video_adapter_name() const = 0; virtual String get_video_adapter_vendor() const = 0; @@ -1414,6 +1435,8 @@ public: virtual float get_frame_setup_time_cpu() const = 0; + virtual void gi_set_use_half_resolution(bool p_enable) = 0; + /* TESTING */ virtual RID get_test_cube() = 0; @@ -1446,6 +1469,8 @@ public: virtual bool is_low_end() const = 0; + virtual void set_print_gpu_profile(bool p_enable) = 0; + RenderingDevice *create_local_rendering_device() const; bool is_render_loop_enabled() const; diff --git a/servers/server_wrap_mt_common.h b/servers/server_wrap_mt_common.h index 5e18dc1e6d..95d2e2254e 100644 --- a/servers/server_wrap_mt_common.h +++ b/servers/server_wrap_mt_common.h @@ -29,245 +29,89 @@ /*************************************************************************/ #define FUNC0R(m_r, m_type) \ - virtual m_r m_type() { \ + virtual m_r m_type() override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(); \ } \ } -#define FUNCRID(m_type) \ - List<RID> m_type##_id_pool; \ - int m_type##allocn() { \ - for (int i = 0; i < pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create()); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - server_name->free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create() { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(); \ - } \ - } - -#define FUNC1RID(m_type, m_arg1) \ - int m_type##allocn() { \ - for (int i = 0; i < m_type##_pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create()); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create(m_arg1 p1) { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, p1, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(p1); \ - } \ - } - -#define FUNC2RID(m_type, m_arg1, m_arg2) \ - int m_type##allocn() { \ - for (int i = 0; i < m_type##_pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create()); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create(m_arg1 p1, m_arg2 p2) { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, p1, p2, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(p1, p2); \ - } \ - } - -#define FUNC3RID(m_type, m_arg1, m_arg2, m_arg3) \ - int m_type##allocn() { \ - for (int i = 0; i < m_type##_pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create()); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create(m_arg1 p1, m_arg2 p2, m_arg3 p3) { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, p1, p2, p3, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(p1, p2, p3); \ - } \ - } - -#define FUNC4RID(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - int m_type##allocn() { \ - for (int i = 0; i < m_type##_pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create()); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, p1, p2, p3, p4, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(p1, p2, p3, p4); \ - } \ +#define FUNCRIDSPLIT(m_type) \ + virtual RID m_type##_create() override { \ + RID ret = server_name->m_type##_allocate(); \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type##_initialize, ret); \ + } else { \ + server_name->m_type##_initialize(ret); \ + } \ + return ret; \ } -#define FUNC5RID(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - List<RID> m_type##_id_pool; \ - int m_type##allocn(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) { \ - for (int i = 0; i < pool_max_size; i++) { \ - m_type##_id_pool.push_back(server_name->m_type##_create(p1, p2, p3, p4, p5)); \ - } \ - return 0; \ - } \ - void m_type##_free_cached_ids() { \ - while (m_type##_id_pool.size()) { \ - free(m_type##_id_pool.front()->get()); \ - m_type##_id_pool.pop_front(); \ - } \ - } \ - virtual RID m_type##_create(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) { \ - if (Thread::get_caller_id() != server_thread) { \ - RID rid; \ - MutexLock lock(alloc_mutex); \ - if (m_type##_id_pool.size() == 0) { \ - int ret; \ - command_queue.push_and_ret(this, &ServerNameWrapMT::m_type##allocn, p1, p2, p3, p4, p5, &ret); \ - SYNC_DEBUG \ - } \ - rid = m_type##_id_pool.front()->get(); \ - m_type##_id_pool.pop_front(); \ - return rid; \ - } else { \ - return server_name->m_type##_create(p1, p2, p3, p4, p5); \ - } \ +//RID now returns directly, ensure thread safety yourself +#define FUNCRID(m_type) \ + virtual RID m_type##_create() override { \ + return server_name->m_type##_create(); \ } #define FUNC0RC(m_r, m_type) \ - virtual m_r m_type() const { \ + virtual m_r m_type() const override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(); \ } \ } #define FUNC0(m_type) \ - virtual void m_type() { \ + virtual void m_type() override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(); \ } \ } #define FUNC0C(m_type) \ - virtual void m_type() const { \ + virtual void m_type() const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(); \ } \ } #define FUNC0S(m_type) \ - virtual void m_type() { \ + virtual void m_type() override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(); \ } \ } #define FUNC0SC(m_type) \ - virtual void m_type() const { \ + virtual void m_type() const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(); \ } \ } @@ -275,560 +119,646 @@ /////////////////////////////////////////////// #define FUNC1R(m_r, m_type, m_arg1) \ - virtual m_r m_type(m_arg1 p1) { \ + virtual m_r m_type(m_arg1 p1) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1); \ } \ } #define FUNC1RC(m_r, m_type, m_arg1) \ - virtual m_r m_type(m_arg1 p1) const { \ + virtual m_r m_type(m_arg1 p1) const override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1); \ } \ } #define FUNC1S(m_type, m_arg1) \ - virtual void m_type(m_arg1 p1) { \ + virtual void m_type(m_arg1 p1) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1); \ } \ } #define FUNC1SC(m_type, m_arg1) \ - virtual void m_type(m_arg1 p1) const { \ + virtual void m_type(m_arg1 p1) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1); \ } \ } #define FUNC1(m_type, m_arg1) \ - virtual void m_type(m_arg1 p1) { \ + virtual void m_type(m_arg1 p1) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1); \ } \ } #define FUNC1C(m_type, m_arg1) \ - virtual void m_type(m_arg1 p1) const { \ + virtual void m_type(m_arg1 p1) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1); \ } \ } #define FUNC2R(m_r, m_type, m_arg1, m_arg2) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2) { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2); \ } \ } #define FUNC2RC(m_r, m_type, m_arg1, m_arg2) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2) const { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2) const override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2); \ } \ } #define FUNC2S(m_type, m_arg1, m_arg2) \ - virtual void m_type(m_arg1 p1, m_arg2 p2) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2); \ } \ } #define FUNC2SC(m_type, m_arg1, m_arg2) \ - virtual void m_type(m_arg1 p1, m_arg2 p2) const { \ + virtual void m_type(m_arg1 p1, m_arg2 p2) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2); \ } \ } #define FUNC2(m_type, m_arg1, m_arg2) \ - virtual void m_type(m_arg1 p1, m_arg2 p2) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1, p2); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2); \ } \ } #define FUNC2C(m_type, m_arg1, m_arg2) \ - virtual void m_type(m_arg1 p1, m_arg2 p2) const { \ + virtual void m_type(m_arg1 p1, m_arg2 p2) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1, p2); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2); \ } \ } #define FUNC3R(m_r, m_type, m_arg1, m_arg2, m_arg3) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3); \ } \ } #define FUNC3RC(m_r, m_type, m_arg1, m_arg2, m_arg3) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3); \ } \ } #define FUNC3S(m_type, m_arg1, m_arg2, m_arg3) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3); \ } \ } #define FUNC3SC(m_type, m_arg1, m_arg2, m_arg3) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3); \ } \ } #define FUNC3(m_type, m_arg1, m_arg2, m_arg3) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1, p2, p3); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3); \ } \ } #define FUNC3C(m_type, m_arg1, m_arg2, m_arg3) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1, p2, p3); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3); \ } \ } #define FUNC4R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3, p4); \ } \ } #define FUNC4RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3, p4); \ } \ } #define FUNC4S(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3, p4); \ } \ } #define FUNC4SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const override { \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3, p4); \ } \ } #define FUNC4(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4); \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3, p4); \ } \ } -#define FUNC4C(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4); \ - } else { \ - server_name->m_type(p1, p2, p3, p4); \ - } \ +#define FUNC4C(m_type, m_arg1, m_arg2, m_arg3, m_arg4) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4); \ + } \ } #define FUNC5R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3, p4, p5); \ } \ } #define FUNC5RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const { \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const override { \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3, p4, p5); \ } \ } #define FUNC5S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) { \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) override { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ SYNC_DEBUG \ } else { \ + command_queue.flush_if_pending(); \ server_name->m_type(p1, p2, p3, p4, p5); \ } \ } -#define FUNC5SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5); \ - } \ +#define FUNC5SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5); \ + } \ } -#define FUNC5(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5); \ - } \ +#define FUNC5(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5); \ + } \ } -#define FUNC5C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5); \ - } \ +#define FUNC5C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5); \ + } \ } #define FUNC6R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) { \ + WRITE_ACTION \ if (Thread::get_caller_id() != server_thread) { \ m_r ret; \ command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, &ret); \ SYNC_DEBUG \ return ret; \ } else { \ + command_queue.flush_if_pending(); \ return server_name->m_type(p1, p2, p3, p4, p5, p6); \ } \ } -#define FUNC6RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const { \ - if (Thread::get_caller_id() != server_thread) { \ - m_r ret; \ - command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, &ret); \ - SYNC_DEBUG \ - return ret; \ - } else { \ - return server_name->m_type(p1, p2, p3, p4, p5, p6); \ - } \ +#define FUNC6RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + m_r ret; \ + command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, &ret); \ + SYNC_DEBUG \ + return ret; \ + } else { \ + command_queue.flush_if_pending(); \ + return server_name->m_type(p1, p2, p3, p4, p5, p6); \ + } \ + } + +#define FUNC6S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6); \ + } \ + } + +#define FUNC6SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6); \ + } \ } -#define FUNC6S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6); \ - } \ +#define FUNC6(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6); \ + } \ } -#define FUNC6SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6); \ - } \ +#define FUNC6C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6); \ + } \ } -#define FUNC6(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6); \ - } \ +#define FUNC7R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + m_r ret; \ + command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, &ret); \ + SYNC_DEBUG \ + return ret; \ + } else { \ + command_queue.flush_if_pending(); \ + return server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ } -#define FUNC6C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6); \ - } \ - } - -#define FUNC7R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) { \ - if (Thread::get_caller_id() != server_thread) { \ - m_r ret; \ - command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, &ret); \ - SYNC_DEBUG \ - return ret; \ - } else { \ - return server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC7RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const { \ - if (Thread::get_caller_id() != server_thread) { \ - m_r ret; \ - command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, &ret); \ - SYNC_DEBUG \ - return ret; \ - } else { \ - return server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC7S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC7SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC7(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC7C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ - } \ - } - -#define FUNC8R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) { \ - if (Thread::get_caller_id() != server_thread) { \ - m_r ret; \ - command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, &ret); \ - SYNC_DEBUG \ - return ret; \ - } else { \ - return server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC8RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const { \ - if (Thread::get_caller_id() != server_thread) { \ - m_r ret; \ - command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, &ret); \ - SYNC_DEBUG \ - return ret; \ - } else { \ - return server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC8S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC8SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ - SYNC_DEBUG \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC8(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC8C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ - } \ - } - -#define FUNC9(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9); \ - } \ - } - -#define FUNC10(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10); \ - } \ - } - -#define FUNC11(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11); \ - } \ - } - -#define FUNC12(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12); \ - } \ - } - -#define FUNC13(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13); \ - } \ - } - -#define FUNC14(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13, m_arg14) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13, m_arg14 p14) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14); \ - } \ - } - -#define FUNC15(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13, m_arg14, m_arg15) \ - virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13, m_arg14 p14, m_arg15 p15) { \ - if (Thread::get_caller_id() != server_thread) { \ - command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14, p15); \ - } else { \ - server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14, p15); \ - } \ +#define FUNC7RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + m_r ret; \ + command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, &ret); \ + SYNC_DEBUG \ + return ret; \ + } else { \ + command_queue.flush_if_pending(); \ + return server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ + } + +#define FUNC7S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ + } + +#define FUNC7SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ + } + +#define FUNC7(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ + } + +#define FUNC7C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7); \ + } \ + } + +#define FUNC8R(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + m_r ret; \ + command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, &ret); \ + SYNC_DEBUG \ + return ret; \ + } else { \ + command_queue.flush_if_pending(); \ + return server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC8RC(m_r, m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual m_r m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + m_r ret; \ + command_queue.push_and_ret(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, &ret); \ + SYNC_DEBUG \ + return ret; \ + } else { \ + command_queue.flush_if_pending(); \ + return server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC8S(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC8SC(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push_and_sync(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ + SYNC_DEBUG \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC8(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC8C(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8) const override { \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8); \ + } \ + } + +#define FUNC9(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9); \ + } \ + } + +#define FUNC10(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10); \ + } \ + } + +#define FUNC11(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11); \ + } \ + } + +#define FUNC12(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12); \ + } \ + } + +#define FUNC13(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13); \ + } \ + } + +#define FUNC14(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13, m_arg14) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13, m_arg14 p14) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14); \ + } \ + } + +#define FUNC15(m_type, m_arg1, m_arg2, m_arg3, m_arg4, m_arg5, m_arg6, m_arg7, m_arg8, m_arg9, m_arg10, m_arg11, m_arg12, m_arg13, m_arg14, m_arg15) \ + virtual void m_type(m_arg1 p1, m_arg2 p2, m_arg3 p3, m_arg4 p4, m_arg5 p5, m_arg6 p6, m_arg7 p7, m_arg8 p8, m_arg9 p9, m_arg10 p10, m_arg11 p11, m_arg12 p12, m_arg13 p13, m_arg14 p14, m_arg15 p15) override { \ + WRITE_ACTION \ + if (Thread::get_caller_id() != server_thread) { \ + command_queue.push(server_name, &ServerName::m_type, p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14, p15); \ + } else { \ + command_queue.flush_if_pending(); \ + server_name->m_type(p1, p2, p3, p4, p5, p6, p7, p8, p9, p10, p11, p12, p13, p14, p15); \ + } \ } diff --git a/servers/text_server.cpp b/servers/text_server.cpp index da68ceb128..8b03565291 100644 --- a/servers/text_server.cpp +++ b/servers/text_server.cpp @@ -219,6 +219,11 @@ void TextServer::_bind_methods() { ClassDB::bind_method(D_METHOD("create_font_system", "name", "base_size"), &TextServer::create_font_system, DEFVAL(16)); ClassDB::bind_method(D_METHOD("create_font_resource", "filename", "base_size"), &TextServer::create_font_resource, DEFVAL(16)); ClassDB::bind_method(D_METHOD("create_font_memory", "data", "type", "base_size"), &TextServer::_create_font_memory, DEFVAL(16)); + ClassDB::bind_method(D_METHOD("create_font_bitmap", "height", "ascent", "base_size"), &TextServer::create_font_bitmap); + + ClassDB::bind_method(D_METHOD("font_bitmap_add_texture", "font", "texture"), &TextServer::font_bitmap_add_texture); + ClassDB::bind_method(D_METHOD("font_bitmap_add_char", "font", "char", "texture_idx", "rect", "align", "advance"), &TextServer::font_bitmap_add_char); + ClassDB::bind_method(D_METHOD("font_bitmap_add_kerning_pair", "font", "A", "B", "kerning"), &TextServer::font_bitmap_add_kerning_pair); ClassDB::bind_method(D_METHOD("font_get_height", "font", "size"), &TextServer::font_get_height); ClassDB::bind_method(D_METHOD("font_get_ascent", "font", "size"), &TextServer::font_get_ascent); @@ -227,6 +232,12 @@ void TextServer::_bind_methods() { ClassDB::bind_method(D_METHOD("font_get_underline_position", "font", "size"), &TextServer::font_get_underline_position); ClassDB::bind_method(D_METHOD("font_get_underline_thickness", "font", "size"), &TextServer::font_get_underline_thickness); + ClassDB::bind_method(D_METHOD("font_get_spacing_space", "font"), &TextServer::font_get_spacing_space); + ClassDB::bind_method(D_METHOD("font_set_spacing_space", "font", "value"), &TextServer::font_set_spacing_space); + + ClassDB::bind_method(D_METHOD("font_get_spacing_glyph", "font"), &TextServer::font_get_spacing_glyph); + ClassDB::bind_method(D_METHOD("font_set_spacing_glyph", "font", "value"), &TextServer::font_set_spacing_glyph); + ClassDB::bind_method(D_METHOD("font_set_antialiased", "font", "antialiased"), &TextServer::font_set_antialiased); ClassDB::bind_method(D_METHOD("font_get_antialiased", "font"), &TextServer::font_get_antialiased); @@ -280,11 +291,13 @@ void TextServer::_bind_methods() { ClassDB::bind_method(D_METHOD("get_hex_code_box_size", "size", "index"), &TextServer::get_hex_code_box_size); ClassDB::bind_method(D_METHOD("draw_hex_code_box", "canvas", "size", "pos", "index", "color"), &TextServer::draw_hex_code_box); + ClassDB::bind_method(D_METHOD("font_get_glyph_contours", "font", "size", "index"), &TextServer::_font_get_glyph_contours); + /* Shaped text buffer interface */ ClassDB::bind_method(D_METHOD("create_shaped_text", "direction", "orientation"), &TextServer::create_shaped_text, DEFVAL(DIRECTION_AUTO), DEFVAL(ORIENTATION_HORIZONTAL)); - ClassDB::bind_method(D_METHOD("shaped_text_clear"), &TextServer::shaped_text_clear); + ClassDB::bind_method(D_METHOD("shaped_text_clear", "rid"), &TextServer::shaped_text_clear); ClassDB::bind_method(D_METHOD("shaped_text_set_direction", "shaped", "direction"), &TextServer::shaped_text_set_direction, DEFVAL(DIRECTION_AUTO)); ClassDB::bind_method(D_METHOD("shaped_text_get_direction", "shaped"), &TextServer::shaped_text_get_direction); @@ -392,6 +405,11 @@ void TextServer::_bind_methods() { BIND_ENUM_CONSTANT(FEATURE_FONT_SYSTEM); BIND_ENUM_CONSTANT(FEATURE_FONT_VARIABLE); BIND_ENUM_CONSTANT(FEATURE_USE_SUPPORT_DATA); + + /* FT Contour Point Types */ + BIND_ENUM_CONSTANT(CONTOUR_CURVE_TAG_ON); + BIND_ENUM_CONSTANT(CONTOUR_CURVE_TAG_OFF_CONIC); + BIND_ENUM_CONSTANT(CONTOUR_CURVE_TAG_OFF_CUBIC); } Vector3 TextServer::hex_code_box_font_size[2] = { Vector3(5, 5, 1), Vector3(10, 10, 2) }; @@ -1201,6 +1219,21 @@ RID TextServer::_create_font_memory(const PackedByteArray &p_data, const String return create_font_memory(p_data.ptr(), p_data.size(), p_type, p_base_size); } +Dictionary TextServer::_font_get_glyph_contours(RID p_font, int p_size, uint32_t p_index) const { + Vector<Vector3> points; + Vector<int32_t> contours; + bool orientation; + bool ok = font_get_glyph_contours(p_font, p_size, p_index, points, contours, orientation); + Dictionary out; + + if (ok) { + out["points"] = points; + out["contours"] = contours; + out["orientation"] = orientation; + } + return out; +} + void TextServer::_shaped_text_set_bidi_override(RID p_shaped, const Array &p_override) { Vector<Vector2i> overrides; for (int i = 0; i < p_override.size(); i++) { diff --git a/servers/text_server.h b/servers/text_server.h index 23367de4c8..7fcfb91151 100644 --- a/servers/text_server.h +++ b/servers/text_server.h @@ -99,6 +99,12 @@ public: FEATURE_USE_SUPPORT_DATA = 1 << 7 }; + enum ContourPointTag { + CONTOUR_CURVE_TAG_ON = 0x01, + CONTOUR_CURVE_TAG_OFF_CONIC = 0x00, + CONTOUR_CURVE_TAG_OFF_CUBIC = 0x02 + }; + struct Glyph { int start = -1; // Start offset in the source string. int end = -1; // End offset in the source string. @@ -192,18 +198,6 @@ public: Vector<TextServer::Glyph> glyphs_logical; }; - struct BitmapFontData { - int height = 0; - int ascent = 0; - int charcount = 0; - const int *char_rects = nullptr; - int kerning_count = 0; - const int *kernings = nullptr; - int w = 0; - int h = 0; - const unsigned char *img = nullptr; - }; - protected: static void _bind_methods(); @@ -236,11 +230,22 @@ public: virtual RID create_font_system(const String &p_name, int p_base_size = 16) = 0; virtual RID create_font_resource(const String &p_filename, int p_base_size = 16) = 0; virtual RID create_font_memory(const uint8_t *p_data, size_t p_size, const String &p_type, int p_base_size = 16) = 0; + virtual RID create_font_bitmap(float p_height, float p_ascent, int p_base_size = 16) = 0; + + virtual void font_bitmap_add_texture(RID p_font, const Ref<Texture> &p_texture) = 0; + virtual void font_bitmap_add_char(RID p_font, char32_t p_char, int p_texture_idx, const Rect2 &p_rect, const Size2 &p_align, float p_advance) = 0; + virtual void font_bitmap_add_kerning_pair(RID p_font, char32_t p_A, char32_t p_B, int p_kerning) = 0; virtual float font_get_height(RID p_font, int p_size) const = 0; virtual float font_get_ascent(RID p_font, int p_size) const = 0; virtual float font_get_descent(RID p_font, int p_size) const = 0; + virtual int font_get_spacing_space(RID p_font) const = 0; + virtual void font_set_spacing_space(RID p_font, int p_value) = 0; + + virtual int font_get_spacing_glyph(RID p_font) const = 0; + virtual void font_set_spacing_glyph(RID p_font, int p_value) = 0; + virtual float font_get_underline_position(RID p_font, int p_size) const = 0; virtual float font_get_underline_thickness(RID p_font, int p_size) const = 0; @@ -287,6 +292,8 @@ public: virtual Vector2 font_draw_glyph(RID p_font, RID p_canvas, int p_size, const Vector2 &p_pos, uint32_t p_index, const Color &p_color = Color(1, 1, 1)) const = 0; virtual Vector2 font_draw_glyph_outline(RID p_font, RID p_canvas, int p_size, int p_outline_size, const Vector2 &p_pos, uint32_t p_index, const Color &p_color = Color(1, 1, 1)) const = 0; + virtual bool font_get_glyph_contours(RID p_font, int p_size, uint32_t p_index, Vector<Vector3> &r_points, Vector<int32_t> &r_contours, bool &r_orientation) const = 0; + virtual float font_get_oversampling() const = 0; virtual void font_set_oversampling(float p_oversampling) = 0; @@ -373,6 +380,8 @@ public: /* GDScript wrappers */ RID _create_font_memory(const PackedByteArray &p_data, const String &p_type, int p_base_size = 16); + Dictionary _font_get_glyph_contours(RID p_font, int p_size, uint32_t p_index) const; + Array _shaped_text_get_glyphs(RID p_shaped) const; Dictionary _shaped_text_get_carets(RID p_shaped, int p_position) const; @@ -455,5 +464,6 @@ VARIANT_ENUM_CAST(TextServer::LineBreakFlag); VARIANT_ENUM_CAST(TextServer::GraphemeFlag); VARIANT_ENUM_CAST(TextServer::Hinting); VARIANT_ENUM_CAST(TextServer::Feature); +VARIANT_ENUM_CAST(TextServer::ContourPointTag); #endif // TEXT_SERVER_H diff --git a/servers/xr/xr_positional_tracker.h b/servers/xr/xr_positional_tracker.h index 420d818342..a5c6459471 100644 --- a/servers/xr/xr_positional_tracker.h +++ b/servers/xr/xr_positional_tracker.h @@ -43,8 +43,8 @@ This is where potentially additional AR/VR interfaces may be active as there are AR/VR SDKs that solely deal with positional tracking. */ -class XRPositionalTracker : public Object { - GDCLASS(XRPositionalTracker, Object); +class XRPositionalTracker : public Reference { + GDCLASS(XRPositionalTracker, Reference); _THREAD_SAFE_CLASS_ public: diff --git a/servers/xr_server.cpp b/servers/xr_server.cpp index 2acc2e398c..5678071857 100644 --- a/servers/xr_server.cpp +++ b/servers/xr_server.cpp @@ -48,12 +48,17 @@ void XRServer::_bind_methods() { ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "world_scale"), "set_world_scale", "get_world_scale"); + ClassDB::bind_method(D_METHOD("add_interface", "interface"), &XRServer::add_interface); + ClassDB::bind_method(D_METHOD("clear_primary_interface_if", "interface"), &XRServer::clear_primary_interface_if); ClassDB::bind_method(D_METHOD("get_interface_count"), &XRServer::get_interface_count); + ClassDB::bind_method(D_METHOD("remove_interface", "interface"), &XRServer::remove_interface); ClassDB::bind_method(D_METHOD("get_interface", "idx"), &XRServer::get_interface); ClassDB::bind_method(D_METHOD("get_interfaces"), &XRServer::get_interfaces); ClassDB::bind_method(D_METHOD("find_interface", "name"), &XRServer::find_interface); ClassDB::bind_method(D_METHOD("get_tracker_count"), &XRServer::get_tracker_count); ClassDB::bind_method(D_METHOD("get_tracker", "idx"), &XRServer::get_tracker); + ClassDB::bind_method(D_METHOD("add_tracker", "tracker"), &XRServer::add_tracker); + ClassDB::bind_method(D_METHOD("remove_tracker", "tracker"), &XRServer::remove_tracker); ClassDB::bind_method(D_METHOD("get_primary_interface"), &XRServer::get_primary_interface); ClassDB::bind_method(D_METHOD("set_primary_interface", "interface"), &XRServer::set_primary_interface); @@ -260,15 +265,15 @@ int XRServer::get_free_tracker_id_for_type(TrackerType p_tracker_type) { return tracker_id; }; -void XRServer::add_tracker(XRPositionalTracker *p_tracker) { - ERR_FAIL_NULL(p_tracker); +void XRServer::add_tracker(Ref<XRPositionalTracker> p_tracker) { + ERR_FAIL_COND(p_tracker.is_null()); trackers.push_back(p_tracker); emit_signal("tracker_added", p_tracker->get_tracker_name(), p_tracker->get_tracker_type(), p_tracker->get_tracker_id()); }; -void XRServer::remove_tracker(XRPositionalTracker *p_tracker) { - ERR_FAIL_NULL(p_tracker); +void XRServer::remove_tracker(Ref<XRPositionalTracker> p_tracker) { + ERR_FAIL_COND(p_tracker.is_null()); int idx = -1; for (int i = 0; i < trackers.size(); i++) { @@ -288,14 +293,14 @@ int XRServer::get_tracker_count() const { return trackers.size(); }; -XRPositionalTracker *XRServer::get_tracker(int p_index) const { - ERR_FAIL_INDEX_V(p_index, trackers.size(), nullptr); +Ref<XRPositionalTracker> XRServer::get_tracker(int p_index) const { + ERR_FAIL_INDEX_V(p_index, trackers.size(), Ref<XRPositionalTracker>()); return trackers[p_index]; }; -XRPositionalTracker *XRServer::find_by_type_and_id(TrackerType p_tracker_type, int p_tracker_id) const { - ERR_FAIL_COND_V(p_tracker_id == 0, nullptr); +Ref<XRPositionalTracker> XRServer::find_by_type_and_id(TrackerType p_tracker_type, int p_tracker_id) const { + ERR_FAIL_COND_V(p_tracker_id == 0, Ref<XRPositionalTracker>()); for (int i = 0; i < trackers.size(); i++) { if (trackers[i]->get_tracker_type() == p_tracker_type && trackers[i]->get_tracker_id() == p_tracker_id) { @@ -303,7 +308,7 @@ XRPositionalTracker *XRServer::find_by_type_and_id(TrackerType p_tracker_type, i }; }; - return nullptr; + return Ref<XRPositionalTracker>(); }; Ref<XRInterface> XRServer::get_primary_interface() const { @@ -311,6 +316,7 @@ Ref<XRInterface> XRServer::get_primary_interface() const { }; void XRServer::set_primary_interface(const Ref<XRInterface> &p_primary_interface) { + ERR_FAIL_COND(p_primary_interface.is_null()); primary_interface = p_primary_interface; print_verbose("XR: Primary interface set to: " + primary_interface->get_name()); diff --git a/servers/xr_server.h b/servers/xr_server.h index d3972be838..46243d7fd0 100644 --- a/servers/xr_server.h +++ b/servers/xr_server.h @@ -77,7 +77,7 @@ public: private: Vector<Ref<XRInterface>> interfaces; - Vector<XRPositionalTracker *> trackers; + Vector<Ref<XRPositionalTracker>> trackers; Ref<XRInterface> primary_interface; /* we'll identify one interface as primary, this will be used by our viewports */ @@ -167,11 +167,11 @@ public: */ bool is_tracker_id_in_use_for_type(TrackerType p_tracker_type, int p_tracker_id) const; int get_free_tracker_id_for_type(TrackerType p_tracker_type); - void add_tracker(XRPositionalTracker *p_tracker); - void remove_tracker(XRPositionalTracker *p_tracker); + void add_tracker(Ref<XRPositionalTracker> p_tracker); + void remove_tracker(Ref<XRPositionalTracker> p_tracker); int get_tracker_count() const; - XRPositionalTracker *get_tracker(int p_index) const; - XRPositionalTracker *find_by_type_and_id(TrackerType p_tracker_type, int p_tracker_id) const; + Ref<XRPositionalTracker> get_tracker(int p_index) const; + Ref<XRPositionalTracker> find_by_type_and_id(TrackerType p_tracker_type, int p_tracker_id) const; uint64_t get_last_process_usec(); uint64_t get_last_commit_usec(); |