diff options
Diffstat (limited to 'servers')
52 files changed, 1977 insertions, 509 deletions
diff --git a/servers/audio/audio_stream.cpp b/servers/audio/audio_stream.cpp index aec6932326..f3fa857682 100644 --- a/servers/audio/audio_stream.cpp +++ b/servers/audio/audio_stream.cpp @@ -48,27 +48,27 @@ void AudioStreamPlaybackResampled::_begin_resample() { void AudioStreamPlaybackResampled::mix(AudioFrame *p_buffer, float p_rate_scale, int p_frames) { float target_rate = AudioServer::get_singleton()->get_mix_rate(); - float global_rate_scale = AudioServer::get_singleton()->get_global_rate_scale(); + float playback_speed_scale = AudioServer::get_singleton()->get_playback_speed_scale(); - uint64_t mix_increment = uint64_t(((get_stream_sampling_rate() * p_rate_scale) / double(target_rate * global_rate_scale)) * double(FP_LEN)); + uint64_t mix_increment = uint64_t(((get_stream_sampling_rate() * p_rate_scale * playback_speed_scale) / double(target_rate)) * double(FP_LEN)); for (int i = 0; i < p_frames; i++) { uint32_t idx = CUBIC_INTERP_HISTORY + uint32_t(mix_offset >> FP_BITS); - // 4 point, 4th order optimal resampling algorithm from: http://yehar.com/blog/wp-content/uploads/2009/08/deip.pdf + //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. 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]; - 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; + float mu2 = mu * mu; + AudioFrame a0 = 3 * y1 - 3 * y2 + y3 - y0; + AudioFrame a1 = 2 * y0 - 5 * y1 + 4 * y2 - y3; + AudioFrame a2 = y2 - y0; + AudioFrame a3 = 2 * y1; + + p_buffer[i] = (a0 * mu * mu2 + a1 * mu2 + a2 * mu + a3) / 2; mix_offset += mix_increment; diff --git a/servers/audio_server.cpp b/servers/audio_server.cpp index 076cbb58af..4c54188cb2 100644 --- a/servers/audio_server.cpp +++ b/servers/audio_server.cpp @@ -913,14 +913,14 @@ bool AudioServer::is_bus_channel_active(int p_bus, int p_channel) const { return buses[p_bus]->channels[p_channel].active; } -void AudioServer::set_global_rate_scale(float p_scale) { +void AudioServer::set_playback_speed_scale(float p_scale) { ERR_FAIL_COND(p_scale <= 0); - global_rate_scale = p_scale; + playback_speed_scale = p_scale; } -float AudioServer::get_global_rate_scale() const { - return global_rate_scale; +float AudioServer::get_playback_speed_scale() const { + return playback_speed_scale; } void AudioServer::init_channels_and_buffers() { @@ -1277,8 +1277,8 @@ void AudioServer::_bind_methods() { ClassDB::bind_method(D_METHOD("get_bus_peak_volume_left_db", "bus_idx", "channel"), &AudioServer::get_bus_peak_volume_left_db); ClassDB::bind_method(D_METHOD("get_bus_peak_volume_right_db", "bus_idx", "channel"), &AudioServer::get_bus_peak_volume_right_db); - ClassDB::bind_method(D_METHOD("set_global_rate_scale", "scale"), &AudioServer::set_global_rate_scale); - ClassDB::bind_method(D_METHOD("get_global_rate_scale"), &AudioServer::get_global_rate_scale); + ClassDB::bind_method(D_METHOD("set_playback_speed_scale", "scale"), &AudioServer::set_playback_speed_scale); + ClassDB::bind_method(D_METHOD("get_playback_speed_scale"), &AudioServer::get_playback_speed_scale); ClassDB::bind_method(D_METHOD("lock"), &AudioServer::lock); ClassDB::bind_method(D_METHOD("unlock"), &AudioServer::unlock); @@ -1302,7 +1302,7 @@ void AudioServer::_bind_methods() { ADD_PROPERTY(PropertyInfo(Variant::INT, "bus_count"), "set_bus_count", "get_bus_count"); ADD_PROPERTY(PropertyInfo(Variant::STRING, "device"), "set_device", "get_device"); - ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "global_rate_scale"), "set_global_rate_scale", "get_global_rate_scale"); + ADD_PROPERTY(PropertyInfo(Variant::FLOAT, "playback_speed_scale"), "set_playback_speed_scale", "get_playback_speed_scale"); ADD_SIGNAL(MethodInfo("bus_layout_changed")); @@ -1322,7 +1322,7 @@ AudioServer::AudioServer() { #endif mix_time = 0; mix_size = 0; - global_rate_scale = 1; + playback_speed_scale = 1; } AudioServer::~AudioServer() { diff --git a/servers/audio_server.h b/servers/audio_server.h index a1a373e1ca..7974c4a2ad 100644 --- a/servers/audio_server.h +++ b/servers/audio_server.h @@ -177,7 +177,7 @@ private: int channel_count; int to_mix; - float global_rate_scale; + float playback_speed_scale; struct Bus { StringName name; @@ -316,8 +316,8 @@ public: bool is_bus_channel_active(int p_bus, int p_channel) const; - void set_global_rate_scale(float p_scale); - float get_global_rate_scale() const; + void set_playback_speed_scale(float p_scale); + float get_playback_speed_scale() const; virtual void init(); virtual void finish(); diff --git a/servers/physics_2d/body_2d_sw.h b/servers/physics_2d/body_2d_sw.h index b4a95651cb..74bef433dc 100644 --- a/servers/physics_2d/body_2d_sw.h +++ b/servers/physics_2d/body_2d_sw.h @@ -266,6 +266,10 @@ public: void integrate_forces(real_t p_step); void integrate_velocities(real_t p_step); + _FORCE_INLINE_ Vector2 get_velocity_in_local_point(const Vector2 &rel_pos) const { + return linear_velocity + Vector2(-angular_velocity * rel_pos.y, angular_velocity * rel_pos.x); + } + _FORCE_INLINE_ Vector2 get_motion() const { if (mode > PhysicsServer2D::BODY_MODE_KINEMATIC) { return new_transform.get_origin() - get_transform().get_origin(); @@ -352,6 +356,8 @@ public: virtual void set_transform(const Transform2D &p_transform) override { body->set_state(PhysicsServer2D::BODY_STATE_TRANSFORM, p_transform); } virtual Transform2D get_transform() const override { return body->get_transform(); } + virtual Vector2 get_velocity_at_local_position(const Vector2 &p_position) const override { return body->get_velocity_in_local_point(p_position); } + virtual void add_central_force(const Vector2 &p_force) override { body->add_central_force(p_force); } virtual void add_force(const Vector2 &p_force, const Vector2 &p_position = Vector2()) override { body->add_force(p_force, p_position); } virtual void add_torque(real_t p_torque) override { body->add_torque(p_torque); } diff --git a/servers/physics_3d/body_3d_sw.h b/servers/physics_3d/body_3d_sw.h index 0fa31c5037..bdc615ab6c 100644 --- a/servers/physics_3d/body_3d_sw.h +++ b/servers/physics_3d/body_3d_sw.h @@ -389,6 +389,8 @@ public: virtual void set_transform(const Transform3D &p_transform) override { body->set_state(PhysicsServer3D::BODY_STATE_TRANSFORM, p_transform); } virtual Transform3D get_transform() const override { return body->get_transform(); } + virtual Vector3 get_velocity_at_local_position(const Vector3 &p_position) const override { return body->get_velocity_in_local_point(p_position); } + virtual void add_central_force(const Vector3 &p_force) override { body->add_central_force(p_force); } virtual void add_force(const Vector3 &p_force, const Vector3 &p_position = Vector3()) override { body->add_force(p_force, p_position); diff --git a/servers/physics_3d/collision_solver_3d_sat.cpp b/servers/physics_3d/collision_solver_3d_sat.cpp index 6a7f2b73c5..852e9ecd74 100644 --- a/servers/physics_3d/collision_solver_3d_sat.cpp +++ b/servers/physics_3d/collision_solver_3d_sat.cpp @@ -956,9 +956,12 @@ static void _collision_sphere_convex_polygon(const Shape3DSW *p_a, const Transfo const Vector3 *vertices = mesh.vertices.ptr(); int vertex_count = mesh.vertices.size(); + // Precalculating this makes the transforms faster. + Basis b_xform_normal = p_transform_b.basis.inverse().transposed(); + // faces of B for (int i = 0; i < face_count; i++) { - Vector3 axis = p_transform_b.xform(faces[i].plane).normal; + Vector3 axis = b_xform_normal.xform(faces[i].plane.normal).normalized(); if (!separator.test_axis(axis)) { return; @@ -1379,9 +1382,12 @@ static void _collision_box_convex_polygon(const Shape3DSW *p_a, const Transform3 } } + // Precalculating this makes the transforms faster. + Basis b_xform_normal = p_transform_b.basis.inverse().transposed(); + // faces of B for (int i = 0; i < face_count; i++) { - Vector3 axis = p_transform_b.xform(faces[i].plane).normal; + Vector3 axis = b_xform_normal.xform(faces[i].plane.normal).normalized(); if (!separator.test_axis(axis)) { return; @@ -1733,9 +1739,12 @@ static void _collision_capsule_convex_polygon(const Shape3DSW *p_a, const Transf int edge_count = mesh.edges.size(); const Vector3 *vertices = mesh.vertices.ptr(); + // Precalculating this makes the transforms faster. + Basis b_xform_normal = p_transform_b.basis.inverse().transposed(); + // faces of B for (int i = 0; i < face_count; i++) { - Vector3 axis = p_transform_b.xform(faces[i].plane).normal; + Vector3 axis = b_xform_normal.xform(faces[i].plane.normal).normalized(); if (!separator.test_axis(axis)) { return; @@ -2057,20 +2066,24 @@ static void _collision_convex_polygon_convex_polygon(const Shape3DSW *p_a, const const Vector3 *vertices_B = mesh_B.vertices.ptr(); int vertex_count_B = mesh_B.vertices.size(); + // Precalculating this makes the transforms faster. + Basis a_xform_normal = p_transform_b.basis.inverse().transposed(); + // faces of A for (int i = 0; i < face_count_A; i++) { - Vector3 axis = p_transform_a.xform(faces_A[i].plane).normal; - //Vector3 axis = p_transform_a.basis.xform( faces_A[i].plane.normal ).normalized(); + Vector3 axis = a_xform_normal.xform(faces_A[i].plane.normal).normalized(); if (!separator.test_axis(axis)) { return; } } + // Precalculating this makes the transforms faster. + Basis b_xform_normal = p_transform_b.basis.inverse().transposed(); + // faces of B for (int i = 0; i < face_count_B; i++) { - Vector3 axis = p_transform_b.xform(faces_B[i].plane).normal; - //Vector3 axis = p_transform_b.basis.xform( faces_B[i].plane.normal ).normalized(); + Vector3 axis = b_xform_normal.xform(faces_B[i].plane.normal).normalized(); if (!separator.test_axis(axis)) { return; diff --git a/servers/physics_3d/shape_3d_sw.cpp b/servers/physics_3d/shape_3d_sw.cpp index 04a174f9c8..5b3721861e 100644 --- a/servers/physics_3d/shape_3d_sw.cpp +++ b/servers/physics_3d/shape_3d_sw.cpp @@ -1635,7 +1635,7 @@ ConcavePolygonShape3DSW::ConcavePolygonShape3DSW() { /* HEIGHT MAP SHAPE */ -Vector<float> HeightMapShape3DSW::get_heights() const { +Vector<real_t> HeightMapShape3DSW::get_heights() const { return heights; } @@ -1932,7 +1932,7 @@ Vector3 HeightMapShape3DSW::get_moment_of_inertia(real_t p_mass) const { (p_mass / 3.0) * (extents.x * extents.x + extents.y * extents.y)); } -void HeightMapShape3DSW::_setup(const Vector<float> &p_heights, int p_width, int p_depth, real_t p_min_height, real_t p_max_height) { +void HeightMapShape3DSW::_setup(const Vector<real_t> &p_heights, int p_width, int p_depth, real_t p_min_height, real_t p_max_height) { heights = p_heights; width = p_width; depth = p_depth; @@ -1966,8 +1966,12 @@ void HeightMapShape3DSW::set_data(const Variant &p_data) { ERR_FAIL_COND(depth <= 0.0); Variant heights_variant = d["heights"]; - Vector<float> heights_buffer; + Vector<real_t> heights_buffer; +#ifdef REAL_T_IS_DOUBLE + if (heights_variant.get_type() == Variant::PACKED_FLOAT64_ARRAY) { +#else if (heights_variant.get_type() == Variant::PACKED_FLOAT32_ARRAY) { +#endif // Ready-to-use heights can be passed. heights_buffer = heights_variant; } else if (heights_variant.get_type() == Variant::OBJECT) { @@ -1980,13 +1984,17 @@ void HeightMapShape3DSW::set_data(const Variant &p_data) { PackedByteArray im_data = image->get_data(); heights_buffer.resize(image->get_width() * image->get_height()); - float *w = heights_buffer.ptrw(); - float *rp = (float *)im_data.ptr(); + real_t *w = heights_buffer.ptrw(); + real_t *rp = (real_t *)im_data.ptr(); for (int i = 0; i < heights_buffer.size(); ++i) { w[i] = rp[i]; } } else { +#ifdef REAL_T_IS_DOUBLE + ERR_FAIL_MSG("Expected PackedFloat64Array or float Image."); +#else ERR_FAIL_MSG("Expected PackedFloat32Array or float Image."); +#endif } // Compute min and max heights or use precomputed values. diff --git a/servers/physics_3d/shape_3d_sw.h b/servers/physics_3d/shape_3d_sw.h index 0d1b7cc3d7..72be34b788 100644 --- a/servers/physics_3d/shape_3d_sw.h +++ b/servers/physics_3d/shape_3d_sw.h @@ -389,12 +389,12 @@ public: }; struct HeightMapShape3DSW : public ConcaveShape3DSW { - Vector<float> heights; + Vector<real_t> heights; int width = 0; int depth = 0; Vector3 local_origin; - _FORCE_INLINE_ float _get_height(int p_x, int p_z) const { + _FORCE_INLINE_ real_t _get_height(int p_x, int p_z) const { return heights[(p_z * width) + p_x]; } @@ -406,10 +406,10 @@ struct HeightMapShape3DSW : public ConcaveShape3DSW { void _get_cell(const Vector3 &p_point, int &r_x, int &r_y, int &r_z) const; - void _setup(const Vector<float> &p_heights, int p_width, int p_depth, real_t p_min_height, real_t p_max_height); + void _setup(const Vector<real_t> &p_heights, int p_width, int p_depth, real_t p_min_height, real_t p_max_height); public: - Vector<float> get_heights() const; + Vector<real_t> get_heights() const; int get_width() const; int get_depth() const; diff --git a/servers/physics_server_2d.cpp b/servers/physics_server_2d.cpp index e8272f377e..2e44a1a10b 100644 --- a/servers/physics_server_2d.cpp +++ b/servers/physics_server_2d.cpp @@ -89,6 +89,8 @@ void PhysicsDirectBodyState2D::_bind_methods() { ClassDB::bind_method(D_METHOD("set_transform", "transform"), &PhysicsDirectBodyState2D::set_transform); ClassDB::bind_method(D_METHOD("get_transform"), &PhysicsDirectBodyState2D::get_transform); + ClassDB::bind_method(D_METHOD("get_velocity_at_local_position", "local_position"), &PhysicsDirectBodyState2D::get_velocity_at_local_position); + ClassDB::bind_method(D_METHOD("add_central_force", "force"), &PhysicsDirectBodyState2D::add_central_force); ClassDB::bind_method(D_METHOD("add_force", "force", "position"), &PhysicsDirectBodyState2D::add_force, Vector2()); ClassDB::bind_method(D_METHOD("add_torque", "torque"), &PhysicsDirectBodyState2D::add_torque); diff --git a/servers/physics_server_2d.h b/servers/physics_server_2d.h index 8542b54838..fafde62599 100644 --- a/servers/physics_server_2d.h +++ b/servers/physics_server_2d.h @@ -60,6 +60,8 @@ public: virtual void set_transform(const Transform2D &p_transform) = 0; virtual Transform2D get_transform() const = 0; + virtual Vector2 get_velocity_at_local_position(const Vector2 &p_position) const = 0; + virtual void add_central_force(const Vector2 &p_force) = 0; virtual void add_force(const Vector2 &p_force, const Vector2 &p_position = Vector2()) = 0; virtual void add_torque(real_t p_torque) = 0; diff --git a/servers/physics_server_3d.cpp b/servers/physics_server_3d.cpp index 3ed8841119..63a51dd9d6 100644 --- a/servers/physics_server_3d.cpp +++ b/servers/physics_server_3d.cpp @@ -91,6 +91,8 @@ void PhysicsDirectBodyState3D::_bind_methods() { ClassDB::bind_method(D_METHOD("set_transform", "transform"), &PhysicsDirectBodyState3D::set_transform); ClassDB::bind_method(D_METHOD("get_transform"), &PhysicsDirectBodyState3D::get_transform); + ClassDB::bind_method(D_METHOD("get_velocity_at_local_position", "local_position"), &PhysicsDirectBodyState3D::get_velocity_at_local_position); + ClassDB::bind_method(D_METHOD("add_central_force", "force"), &PhysicsDirectBodyState3D::add_central_force, Vector3()); ClassDB::bind_method(D_METHOD("add_force", "force", "position"), &PhysicsDirectBodyState3D::add_force, Vector3()); ClassDB::bind_method(D_METHOD("add_torque", "torque"), &PhysicsDirectBodyState3D::add_torque); diff --git a/servers/physics_server_3d.h b/servers/physics_server_3d.h index 17bae9a057..3fb01b301b 100644 --- a/servers/physics_server_3d.h +++ b/servers/physics_server_3d.h @@ -62,6 +62,8 @@ public: virtual void set_transform(const Transform3D &p_transform) = 0; virtual Transform3D get_transform() const = 0; + virtual Vector3 get_velocity_at_local_position(const Vector3 &p_position) const = 0; + virtual void add_central_force(const Vector3 &p_force) = 0; virtual void add_force(const Vector3 &p_force, const Vector3 &p_position = Vector3()) = 0; virtual void add_torque(const Vector3 &p_torque) = 0; diff --git a/servers/rendering/rasterizer_dummy.h b/servers/rendering/rasterizer_dummy.h index b7cf0983af..b2fffca3ea 100644 --- a/servers/rendering/rasterizer_dummy.h +++ b/servers/rendering/rasterizer_dummy.h @@ -540,13 +540,13 @@ public: void particles_emit(RID p_particles, const Transform3D &p_transform, const Vector3 &p_velocity, const Color &p_color, const Color &p_custom, uint32_t p_emit_flags) override {} void particles_set_emitting(RID p_particles, bool p_emitting) override {} void particles_set_amount(RID p_particles, int p_amount) override {} - void particles_set_lifetime(RID p_particles, float p_lifetime) override {} + void particles_set_lifetime(RID p_particles, double p_lifetime) override {} void particles_set_one_shot(RID p_particles, bool p_one_shot) override {} - void particles_set_pre_process_time(RID p_particles, float p_time) override {} + void particles_set_pre_process_time(RID p_particles, double p_time) override {} void particles_set_explosiveness_ratio(RID p_particles, float p_ratio) override {} void particles_set_randomness_ratio(RID p_particles, float p_ratio) override {} void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb) override {} - void particles_set_speed_scale(RID p_particles, float p_scale) override {} + void particles_set_speed_scale(RID p_particles, double p_scale) override {} void particles_set_use_local_coordinates(RID p_particles, bool p_enable) override {} void particles_set_process_material(RID p_particles, RID p_material) override {} void particles_set_fixed_fps(RID p_particles, int p_fps) override {} @@ -724,7 +724,7 @@ public: class RasterizerDummy : public RendererCompositor { private: uint64_t frame = 1; - float delta = 0; + double delta = 0; protected: RasterizerCanvasDummy canvas; @@ -765,7 +765,7 @@ public: bool is_low_end() const override { return true; } uint64_t get_frame_number() const override { return frame; } - float get_frame_delta_time() const override { return delta; } + double get_frame_delta_time() const override { return delta; } RasterizerDummy() {} ~RasterizerDummy() {} diff --git a/servers/rendering/renderer_canvas_cull.cpp b/servers/rendering/renderer_canvas_cull.cpp index ec0a8347f8..fd7d5b91fa 100644 --- a/servers/rendering/renderer_canvas_cull.cpp +++ b/servers/rendering/renderer_canvas_cull.cpp @@ -639,17 +639,19 @@ void RendererCanvasCull::canvas_item_add_polyline(RID p_item, const Vector<Point j2 = j + 1; - Vector2 tangent = ((t + prev_t).normalized()) * p_width * 0.5; + Vector2 dir = (t + prev_t).normalized(); + Vector2 tangent = dir * p_width * 0.5; + Vector2 border = dir * 2.0; Vector2 pos = p_points[i]; points_ptr[j] = pos + tangent; points_ptr[j2] = pos - tangent; - points_top_ptr[j] = pos + tangent + tangent; + points_top_ptr[j] = pos + tangent + border; points_top_ptr[j2] = pos + tangent; points_bottom_ptr[j] = pos - tangent; - points_bottom_ptr[j2] = pos - tangent - tangent; + points_bottom_ptr[j2] = pos - tangent - border; if (i < p_colors.size()) { color = p_colors[i]; diff --git a/servers/rendering/renderer_compositor.h b/servers/rendering/renderer_compositor.h index 5fe9cdffba..6206849b66 100644 --- a/servers/rendering/renderer_compositor.h +++ b/servers/rendering/renderer_compositor.h @@ -85,7 +85,7 @@ public: virtual void end_frame(bool p_swap_buffers) = 0; virtual void finalize() = 0; virtual uint64_t get_frame_number() const = 0; - virtual float get_frame_delta_time() const = 0; + virtual double get_frame_delta_time() const = 0; virtual bool is_low_end() const = 0; virtual bool is_xr_enabled() const; diff --git a/servers/rendering/renderer_rd/effects_rd.cpp b/servers/rendering/renderer_rd/effects_rd.cpp index 47bb756d55..c609f25e31 100644 --- a/servers/rendering/renderer_rd/effects_rd.cpp +++ b/servers/rendering/renderer_rd/effects_rd.cpp @@ -37,6 +37,10 @@ #include "servers/rendering/renderer_rd/renderer_compositor_rd.h" #include "thirdparty/misc/cubemap_coeffs.h" +bool EffectsRD::get_prefer_raster_effects() { + return prefer_raster_effects; +} + 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++) { @@ -66,6 +70,28 @@ RID EffectsRD::_get_uniform_set_from_image(RID p_image) { return uniform_set; } +RID EffectsRD::_get_uniform_set_for_input(RID p_texture) { + if (input_to_uniform_set_cache.has(p_texture)) { + RID uniform_set = input_to_uniform_set_cache[p_texture]; + if (RD::get_singleton()->uniform_set_is_valid(uniform_set)) { + return uniform_set; + } + } + + Vector<RD::Uniform> uniforms; + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_INPUT_ATTACHMENT; + u.binding = 0; + u.ids.push_back(p_texture); + uniforms.push_back(u); + // This is specific to our subpass shader + RID uniform_set = RD::get_singleton()->uniform_set_create(uniforms, tonemap.shader.version_get_shader(tonemap.shader_version, TONEMAP_MODE_SUBPASS), 0); + + input_to_uniform_set_cache[p_texture] = uniform_set; + + return uniform_set; +} + RID EffectsRD::_get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps) { if (texture_to_uniform_set_cache.has(p_texture)) { RID uniform_set = texture_to_uniform_set_cache[p_texture]; @@ -81,7 +107,7 @@ RID EffectsRD::_get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps) u.ids.push_back(p_use_mipmaps ? default_mipmap_sampler : default_sampler); u.ids.push_back(p_texture); uniforms.push_back(u); - //anything with the same configuration (one texture in binding 0 for set 0), is good + // anything with the same configuration (one texture in binding 0 for set 0), is good RID uniform_set = RD::get_singleton()->uniform_set_create(uniforms, tonemap.shader.version_get_shader(tonemap.shader_version, 0), 0); texture_to_uniform_set_cache[p_texture] = uniform_set; @@ -454,7 +480,7 @@ void EffectsRD::gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const } void EffectsRD::gaussian_glow_raster(RID p_source_rd_texture, RID p_framebuffer_half, RID p_rd_texture_half, RID p_dest_framebuffer, const Vector2 &p_pixel_size, float p_strength, bool p_high_quality, bool p_first_pass, float p_luminance_cap, float p_exposure, float p_bloom, float p_hdr_bleed_treshold, float p_hdr_bleed_scale, RID p_auto_exposure, float p_auto_exposure_grey) { - ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use the fragment version of the gaussian glow with the clustered renderer."); + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use the raster version of the gaussian glow with the clustered renderer."); memset(&blur_raster.push_constant, 0, sizeof(BlurRasterPushConstant)); @@ -774,11 +800,11 @@ void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const Tone if (p_settings.view_count > 1) { // Use MULTIVIEW versions - mode += 4; + mode += 6; } RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dst_framebuffer, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_DROP, RD::FINAL_ACTION_DISCARD); - RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer))); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dst_framebuffer), false, RD::get_singleton()->draw_list_get_current_pass())); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_color), 0); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_settings.exposure_texture), 1); RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_settings.glow_texture, true), 2); @@ -790,6 +816,45 @@ void EffectsRD::tonemapper(RID p_source_color, RID p_dst_framebuffer, const Tone RD::get_singleton()->draw_list_end(); } +void EffectsRD::tonemapper(RD::DrawListID p_subpass_draw_list, RID p_source_color, RD::FramebufferFormatID p_dst_format_id, const TonemapSettings &p_settings) { + memset(&tonemap.push_constant, 0, sizeof(TonemapPushConstant)); + + tonemap.push_constant.use_bcs = p_settings.use_bcs; + tonemap.push_constant.bcs[0] = p_settings.brightness; + tonemap.push_constant.bcs[1] = p_settings.contrast; + tonemap.push_constant.bcs[2] = p_settings.saturation; + + ERR_FAIL_COND_MSG(p_settings.use_glow, "Glow is not supported when using subpasses."); + tonemap.push_constant.use_glow = p_settings.use_glow; + + int mode = p_settings.use_1d_color_correction ? TONEMAP_MODE_SUBPASS_1D_LUT : TONEMAP_MODE_SUBPASS; + if (p_settings.view_count > 1) { + // Use MULTIVIEW versions + mode += 6; + } + + tonemap.push_constant.tonemapper = p_settings.tonemap_mode; + tonemap.push_constant.use_auto_exposure = p_settings.use_auto_exposure; + tonemap.push_constant.exposure = p_settings.exposure; + tonemap.push_constant.white = p_settings.white; + tonemap.push_constant.auto_exposure_grey = p_settings.auto_exposure_grey; + + tonemap.push_constant.use_color_correction = p_settings.use_color_correction; + + tonemap.push_constant.use_debanding = p_settings.use_debanding; + + RD::get_singleton()->draw_list_bind_render_pipeline(p_subpass_draw_list, tonemap.pipelines[mode].get_render_pipeline(RD::INVALID_ID, p_dst_format_id, false, RD::get_singleton()->draw_list_get_current_pass())); + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_for_input(p_source_color), 0); + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.exposure_texture), 1); // should be set to a default texture, it's ignored + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.glow_texture, true), 2); // should be set to a default texture, it's ignored + RD::get_singleton()->draw_list_bind_uniform_set(p_subpass_draw_list, _get_uniform_set_from_texture(p_settings.color_correction_texture), 3); + + RD::get_singleton()->draw_list_bind_index_array(p_subpass_draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(p_subpass_draw_list, &tonemap.push_constant, sizeof(TonemapPushConstant)); + RD::get_singleton()->draw_list_draw(p_subpass_draw_list, true); +} + void EffectsRD::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) { ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute version of luminance reduction with the mobile renderer."); @@ -832,7 +897,7 @@ void EffectsRD::luminance_reduction(RID p_source_texture, const Size2i p_source_ } void EffectsRD::luminance_reduction_raster(RID p_source_texture, const Size2i p_source_size, const Vector<RID> p_reduce, Vector<RID> p_fb, RID p_prev_luminance, float p_min_luminance, float p_max_luminance, float p_adjust, bool p_set) { - ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use fragment version of luminance reduction with the clustered renderer."); + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster version of luminance reduction with the clustered renderer."); ERR_FAIL_COND_MSG(p_reduce.size() != p_fb.size(), "Incorrect frame buffer account for luminance reduction."); luminance_reduce_raster.push_constant.max_luminance = p_max_luminance; @@ -1447,7 +1512,9 @@ void EffectsRD::roughness_limit(RID p_source_normal, RID p_roughness, const Size RD::get_singleton()->compute_list_end(); } -void EffectsRD::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 EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_texture, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap roughness with the mobile renderer."); + memset(&roughness.push_constant, 0, sizeof(CubemapRoughnessPushConstant)); roughness.push_constant.face_id = p_face_id > 9 ? 0 : p_face_id; @@ -1457,10 +1524,10 @@ void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffe roughness.push_constant.face_size = p_size; RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, roughness.pipeline); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, roughness.compute_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_framebuffer), 1); + 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, &roughness.push_constant, sizeof(CubemapRoughnessPushConstant)); @@ -1472,11 +1539,37 @@ void EffectsRD::cubemap_roughness(RID p_source_rd_texture, RID p_dest_framebuffe RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_roughness_raster(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) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap roughness with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap roughness must process one side at a time."); + + memset(&roughness.push_constant, 0, sizeof(CubemapRoughnessPushConstant)); + + roughness.push_constant.face_id = p_face_id; + roughness.push_constant.roughness = p_roughness; + roughness.push_constant.sample_count = p_sample_count; + roughness.push_constant.use_direct_write = p_roughness == 0.0; + roughness.push_constant.face_size = p_size; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, roughness.raster_pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_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::get_singleton()->draw_list_set_push_constant(draw_list, &roughness.push_constant, sizeof(CubemapRoughnessPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, const Size2i &p_size) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap downsample with the mobile renderer."); + cubemap_downsampler.push_constant.face_size = p_size.x; + cubemap_downsampler.push_constant.face_id = 0; // we render all 6 sides to each layer in one call RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, cubemap_downsampler.pipeline); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, cubemap_downsampler.compute_pipeline); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_cubemap), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_uniform_set_from_image(p_dest_cubemap), 1); @@ -1490,7 +1583,27 @@ void EffectsRD::cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, con RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_downsample_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, const Size2i &p_size) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap downsample with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap downsample must process one side at a time."); + + cubemap_downsampler.push_constant.face_size = p_size.x; + cubemap_downsampler.push_constant.face_id = p_face_id; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, cubemap_downsampler.raster_pipeline.get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_cubemap), 0); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &cubemap_downsampler.push_constant, sizeof(CubemapDownsamplerPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, bool p_use_array) { + ERR_FAIL_COND_MSG(prefer_raster_effects, "Can't use compute based cubemap filter with the mobile renderer."); + Vector<RD::Uniform> uniforms; for (int i = 0; i < p_dest_cubemap.size(); i++) { RD::Uniform u; @@ -1502,12 +1615,12 @@ void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, if (RD::get_singleton()->uniform_set_is_valid(filter.image_uniform_set)) { RD::get_singleton()->free(filter.image_uniform_set); } - filter.image_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.shader.version_get_shader(filter.shader_version, 0), 2); + filter.image_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.compute_shader.version_get_shader(filter.shader_version, 0), 2); int pipeline = p_use_array ? FILTER_MODE_HIGH_QUALITY_ARRAY : FILTER_MODE_HIGH_QUALITY; pipeline = filter.use_high_quality ? pipeline : pipeline + 1; RD::ComputeListID compute_list = RD::get_singleton()->compute_list_begin(); - RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, filter.pipelines[pipeline]); + RD::get_singleton()->compute_list_bind_compute_pipeline(compute_list, filter.compute_pipelines[pipeline]); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, _get_compute_uniform_set_from_texture(p_source_cubemap, true), 0); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, filter.uniform_set, 1); RD::get_singleton()->compute_list_bind_uniform_set(compute_list, filter.image_uniform_set, 2); @@ -1519,6 +1632,29 @@ void EffectsRD::cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, RD::get_singleton()->compute_list_end(); } +void EffectsRD::cubemap_filter_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_mip_level) { + ERR_FAIL_COND_MSG(!prefer_raster_effects, "Can't use raster based cubemap filter with the clustered renderer."); + ERR_FAIL_COND_MSG(p_face_id >= 6, "Raster implementation of cubemap filter must process one side at a time."); + + // TODO implement! + CubemapFilterRasterPushConstant push_constant; + push_constant.mip_level = p_mip_level; + push_constant.face_id = p_face_id; + + CubemapFilterMode mode = filter.use_high_quality ? FILTER_MODE_HIGH_QUALITY : FILTER_MODE_LOW_QUALITY; + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(p_dest_framebuffer, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_KEEP, RD::FINAL_ACTION_DISCARD); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, filter.raster_pipelines[mode].get_render_pipeline(RD::INVALID_ID, RD::get_singleton()->framebuffer_get_format(p_dest_framebuffer))); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, _get_uniform_set_from_texture(p_source_cubemap), 0); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, filter.uniform_set, 1); + RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); + + RD::get_singleton()->draw_list_set_push_constant(draw_list, &push_constant, sizeof(CubemapFilterRasterPushConstant)); + + RD::get_singleton()->draw_list_draw(draw_list, true); + RD::get_singleton()->draw_list_end(); +} + void EffectsRD::resolve_gi(RID p_source_depth, RID p_source_normal_roughness, RID p_source_voxel_gi, RID p_dest_depth, RID p_dest_normal_roughness, RID p_dest_voxel_gi, Vector2i p_screen_size, int p_samples, uint32_t p_barrier) { ResolvePushConstant push_constant; push_constant.screen_size[0] = p_screen_size.x; @@ -1713,11 +1849,22 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { // Initialize roughness Vector<String> cubemap_roughness_modes; cubemap_roughness_modes.push_back(""); - roughness.shader.initialize(cubemap_roughness_modes); - roughness.shader_version = roughness.shader.version_create(); + if (prefer_raster_effects) { + roughness.raster_shader.initialize(cubemap_roughness_modes); + + roughness.shader_version = roughness.raster_shader.version_create(); + + roughness.raster_pipeline.setup(roughness.raster_shader.version_get_shader(roughness.shader_version, 0), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); - roughness.pipeline = RD::get_singleton()->compute_pipeline_create(roughness.shader.version_get_shader(roughness.shader_version, 0)); + } else { + roughness.compute_shader.initialize(cubemap_roughness_modes); + + roughness.shader_version = roughness.compute_shader.version_create(); + + roughness.compute_pipeline = RD::get_singleton()->compute_pipeline_create(roughness.compute_shader.version_get_shader(roughness.shader_version, 0)); + roughness.raster_pipeline.clear(); + } } { @@ -1727,12 +1874,16 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { tonemap_modes.push_back("\n#define USE_GLOW_FILTER_BICUBIC\n"); tonemap_modes.push_back("\n#define USE_1D_LUT\n"); tonemap_modes.push_back("\n#define USE_GLOW_FILTER_BICUBIC\n#define USE_1D_LUT\n"); + tonemap_modes.push_back("\n#define SUBPASS\n"); + tonemap_modes.push_back("\n#define SUBPASS\n#define USE_1D_LUT\n"); // multiview versions of our shaders tonemap_modes.push_back("\n#define MULTIVIEW\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_GLOW_FILTER_BICUBIC\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_1D_LUT\n"); tonemap_modes.push_back("\n#define MULTIVIEW\n#define USE_GLOW_FILTER_BICUBIC\n#define USE_1D_LUT\n"); + tonemap_modes.push_back("\n#define MULTIVIEW\n#define SUBPASS\n"); + tonemap_modes.push_back("\n#define MULTIVIEW\n#define SUBPASS\n#define USE_1D_LUT\n"); tonemap.shader.initialize(tonemap_modes); @@ -1741,6 +1892,8 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { tonemap.shader.set_variant_enabled(TONEMAP_MODE_BICUBIC_GLOW_FILTER_MULTIVIEW, false); tonemap.shader.set_variant_enabled(TONEMAP_MODE_1D_LUT_MULTIVIEW, false); tonemap.shader.set_variant_enabled(TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT_MULTIVIEW, false); + tonemap.shader.set_variant_enabled(TONEMAP_MODE_SUBPASS_MULTIVIEW, false); + tonemap.shader.set_variant_enabled(TONEMAP_MODE_SUBPASS_1D_LUT_MULTIVIEW, false); } tonemap.shader_version = tonemap.shader.version_create(); @@ -1983,11 +2136,21 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { //Initialize cubemap downsampler Vector<String> cubemap_downsampler_modes; cubemap_downsampler_modes.push_back(""); - cubemap_downsampler.shader.initialize(cubemap_downsampler_modes); - cubemap_downsampler.shader_version = cubemap_downsampler.shader.version_create(); + if (prefer_raster_effects) { + cubemap_downsampler.raster_shader.initialize(cubemap_downsampler_modes); + + cubemap_downsampler.shader_version = cubemap_downsampler.raster_shader.version_create(); + + cubemap_downsampler.raster_pipeline.setup(cubemap_downsampler.raster_shader.version_get_shader(cubemap_downsampler.shader_version, 0), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } else { + cubemap_downsampler.compute_shader.initialize(cubemap_downsampler_modes); + + cubemap_downsampler.shader_version = cubemap_downsampler.compute_shader.version_create(); - cubemap_downsampler.pipeline = RD::get_singleton()->compute_pipeline_create(cubemap_downsampler.shader.version_get_shader(cubemap_downsampler.shader_version, 0)); + cubemap_downsampler.compute_pipeline = RD::get_singleton()->compute_pipeline_create(cubemap_downsampler.compute_shader.version_get_shader(cubemap_downsampler.shader_version, 0)); + cubemap_downsampler.raster_pipeline.clear(); + } } { @@ -1999,12 +2162,6 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { cubemap_filter_modes.push_back("\n#define USE_LOW_QUALITY\n"); cubemap_filter_modes.push_back("\n#define USE_HIGH_QUALITY\n#define USE_TEXTURE_ARRAY\n"); cubemap_filter_modes.push_back("\n#define USE_LOW_QUALITY\n#define USE_TEXTURE_ARRAY\n"); - filter.shader.initialize(cubemap_filter_modes); - filter.shader_version = filter.shader.version_create(); - - for (int i = 0; i < FILTER_MODE_MAX; i++) { - filter.pipelines[i] = RD::get_singleton()->compute_pipeline_create(filter.shader.version_get_shader(filter.shader_version, i)); - } if (filter.use_high_quality) { filter.coefficient_buffer = RD::get_singleton()->storage_buffer_create(sizeof(high_quality_coeffs)); @@ -2014,15 +2171,50 @@ EffectsRD::EffectsRD(bool p_prefer_raster_effects) { RD::get_singleton()->buffer_update(filter.coefficient_buffer, 0, sizeof(low_quality_coeffs), &low_quality_coeffs[0]); } - Vector<RD::Uniform> uniforms; - { - RD::Uniform u; - u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; - u.binding = 0; - u.ids.push_back(filter.coefficient_buffer); - uniforms.push_back(u); + if (prefer_raster_effects) { + filter.raster_shader.initialize(cubemap_filter_modes); + filter.shader_version = filter.raster_shader.version_create(); + + // array variants are not supported in raster + filter.raster_shader.set_variant_enabled(FILTER_MODE_HIGH_QUALITY_ARRAY, false); + filter.raster_shader.set_variant_enabled(FILTER_MODE_LOW_QUALITY_ARRAY, false); + + for (int i = 0; i < FILTER_MODE_MAX; i++) { + if (filter.raster_shader.is_variant_enabled(i)) { + filter.raster_pipelines[i].setup(filter.raster_shader.version_get_shader(filter.shader_version, i), RD::RENDER_PRIMITIVE_TRIANGLES, RD::PipelineRasterizationState(), RD::PipelineMultisampleState(), RD::PipelineDepthStencilState(), RD::PipelineColorBlendState::create_disabled(), 0); + } else { + filter.raster_pipelines[i].clear(); + } + } + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 0; + u.ids.push_back(filter.coefficient_buffer); + uniforms.push_back(u); + } + filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.raster_shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); + } else { + filter.compute_shader.initialize(cubemap_filter_modes); + filter.shader_version = filter.compute_shader.version_create(); + + for (int i = 0; i < FILTER_MODE_MAX; i++) { + filter.compute_pipelines[i] = RD::get_singleton()->compute_pipeline_create(filter.compute_shader.version_get_shader(filter.shader_version, i)); + filter.raster_pipelines[i].clear(); + } + + Vector<RD::Uniform> uniforms; + { + RD::Uniform u; + u.uniform_type = RD::UNIFORM_TYPE_STORAGE_BUFFER; + u.binding = 0; + u.ids.push_back(filter.coefficient_buffer); + uniforms.push_back(u); + } + filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.compute_shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); } - filter.uniform_set = RD::get_singleton()->uniform_set_create(uniforms, filter.shader.version_get_shader(filter.shader_version, filter.use_high_quality ? 0 : 1), 1); } { @@ -2199,17 +2391,20 @@ EffectsRD::~EffectsRD() { if (prefer_raster_effects) { blur_raster.shader.version_free(blur_raster.shader_version); luminance_reduce_raster.shader.version_free(luminance_reduce_raster.shader_version); + roughness.raster_shader.version_free(roughness.shader_version); + cubemap_downsampler.raster_shader.version_free(cubemap_downsampler.shader_version); + filter.raster_shader.version_free(filter.shader_version); } else { bokeh.shader.version_free(bokeh.shader_version); luminance_reduce.shader.version_free(luminance_reduce.shader_version); + roughness.compute_shader.version_free(roughness.shader_version); + cubemap_downsampler.compute_shader.version_free(cubemap_downsampler.shader_version); + filter.compute_shader.version_free(filter.shader_version); } copy.shader.version_free(copy.shader_version); copy_to_fb.shader.version_free(copy_to_fb.shader_version); cube_to_dp.shader.version_free(cube_to_dp.shader_version); - cubemap_downsampler.shader.version_free(cubemap_downsampler.shader_version); - filter.shader.version_free(filter.shader_version); resolve.shader.version_free(resolve.shader_version); - roughness.shader.version_free(roughness.shader_version); roughness_limiter.shader.version_free(roughness_limiter.shader_version); sort.shader.version_free(sort.shader_version); specular_merge.shader.version_free(specular_merge.shader_version); diff --git a/servers/rendering/renderer_rd/effects_rd.h b/servers/rendering/renderer_rd/effects_rd.h index d072564c23..79d9a8b738 100644 --- a/servers/rendering/renderer_rd/effects_rd.h +++ b/servers/rendering/renderer_rd/effects_rd.h @@ -39,8 +39,11 @@ #include "servers/rendering/renderer_rd/shaders/copy_to_fb.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cube_to_dp.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_filter.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl.gen.h" +#include "servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/luminance_reduce.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/luminance_reduce_raster.glsl.gen.h" #include "servers/rendering/renderer_rd/shaders/resolve.glsl.gen.h" @@ -62,6 +65,9 @@ #include "servers/rendering_server.h" class EffectsRD { +private: + bool prefer_raster_effects; + enum BlurRasterMode { BLUR_MODE_GAUSSIAN_BLUR, BLUR_MODE_GAUSSIAN_GLOW, @@ -220,9 +226,11 @@ class EffectsRD { struct CubemapRoughness { CubemapRoughnessPushConstant push_constant; - CubemapRoughnessShaderRD shader; + CubemapRoughnessShaderRD compute_shader; + CubemapRoughnessRasterShaderRD raster_shader; RID shader_version; - RID pipeline; + RID compute_pipeline; + PipelineCacheRD raster_pipeline; } roughness; enum TonemapMode { @@ -230,11 +238,15 @@ class EffectsRD { TONEMAP_MODE_BICUBIC_GLOW_FILTER, TONEMAP_MODE_1D_LUT, TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT, + TONEMAP_MODE_SUBPASS, + TONEMAP_MODE_SUBPASS_1D_LUT, TONEMAP_MODE_NORMAL_MULTIVIEW, TONEMAP_MODE_BICUBIC_GLOW_FILTER_MULTIVIEW, TONEMAP_MODE_1D_LUT_MULTIVIEW, TONEMAP_MODE_BICUBIC_GLOW_FILTER_1D_LUT_MULTIVIEW, + TONEMAP_MODE_SUBPASS_MULTIVIEW, + TONEMAP_MODE_SUBPASS_1D_LUT_MULTIVIEW, TONEMAP_MODE_MAX }; @@ -508,15 +520,17 @@ class EffectsRD { struct CubemapDownsamplerPushConstant { uint32_t face_size; - float pad[3]; + uint32_t face_id; + float pad[2]; }; struct CubemapDownsampler { CubemapDownsamplerPushConstant push_constant; - CubemapDownsamplerShaderRD shader; + CubemapDownsamplerShaderRD compute_shader; + CubemapDownsamplerRasterShaderRD raster_shader; RID shader_version; - RID pipeline; - + RID compute_pipeline; + PipelineCacheRD raster_pipeline; } cubemap_downsampler; enum CubemapFilterMode { @@ -527,10 +541,19 @@ class EffectsRD { FILTER_MODE_MAX, }; + struct CubemapFilterRasterPushConstant { + uint32_t mip_level; + uint32_t face_id; + float pad[2]; + }; + struct CubemapFilter { - CubemapFilterShaderRD shader; + CubemapFilterShaderRD compute_shader; + CubemapFilterRasterShaderRD raster_shader; RID shader_version; - RID pipelines[FILTER_MODE_MAX]; + RID compute_pipelines[FILTER_MODE_MAX]; + PipelineCacheRD raster_pipelines[FILTER_MODE_MAX]; + RID uniform_set; RID image_uniform_set; RID coefficient_buffer; @@ -699,6 +722,7 @@ class EffectsRD { RID index_array; Map<RID, RID> texture_to_uniform_set_cache; + Map<RID, RID> input_to_uniform_set_cache; Map<RID, RID> image_to_uniform_set_cache; @@ -732,15 +756,16 @@ class EffectsRD { Map<TextureSamplerPair, RID> texture_sampler_to_compute_uniform_set_cache; RID _get_uniform_set_from_image(RID p_texture); + RID _get_uniform_set_for_input(RID p_texture); RID _get_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps = false); RID _get_compute_uniform_set_from_texture(RID p_texture, bool p_use_mipmaps = false); RID _get_compute_uniform_set_from_texture_and_sampler(RID p_texture, RID p_sampler); RID _get_compute_uniform_set_from_texture_pair(RID p_texture, RID p_texture2, bool p_use_mipmaps = false); RID _get_compute_uniform_set_from_image_pair(RID p_texture, RID p_texture2); - bool prefer_raster_effects; - public: + bool get_prefer_raster_effects(); + void copy_to_fb_rect(RID p_source_rd_texture, RID p_dest_framebuffer, const Rect2i &p_rect, bool p_flip_y = false, bool p_force_luminance = false, bool p_alpha_to_zero = false, bool p_srgb = false, RID p_secondary = RID()); void copy_to_rect(RID p_source_rd_texture, RID p_dest_texture, const Rect2i &p_rect, bool p_flip_y = false, bool p_force_luminance = false, bool p_all_source = false, bool p_8_bit_dst = false, bool p_alpha_to_one = false); void copy_cubemap_to_panorama(RID p_source_cube, RID p_dest_panorama, const Size2i &p_panorama_size, float p_lod, bool p_is_array); @@ -752,7 +777,8 @@ public: void gaussian_glow(RID p_source_rd_texture, RID p_back_texture, const Size2i &p_size, float p_strength = 1.0, bool p_high_quality = false, bool p_first_pass = false, float p_luminance_cap = 16.0, float p_exposure = 1.0, float p_bloom = 0.0, float p_hdr_bleed_treshold = 1.0, float p_hdr_bleed_scale = 1.0, RID p_auto_exposure = RID(), float p_auto_exposure_grey = 1.0); void gaussian_glow_raster(RID p_source_rd_texture, RID p_framebuffer_half, RID p_rd_texture_half, RID p_dest_framebuffer, const Vector2 &p_pixel_size, float p_strength = 1.0, bool p_high_quality = false, bool p_first_pass = false, float p_luminance_cap = 16.0, float p_exposure = 1.0, float p_bloom = 0.0, float p_hdr_bleed_treshold = 1.0, float p_hdr_bleed_scale = 1.0, RID p_auto_exposure = RID(), float p_auto_exposure_grey = 1.0); - 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 cubemap_roughness(RID p_source_rd_texture, RID p_dest_texture, uint32_t p_face_id, uint32_t p_sample_count, float p_roughness, float p_size); + void cubemap_roughness_raster(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 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); @@ -822,13 +848,16 @@ public: }; void tonemapper(RID p_source_color, RID p_dst_framebuffer, const TonemapSettings &p_settings); + void tonemapper(RD::DrawListID p_subpass_draw_list, RID p_source_color, RD::FramebufferFormatID p_dst_format_id, const TonemapSettings &p_settings); void gather_ssao(RD::ComputeListID p_compute_list, const Vector<RID> p_ao_slices, const SSAOSettings &p_settings, bool p_adaptive_base_pass, RID p_gather_uniform_set, RID p_importance_map_uniform_set); void generate_ssao(RID p_depth_buffer, RID p_normal_buffer, RID p_depth_mipmaps_texture, const Vector<RID> &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, RID &r_downsample_uniform_set, RID &r_gather_uniform_set, RID &r_importance_map_uniform_set); void roughness_limit(RID p_source_normal, RID p_roughness, const Size2i &p_size, float p_curve); void cubemap_downsample(RID p_source_cubemap, RID p_dest_cubemap, const Size2i &p_size); + void cubemap_downsample_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, const Size2i &p_size); void cubemap_filter(RID p_source_cubemap, Vector<RID> p_dest_cubemap, bool p_use_array); + void cubemap_filter_raster(RID p_source_cubemap, RID p_dest_framebuffer, uint32_t p_face_id, uint32_t p_mip_level); void screen_space_reflection(RID p_diffuse, RID p_normal_roughness, RS::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); void merge_specular(RID p_dest_framebuffer, RID p_specular, RID p_base, RID p_reflection); diff --git a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp index ac20515c28..0d304556c2 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.cpp @@ -183,7 +183,7 @@ void RenderForwardClustered::RenderBufferDataForwardClustered::clear() { } } -void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { +void RenderForwardClustered::RenderBufferDataForwardClustered::configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { clear(); ERR_FAIL_COND_MSG(p_view_count != 1, "Multiple views is currently not supported in this renderer, please use the mobile renderer for VR support"); @@ -483,8 +483,8 @@ void RenderForwardClustered::_render_list_template(RenderingDevice::DrawListID p } if (material_uniform_set != prev_material_uniform_set) { - //update uniform set - if (material_uniform_set.is_valid()) { + // Update uniform set. + if (RD::get_singleton()->uniform_set_is_valid(material_uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_uniform_set, MATERIAL_UNIFORM_SET); } @@ -1555,6 +1555,14 @@ void RenderForwardClustered::_render_scene(RenderDataRD *p_render_data, const Co } RD::get_singleton()->draw_command_end_label(); + + if (p_render_data->render_buffers.is_valid()) { + _debug_draw_cluster(p_render_data->render_buffers); + + RENDER_TIMESTAMP("Tonemap"); + + _render_buffers_post_process_and_tonemap(p_render_data); + } } void RenderForwardClustered::_render_shadow_begin() { diff --git a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h index 6682c5e9b0..676f633d33 100644 --- a/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h +++ b/servers/rendering/renderer_rd/forward_clustered/render_forward_clustered.h @@ -112,7 +112,7 @@ class RenderForwardClustered : public RendererSceneRenderRD { void ensure_specular(); void ensure_voxelgi(); void clear(); - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); ~RenderBufferDataForwardClustered(); }; diff --git a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp index 5f6d9465c7..dea2975660 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.cpp @@ -78,12 +78,16 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::clear() { color = RID(); depth = RID(); - color_fb = RID(); + for (int i = 0; i < FB_CONFIG_MAX; i++) { + color_fbs[i] = RID(); + } } -void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { +void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) { clear(); + bool is_half_resolution = false; // Set this once we support this feature. + msaa = p_msaa; width = p_width; @@ -93,15 +97,50 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_b color = p_color_buffer; depth = p_depth_buffer; - RD::DataFormat color_format = RenderForwardMobile::singleton->_render_buffers_get_color_format(); + // We are creating 4 configurations here for our framebuffers. if (p_msaa == RS::VIEWPORT_MSAA_DISABLED) { Vector<RID> fb; - fb.push_back(p_color_buffer); - fb.push_back(depth); + fb.push_back(p_color_buffer); // 0 - color buffer + fb.push_back(depth); // 1 - depth buffer + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + color_fbs[FB_CONFIG_ONE_PASS] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + + // - add sky pass + passes.push_back(pass); + color_fbs[FB_CONFIG_TWO_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + + // - add alpha pass + passes.push_back(pass); + color_fbs[FB_CONFIG_THREE_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); - color_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, view_count); + if (!is_half_resolution) { + // - add blit to 2D pass + fb.push_back(p_target_buffer); // 2 - target buffer + + RD::FramebufferPass blit_pass; + blit_pass.color_attachments.push_back(2); + blit_pass.input_attachments.push_back(0); + passes.push_back(blit_pass); + + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + } else { + // can't do our blit pass if resolutions don't match + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RID(); + } } else { + RD::DataFormat color_format = RenderForwardMobile::singleton->_render_buffers_get_color_format(); + RD::TextureFormat tf; if (view_count > 1) { tf.texture_type = RD::TEXTURE_TYPE_2D_ARRAY; @@ -134,14 +173,87 @@ void RenderForwardMobile::RenderBufferDataForwardMobile::configure(RID p_color_b { Vector<RID> fb; - fb.push_back(color_msaa); - fb.push_back(depth_msaa); + fb.push_back(color_msaa); // 0 - msaa color buffer + fb.push_back(depth_msaa); // 1 - msaa depth buffer + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + + // - add sky pass + fb.push_back(color); // 2 - color buffer + passes.push_back(pass); // without resolve for our 3 + 4 subpass config + { + // but with resolve for our 2 subpass config + Vector<RD::FramebufferPass> two_passes; + two_passes.push_back(pass); // opaque subpass without resolve + pass.resolve_attachments.push_back(2); + two_passes.push_back(pass); // sky subpass with resolve + + color_fbs[FB_CONFIG_TWO_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, two_passes, RenderingDevice::INVALID_ID, view_count); + } + + // - add alpha pass (with resolve, we just added that above) + passes.push_back(pass); + color_fbs[FB_CONFIG_THREE_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); - color_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, view_count); + { + // we also need our one pass with resolve + Vector<RD::FramebufferPass> one_pass_with_resolve; + one_pass_with_resolve.push_back(pass); // note our pass configuration already has resolve.. + color_fbs[FB_CONFIG_ONE_PASS] = RD::get_singleton()->framebuffer_create_multipass(fb, one_pass_with_resolve, RenderingDevice::INVALID_ID, view_count); + } + + if (!is_half_resolution) { + // - add blit to 2D pass + fb.push_back(p_target_buffer); // 3 - target buffer + RD::FramebufferPass blit_pass; + blit_pass.color_attachments.push_back(3); + blit_pass.input_attachments.push_back(2); + passes.push_back(blit_pass); + + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RD::get_singleton()->framebuffer_create_multipass(fb, passes, RenderingDevice::INVALID_ID, view_count); + } else { + // can't do our blit pass if resolutions don't match + color_fbs[FB_CONFIG_FOUR_SUBPASSES] = RID(); + } } } } +RID RenderForwardMobile::reflection_probe_create_framebuffer(RID p_color, RID p_depth) { + // Our attachments + Vector<RID> fb; + fb.push_back(p_color); // 0 + fb.push_back(p_depth); // 1 + + // Now define our subpasses + Vector<RD::FramebufferPass> passes; + RD::FramebufferPass pass; + + // re-using the same attachments + pass.color_attachments.push_back(0); + pass.depth_attachment = 1; + + // - opaque pass + passes.push_back(pass); + + // - sky pass + passes.push_back(pass); + + // - alpha pass + passes.push_back(pass); + + return RD::get_singleton()->framebuffer_create_multipass(fb, passes); +} + RenderForwardMobile::RenderBufferDataForwardMobile::~RenderBufferDataForwardMobile() { clear(); } @@ -365,14 +477,29 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color scene_state.ubo.viewport_size[1] = vp_he.y; scene_state.ubo.directional_light_count = 0; + // We can only use our full subpass approach if we're: + // - not reading from SCREEN_TEXTURE/DEPTH_TEXTURE + // - not using ssr/sss (currently not supported) + // - not using glow or other post effects (can't do 4th subpass) + // - rendering to a half sized render buffer (can't do 4th subpass) + // We'll need to restrict how far we're going with subpasses based on this. + Size2i screen_size; - RID opaque_framebuffer; - RID alpha_framebuffer; + RID framebuffer; bool reverse_cull = false; + bool using_subpass_transparent = true; + bool using_subpass_post_process = true; - // I don't think we support either of these in our mobile renderer so probably should phase them out - bool using_ssr = false; - bool using_sss = false; + bool is_half_resolution = false; // Set this once we support this feature. + bool using_ssr = false; // I don't think we support this in our mobile renderer so probably should phase it out + bool using_sss = false; // I don't think we support this in our mobile renderer so probably should phase it out + + // fill our render lists early so we can find out if we use various features + _fill_render_list(RENDER_LIST_OPAQUE, p_render_data, PASS_MODE_COLOR); + render_list[RENDER_LIST_OPAQUE].sort_by_key(); + render_list[RENDER_LIST_ALPHA].sort_by_reverse_depth_and_priority(); + _fill_element_info(RENDER_LIST_OPAQUE); + _fill_element_info(RENDER_LIST_ALPHA); if (p_render_data->render_info) { p_render_data->render_info->info[RS::VIEWPORT_RENDER_INFO_TYPE_VISIBLE][RS::VIEWPORT_RENDER_INFO_DRAW_CALLS_IN_FRAME] = p_render_data->instances->size(); @@ -384,15 +511,36 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color screen_size.x = render_buffer->width; screen_size.y = render_buffer->height; - opaque_framebuffer = render_buffer->color_fb; - alpha_framebuffer = opaque_framebuffer; + if (is_half_resolution) { + // can't do blit subpass + using_subpass_post_process = false; + } else if (env && (env->glow_enabled || env->auto_exposure || camera_effects_uses_dof(p_render_data->camera_effects))) { + // can't do blit subpass + using_subpass_post_process = false; + } + + if (scene_state.used_screen_texture || scene_state.used_depth_texture) { + // can't use our last two subpasses + using_subpass_transparent = false; + using_subpass_post_process = false; + } + + if (using_subpass_post_process) { + // all as subpasses + framebuffer = render_buffer->color_fbs[FB_CONFIG_FOUR_SUBPASSES]; + } else if (using_subpass_transparent) { + // our tonemap pass is separate + framebuffer = render_buffer->color_fbs[FB_CONFIG_THREE_SUBPASSES]; + } else { + // only opaque and sky as subpasses + framebuffer = render_buffer->color_fbs[FB_CONFIG_TWO_SUBPASSES]; + } } else if (p_render_data->reflection_probe.is_valid()) { uint32_t resolution = reflection_probe_instance_get_resolution(p_render_data->reflection_probe); screen_size.x = resolution; screen_size.y = resolution; - opaque_framebuffer = reflection_probe_instance_get_framebuffer(p_render_data->reflection_probe, p_render_data->reflection_probe_pass); - alpha_framebuffer = opaque_framebuffer; + framebuffer = reflection_probe_instance_get_framebuffer(p_render_data->reflection_probe, p_render_data->reflection_probe_pass); if (storage->reflection_probe_is_interior(reflection_probe_instance_get_probe(p_render_data->reflection_probe))) { p_render_data->environment = RID(); //no environment on interiors @@ -400,6 +548,8 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color } reverse_cull = true; + using_subpass_transparent = true; // we ignore our screen/depth texture here + using_subpass_post_process = false; // not applicable at all for reflection probes. } else { ERR_FAIL(); //bug? } @@ -411,17 +561,7 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color _update_render_base_uniform_set(); //may have changed due to the above (light buffer enlarged, as an example) - _fill_render_list(RENDER_LIST_OPAQUE, p_render_data, PASS_MODE_COLOR); - render_list[RENDER_LIST_OPAQUE].sort_by_key(); - render_list[RENDER_LIST_ALPHA].sort_by_reverse_depth_and_priority(); - - // we no longer use this... - _fill_instance_data(RENDER_LIST_OPAQUE); - _fill_instance_data(RENDER_LIST_ALPHA); - - RD::get_singleton()->draw_command_end_label(); - - // note, no depth prepass here! + RD::get_singleton()->draw_command_end_label(); // Render Setup // setup environment RID radiance_texture; @@ -496,97 +636,199 @@ void RenderForwardMobile::_render_scene(RenderDataRD *p_render_data, const Color // do not try to draw sky if invalid draw_sky = false; } - RD::get_singleton()->draw_command_end_label(); + RD::get_singleton()->draw_command_end_label(); // Setup Sky } } else { clear_color = p_default_bg_color; } - // opaque pass + // update sky buffers (if required) + if (draw_sky || draw_sky_fog_only) { + // !BAS! @TODO See if we can limit doing some things double and maybe even move this into _pre_opaque_render + // and change Forward Clustered in the same way as we have here (but without using subpasses) + RENDER_TIMESTAMP("Setup Sky resolution buffers"); + + RD::get_singleton()->draw_command_begin_label("Setup Sky resolution buffers"); + + if (p_render_data->reflection_probe.is_valid()) { + CameraMatrix correction; + correction.set_depth_correction(true); + CameraMatrix projection = correction * p_render_data->cam_projection; + sky.update_res_buffers(env, 1, &projection, p_render_data->cam_transform, time); + } else { + sky.update_res_buffers(env, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time); + } + + RD::get_singleton()->draw_command_end_label(); // Setup Sky resolution buffers + } - // !BAS! Look into this, seems most of the code in here related to clustered only, may want to move this code into ForwardClustered/RenderForwardMobile before calling it from here - // does trigger shadow map rendering so kinda important _pre_opaque_render(p_render_data, false, false, RID(), RID()); - RD::get_singleton()->draw_command_begin_label("Render Opaque Pass"); + { + if (render_buffer) { + RD::get_singleton()->draw_command_begin_label("Render 3D Pass"); + } else { + RD::get_singleton()->draw_command_begin_label("Render Reflection Probe Pass"); + } - scene_state.ubo.directional_light_count = p_render_data->directional_light_count; + // opaque pass - _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, p_render_data->render_buffers.is_valid()); + RD::get_singleton()->draw_command_begin_label("Render Opaque Subpass"); - RENDER_TIMESTAMP("Render Opaque Pass"); + scene_state.ubo.directional_light_count = p_render_data->directional_light_count; - RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, p_render_data, radiance_texture, true); + _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, p_render_data->render_buffers.is_valid()); - 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; + RENDER_TIMESTAMP("Render Opaque Subpass"); - { - bool will_continue_color = (can_continue_color || draw_sky || draw_sky_fog_only); - bool will_continue_depth = (can_continue_depth || draw_sky || draw_sky_fog_only); + RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_OPAQUE, p_render_data, radiance_texture, true); - // regular forward for now - Vector<Color> c; - c.push_back(clear_color.to_linear()); + 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; - 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(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); - _render_list_with_threads(&render_list_params, opaque_framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); - } + { + bool will_continue_color = (can_continue_color || draw_sky || draw_sky_fog_only); + bool will_continue_depth = (can_continue_depth || draw_sky || draw_sky_fog_only); + + // regular forward for now + Vector<Color> c; + c.push_back(clear_color.to_linear()); // our render buffer + if (render_buffer) { + if (render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { + c.push_back(clear_color.to_linear()); // our resolve buffer + } + if (using_subpass_post_process) { + c.push_back(Color()); // our 2D buffer we're copying into + } + } - RD::get_singleton()->draw_command_end_label(); + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + 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(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_begin_split(framebuffer, thread_draw_lists.size(), thread_draw_lists.ptr(), keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(framebuffer, keep_color ? RD::INITIAL_ACTION_KEEP : RD::INITIAL_ACTION_CLEAR, will_continue_color ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, RD::INITIAL_ACTION_CLEAR, will_continue_depth ? RD::FINAL_ACTION_CONTINUE : RD::FINAL_ACTION_READ, c, 1.0, 0); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + } + } - if (draw_sky || draw_sky_fog_only) { - RENDER_TIMESTAMP("Render Sky"); + RD::get_singleton()->draw_command_end_label(); //Render Opaque Subpass - RD::get_singleton()->draw_command_begin_label("Draw Sky"); + if (draw_sky || draw_sky_fog_only) { + RENDER_TIMESTAMP("Render Sky Subpass"); - if (p_render_data->reflection_probe.is_valid()) { - CameraMatrix correction; - correction.set_depth_correction(true); - CameraMatrix projection = correction * p_render_data->cam_projection; - sky.draw(env, can_continue_color, can_continue_depth, opaque_framebuffer, 1, &projection, p_render_data->cam_transform, time); + RD::get_singleton()->draw_command_begin_label("Draw Sky Subpass"); + + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + + if (p_render_data->reflection_probe.is_valid()) { + CameraMatrix correction; + correction.set_depth_correction(true); + CameraMatrix projection = correction * p_render_data->cam_projection; + sky.draw(draw_list, env, framebuffer, 1, &projection, p_render_data->cam_transform, time); + } else { + sky.draw(draw_list, env, framebuffer, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time); + } + + RD::get_singleton()->draw_command_end_label(); // Draw Sky Subpass + + // note, if MSAA is used in 2-subpass approach we should get an automatic resolve here } else { - sky.draw(env, can_continue_color, can_continue_depth, opaque_framebuffer, p_render_data->view_count, p_render_data->view_projection, p_render_data->cam_transform, time); + // switch to subpass but we do nothing here so basically we skip (though this should trigger resolve with 2-subpass MSAA). + RD::get_singleton()->draw_list_switch_to_next_pass(); } - 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); - /* - if (using_separate_specular) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->specular_msaa, render_buffer->specular); + if (!using_subpass_transparent) { + // We're done with our subpasses so end our container pass + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + + RD::get_singleton()->draw_command_end_label(); // Render 3D Pass / Render Reflection Probe Pass } - */ - } - 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); - } + // transparent pass + RENDER_TIMESTAMP("Render Transparent Subpass"); - // transparent pass - RENDER_TIMESTAMP("Render Transparent Pass"); + RD::get_singleton()->draw_command_begin_label("Render Transparent Subpass"); - RD::get_singleton()->draw_command_begin_label("Render Transparent Pass"); + rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_ALPHA, p_render_data, radiance_texture, true); - rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_ALPHA, p_render_data, radiance_texture, true); + if (using_subpass_transparent) { + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + 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(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_switch_to_next_pass_split(thread_draw_lists.size(), thread_draw_lists.ptr()); + render_list_params.subpass = RD::get_singleton()->draw_list_get_current_pass(); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + render_list_params.subpass = RD::get_singleton()->draw_list_get_current_pass(); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + } - _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, false); + RD::get_singleton()->draw_command_end_label(); // Render Transparent Subpass - { - 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(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); - _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); + // note if we are using MSAA we should get an automatic resolve through our subpass configuration. + + // blit to tonemap + if (render_buffer && using_subpass_post_process) { + _post_process_subpass(render_buffer->color, framebuffer, p_render_data); + } + + RD::get_singleton()->draw_command_end_label(); // Render 3D Pass / Render Reflection Probe Pass + + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } else { + framebuffer = render_buffer->color_fbs[FB_CONFIG_ONE_PASS]; + + // this may be needed if we re-introduced steps that change info, not sure which do so in the previous implementation + // _setup_environment(p_render_data, p_render_data->reflection_probe.is_valid(), screen_size, !p_render_data->reflection_probe.is_valid(), p_default_bg_color, false); + + RD::FramebufferFormatID fb_format = RD::get_singleton()->framebuffer_get_format(framebuffer); + 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(), reverse_cull, PASS_MODE_COLOR, rp_uniform_set, get_debug_draw_mode() == RS::VIEWPORT_DEBUG_DRAW_WIREFRAME, Vector2(), p_render_data->lod_camera_plane, p_render_data->lod_distance_multiplier, p_render_data->screen_lod_threshold, p_render_data->view_count); + render_list_params.framebuffer_format = fb_format; + if ((uint32_t)render_list_params.element_count > render_list_thread_threshold && false) { + // secondary command buffers need more testing at this time + //multi threaded + thread_draw_lists.resize(RendererThreadPool::singleton->thread_work_pool.get_thread_count()); + RD::get_singleton()->draw_list_begin_split(framebuffer, thread_draw_lists.size(), thread_draw_lists.ptr(), 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); + RendererThreadPool::singleton->thread_work_pool.do_work(thread_draw_lists.size(), this, &RenderForwardMobile::_render_list_thread_function, &render_list_params); + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } else { + //single threaded + RD::DrawListID draw_list = RD::get_singleton()->draw_list_begin(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); + _render_list(draw_list, fb_format, &render_list_params, 0, render_list_params.element_count); + RD::get_singleton()->draw_list_end(RD::BARRIER_MASK_ALL); + } + + RD::get_singleton()->draw_command_end_label(); // Render Transparent Subpass + } } - RD::get_singleton()->draw_command_end_label(); + if (render_buffer && !using_subpass_post_process) { + RD::get_singleton()->draw_command_begin_label("Post process pass"); - RD::get_singleton()->draw_command_begin_label("Resolve"); + // If we need extra effects we do this in its own pass + RENDER_TIMESTAMP("Tonemap"); - if (render_buffer && render_buffer->msaa != RS::VIEWPORT_MSAA_DISABLED) { - RD::get_singleton()->texture_resolve_multisample(render_buffer->color_msaa, render_buffer->color); + _render_buffers_post_process_and_tonemap(p_render_data); + + RD::get_singleton()->draw_command_end_label(); // Post process pass } - RD::get_singleton()->draw_command_end_label(); + if (render_buffer) { + _disable_clear_request(p_render_data); + } } /* these are being called from RendererSceneRenderRD::_pre_opaque_render */ @@ -635,7 +877,7 @@ void RenderForwardMobile::_render_shadow_append(RID p_framebuffer, const PagedAr _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode, 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); + _fill_element_info(RENDER_LIST_SECONDARY, render_list_from, render_list_size); { //regular forward for now @@ -713,7 +955,7 @@ void RenderForwardMobile::_render_material(const Transform3D &p_cam_transform, c PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); @@ -754,7 +996,7 @@ void RenderForwardMobile::_render_uv2(const PagedArray<GeometryInstance *> &p_in PassMode pass_mode = PASS_MODE_DEPTH_MATERIAL; _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); @@ -827,7 +1069,7 @@ void RenderForwardMobile::_render_particle_collider_heightfield(RID p_fb, const _fill_render_list(RENDER_LIST_SECONDARY, &render_data, pass_mode); render_list[RENDER_LIST_SECONDARY].sort_by_key(); - _fill_instance_data(RENDER_LIST_SECONDARY); + _fill_element_info(RENDER_LIST_SECONDARY); RID rp_uniform_set = _setup_render_pass_uniform_set(RENDER_LIST_SECONDARY, nullptr, RID()); @@ -1413,9 +1655,7 @@ void RenderForwardMobile::_setup_environment(const RenderDataRD *p_render_data, RD::get_singleton()->buffer_update(scene_state.uniform_buffers[p_index], 0, sizeof(SceneState::UBO), &scene_state.ubo, RD::BARRIER_MASK_RASTER); } -void RenderForwardMobile::_fill_instance_data(RenderListType p_render_list, uint32_t p_offset, int32_t p_max_elements, bool p_update_buffer) { - // !BAS! Rename this to make clear this is not the same as with the forward renderer and remove p_update_buffer? - +void RenderForwardMobile::_fill_element_info(RenderListType p_render_list, uint32_t p_offset, int32_t p_max_elements) { RenderList *rl = &render_list[p_render_list]; uint32_t element_total = p_max_elements >= 0 ? uint32_t(p_max_elements) : rl->elements.size(); @@ -1669,7 +1909,7 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr prev_index_array_rd = index_array_rd; } - RID pipeline_rd = pipeline->get_render_pipeline(vertex_format, framebuffer_format, p_params->force_wireframe); + RID pipeline_rd = pipeline->get_render_pipeline(vertex_format, framebuffer_format, p_params->force_wireframe, p_params->subpass); if (pipeline_rd != prev_pipeline_rd) { // checking with prev shader does not make so much sense, as @@ -1684,8 +1924,8 @@ void RenderForwardMobile::_render_list_template(RenderingDevice::DrawListID p_dr } if (material_uniform_set != prev_material_uniform_set) { - //update uniform set - if (material_uniform_set.is_valid()) { + // Update uniform set. + if (RD::get_singleton()->uniform_set_is_valid(material_uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_uniform_set, MATERIAL_UNIFORM_SET); } @@ -2325,6 +2565,8 @@ RenderForwardMobile::RenderForwardMobile(RendererStorageRD *p_storage) : RendererSceneRenderRD(p_storage) { singleton = this; + sky.set_texture_format(_render_buffers_get_color_format()); + String defines; defines += "\n#define MAX_ROUGHNESS_LOD " + itos(get_roughness_layers() - 1) + ".0\n"; diff --git a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h index 973925d562..b981592840 100644 --- a/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h +++ b/servers/rendering/renderer_rd/forward_mobile/render_forward_mobile.h @@ -92,6 +92,18 @@ protected: /* Render Buffer */ + // We can have: + // - 4 subpasses combining the full render cycle + // - 3 subpasses + 1 normal pass for tonemapping/glow/dof/etc (using fb for 2D buffer) + // - 2 subpasses + 1 normal pass for transparent + 1 normal pass for tonemapping/glow/dof/etc (using fb for 2D buffer) + enum RenderBufferMobileFramebufferConfigType { + FB_CONFIG_ONE_PASS, // Single pass frame buffer for alpha pass + FB_CONFIG_TWO_SUBPASSES, // Opaque + Sky sub pass + FB_CONFIG_THREE_SUBPASSES, // Opaque + Sky + Alpha sub pass + FB_CONFIG_FOUR_SUBPASSES, // Opaque + Sky + Alpha sub pass + Tonemap pass + FB_CONFIG_MAX + }; + struct RenderBufferDataForwardMobile : public RenderBufferData { RID color; RID depth; @@ -104,12 +116,12 @@ protected: RID depth_msaa; // RID normal_roughness_buffer_msaa; - RID color_fb; + RID color_fbs[FB_CONFIG_MAX]; int width, height; uint32_t view_count; void clear(); - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count); ~RenderBufferDataForwardMobile(); }; @@ -152,6 +164,7 @@ protected: RD::FramebufferFormatID framebuffer_format = 0; uint32_t element_offset = 0; uint32_t barrier = RD::BARRIER_MASK_ALL; + uint32_t subpass = 0; RenderListParameters(GeometryInstanceSurfaceDataCache **p_elements, RenderElementInfo *p_element_info, int p_element_count, bool p_reverse_cull, PassMode p_pass_mode, 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_view_count = 1, uint32_t p_element_offset = 0, uint32_t p_barrier = RD::BARRIER_MASK_ALL) { elements = p_elements; @@ -195,7 +208,7 @@ protected: virtual RID _render_buffers_get_normal_texture(RID p_render_buffers) override; void _fill_render_list(RenderListType p_render_list, const RenderDataRD *p_render_data, PassMode p_pass_mode, bool p_append = false); - 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_element_info(RenderListType p_render_list, uint32_t p_offset = 0, int32_t p_max_elements = -1); // void _update_instance_data_buffer(RenderListType p_render_list); static RenderForwardMobile *singleton; @@ -576,6 +589,8 @@ protected: void _update_shader_quality_settings() override; public: + virtual RID reflection_probe_create_framebuffer(RID p_color, RID p_depth) override; + 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); diff --git a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp index 18c1fe02a0..14b2768c29 100644 --- a/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_canvas_render_rd.cpp @@ -480,6 +480,10 @@ void RendererCanvasRenderRD::_render_item(RD::DrawListID p_draw_list, RID p_rend case Item::Command::TYPE_RECT: { const Item::CommandRect *rect = static_cast<const Item::CommandRect *>(c); + if (rect->flags & CANVAS_RECT_TILE) { + current_repeat = RenderingServer::CanvasItemTextureRepeat::CANVAS_ITEM_TEXTURE_REPEAT_ENABLED; + } + //bind pipeline { RID pipeline = pipeline_variants->variants[light_mode][PIPELINE_VARIANT_QUAD].get_render_pipeline(RD::INVALID_ID, p_framebuffer_format); @@ -1089,7 +1093,8 @@ void RendererCanvasRenderRD::_render_items(RID p_to_render_target, int p_item_co if (material_data) { if (material_data->shader_data->version.is_valid() && material_data->shader_data->valid) { pipeline_variants = &material_data->shader_data->pipeline_variants; - if (material_data->uniform_set.is_valid()) { + // Update uniform set. + if (RD::get_singleton()->uniform_set_is_valid(material_data->uniform_set)) { // Material may not have a uniform set. RD::get_singleton()->draw_list_bind_uniform_set(draw_list, material_data->uniform_set, MATERIAL_UNIFORM_SET); } } else { diff --git a/servers/rendering/renderer_rd/renderer_compositor_rd.h b/servers/rendering/renderer_rd/renderer_compositor_rd.h index 15b3b77ed9..8639362da9 100644 --- a/servers/rendering/renderer_rd/renderer_compositor_rd.h +++ b/servers/rendering/renderer_rd/renderer_compositor_rd.h @@ -80,7 +80,7 @@ protected: Map<RID, RID> render_target_descriptors; double time; - float delta; + double delta; static uint64_t frame; @@ -100,7 +100,7 @@ public: void finalize(); _ALWAYS_INLINE_ uint64_t get_frame_number() const { return frame; } - _ALWAYS_INLINE_ float get_frame_delta_time() const { return delta; } + _ALWAYS_INLINE_ double get_frame_delta_time() const { return delta; } _ALWAYS_INLINE_ double get_total_time() const { return time; } static Error is_viable() { diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp index 2e0e7c3c43..5b734fc603 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.cpp @@ -629,6 +629,8 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); ERR_FAIL_COND_V(!rpi, false); + RD::get_singleton()->draw_command_begin_label("Reflection probe render"); + if (storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS && atlas->reflection.is_valid() && atlas->size != 256) { WARN_PRINT("ReflectionProbes set to UPDATE_ALWAYS must have an atlas size of 256. Please update the atlas size in the ProjectSettings."); reflection_atlas_set_size(p_reflection_atlas, 256, atlas->count); @@ -675,12 +677,9 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc } atlas->reflections.resize(atlas->count); for (int i = 0; i < atlas->count; i++) { - 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); + atlas->reflections.write[i].data.update_reflection_data(storage, atlas->size, mipmaps, false, atlas->reflection, i * 6, storage->reflection_probe_get_update_mode(rpi->probe) == RS::REFLECTION_PROBE_UPDATE_ALWAYS, sky.roughness_layers, _render_buffers_get_color_format()); for (int j = 0; j < 6; j++) { - Vector<RID> fb; - fb.push_back(atlas->reflections.write[i].data.layers[0].mipmaps[0].views[j]); - fb.push_back(atlas->depth_buffer); - atlas->reflections.write[i].fbs[j] = RD::get_singleton()->framebuffer_create(fb); + atlas->reflections.write[i].fbs[j] = reflection_probe_create_framebuffer(atlas->reflections.write[i].data.layers[0].mipmaps[0].views[j], atlas->depth_buffer); } } @@ -721,9 +720,18 @@ bool RendererSceneRenderRD::reflection_probe_instance_begin_render(RID p_instanc rpi->processing_layer = 1; rpi->processing_side = 0; + RD::get_singleton()->draw_command_end_label(); + return true; } +RID RendererSceneRenderRD::reflection_probe_create_framebuffer(RID p_color, RID p_depth) { + Vector<RID> fb; + fb.push_back(p_color); + fb.push_back(p_depth); + return RD::get_singleton()->framebuffer_create(fb); +} + bool RendererSceneRenderRD::reflection_probe_instance_postprocess_step(RID p_instance) { ReflectionProbeInstance *rpi = reflection_probe_instance_owner.getornull(p_instance); ERR_FAIL_COND_V(!rpi, false); @@ -1992,6 +2000,75 @@ void RendererSceneRenderRD::_render_buffers_post_process_and_tonemap(const Rende storage->render_target_disable_clear_request(rb->render_target); } +void RendererSceneRenderRD::_post_process_subpass(RID p_source_texture, RID p_framebuffer, const RenderDataRD *p_render_data) { + RD::get_singleton()->draw_command_begin_label("Post Process Subpass"); + + RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + + RendererSceneEnvironmentRD *env = environment_owner.getornull(p_render_data->environment); + + bool can_use_effects = rb->width >= 8 && rb->height >= 8; + + RENDER_TIMESTAMP("Tonemap"); + RD::DrawListID draw_list = RD::get_singleton()->draw_list_switch_to_next_pass(); + + EffectsRD::TonemapSettings tonemap; + + if (env) { + tonemap.tonemap_mode = env->tone_mapper; + tonemap.exposure = env->exposure; + tonemap.white = env->white; + } + + // We don't support glow or auto exposure here, if they are needed, don't use subpasses! + // The problem is that we need to use the result so far and process them before we can + // apply this to our results. + if (can_use_effects && env && env->glow_enabled) { + ERR_FAIL_MSG("Glow is not supported when using subpasses."); + } + if (can_use_effects && env && env->auto_exposure) { + ERR_FAIL_MSG("Glow is not supported when using subpasses."); + } + + tonemap.use_glow = false; + tonemap.glow_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_BLACK); + tonemap.use_auto_exposure = false; + tonemap.exposure_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_WHITE); + + tonemap.use_color_correction = false; + tonemap.use_1d_color_correction = false; + tonemap.color_correction_texture = storage->texture_rd_get_default(RendererStorageRD::DEFAULT_RD_TEXTURE_3D_WHITE); + + if (can_use_effects && env) { + tonemap.use_bcs = env->adjustments_enabled; + tonemap.brightness = env->adjustments_brightness; + tonemap.contrast = env->adjustments_contrast; + tonemap.saturation = env->adjustments_saturation; + if (env->adjustments_enabled && env->color_correction.is_valid()) { + tonemap.use_color_correction = true; + tonemap.use_1d_color_correction = env->use_1d_color_correction; + tonemap.color_correction_texture = storage->texture_get_rd_texture(env->color_correction); + } + } + + tonemap.use_debanding = rb->use_debanding; + tonemap.texture_size = Vector2i(rb->width, rb->height); + + tonemap.view_count = p_render_data->view_count; + + storage->get_effects()->tonemapper(draw_list, p_source_texture, RD::get_singleton()->framebuffer_get_format(p_framebuffer), tonemap); + + RD::get_singleton()->draw_command_end_label(); +} + +void RendererSceneRenderRD::_disable_clear_request(const RenderDataRD *p_render_data) { + RenderBuffers *rb = render_buffers_owner.getornull(p_render_data->render_buffers); + ERR_FAIL_COND(!rb); + + storage->render_target_disable_clear_request(rb->render_target); +} + void RendererSceneRenderRD::_render_buffers_debug_draw(RID p_render_buffers, RID p_shadow_atlas, RID p_occlusion_buffer) { EffectsRD *effects = storage->get_effects(); @@ -2279,12 +2356,11 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p tf.width = rb->width; tf.height = rb->height; tf.array_layers = rb->view_count; // create a layer for every view - tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0); + tf.usage_bits = RD::TEXTURE_USAGE_SAMPLING_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0) | RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; if (rb->msaa != RS::VIEWPORT_MSAA_DISABLED) { - tf.usage_bits |= RD::TEXTURE_USAGE_CAN_COPY_TO_BIT | (_render_buffers_can_be_storage() ? RD::TEXTURE_USAGE_STORAGE_BIT : 0); - } else { - tf.usage_bits |= RD::TEXTURE_USAGE_COLOR_ATTACHMENT_BIT; + tf.usage_bits |= RD::TEXTURE_USAGE_CAN_COPY_TO_BIT; } + tf.usage_bits |= RD::TEXTURE_USAGE_INPUT_ATTACHMENT_BIT; // only needed when using subpasses in the mobile renderer rb->texture = RD::get_singleton()->texture_create(tf, RD::TextureView()); } @@ -2322,7 +2398,8 @@ void RendererSceneRenderRD::render_buffers_configure(RID p_render_buffers, RID p rb->texture_fb = RD::get_singleton()->framebuffer_create(fb, RenderingDevice::INVALID_ID, rb->view_count); } - rb->data->configure(rb->texture, rb->depth_texture, p_width, p_height, p_msaa, p_view_count); + RID target_texture = storage->render_target_get_rd_texture(rb->render_target); + rb->data->configure(rb->texture, rb->depth_texture, target_texture, p_width, p_height, p_msaa, p_view_count); 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); @@ -3842,9 +3919,28 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData _render_scene(&render_data, clear_color); 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) { + /* + _debug_draw_cluster(p_render_buffers); + + RENDER_TIMESTAMP("Tonemap"); + + _render_buffers_post_process_and_tonemap(&render_data); + */ + + _render_buffers_debug_draw(p_render_buffers, p_shadow_atlas, p_occluder_debug_tex); + if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI && rb != nullptr && rb->sdfgi != nullptr) { + rb->sdfgi->debug_draw(render_data.cam_projection, render_data.cam_transform, rb->width, rb->height, rb->render_target, rb->texture); + } + } +} + +void RendererSceneRenderRD::_debug_draw_cluster(RID p_render_buffers) { + if (p_render_buffers.is_valid() && current_cluster_builder != nullptr) { + RS::ViewportDebugDraw dd = get_debug_draw_mode(); + + if (dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_SPOT_LIGHTS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_DECALS || dd == RS::VIEWPORT_DEBUG_DRAW_CLUSTER_REFLECTION_PROBES) { ClusterBuilderRD::ElementType elem_type = ClusterBuilderRD::ELEMENT_TYPE_MAX; - switch (debug_draw) { + switch (dd) { case RS::VIEWPORT_DEBUG_DRAW_CLUSTER_OMNI_LIGHTS: elem_type = ClusterBuilderRD::ELEMENT_TYPE_OMNI_LIGHT; break; @@ -3860,17 +3956,7 @@ void RendererSceneRenderRD::render_scene(RID p_render_buffers, const CameraData default: { } } - if (current_cluster_builder != nullptr) { - current_cluster_builder->debug(elem_type); - } - } - - RENDER_TIMESTAMP("Tonemap"); - - _render_buffers_post_process_and_tonemap(&render_data); - _render_buffers_debug_draw(p_render_buffers, p_shadow_atlas, p_occluder_debug_tex); - if (debug_draw == RS::VIEWPORT_DEBUG_DRAW_SDFGI && rb != nullptr && rb->sdfgi != nullptr) { - rb->sdfgi->debug_draw(render_data.cam_projection, render_data.cam_transform, rb->width, rb->height, rb->render_target, rb->texture); + current_cluster_builder->debug(elem_type); } } } diff --git a/servers/rendering/renderer_rd/renderer_scene_render_rd.h b/servers/rendering/renderer_rd/renderer_scene_render_rd.h index bb06eb608f..336dd6e1a7 100644 --- a/servers/rendering/renderer_rd/renderer_scene_render_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_render_rd.h @@ -95,7 +95,7 @@ protected: double time_step = 0; struct RenderBufferData { - virtual void configure(RID p_color_buffer, RID p_depth_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) = 0; + virtual void configure(RID p_color_buffer, RID p_depth_buffer, RID p_target_buffer, int p_width, int p_height, RS::ViewportMSAA p_msaa, uint32_t p_view_count) = 0; virtual ~RenderBufferData() {} }; virtual RenderBufferData *_create_render_buffer_data() = 0; @@ -117,6 +117,7 @@ protected: virtual void _render_particle_collider_heightfield(RID p_fb, const Transform3D &p_cam_transform, const CameraMatrix &p_cam_projection, const PagedArray<GeometryInstance *> &p_instances) = 0; void _debug_sdfgi_probes(RID p_render_buffers, RD::DrawListID p_draw_list, RID p_framebuffer, const CameraMatrix &p_camera_with_transform); + void _debug_draw_cluster(RID p_render_buffers); RenderBufferData *render_buffers_get_data(RID p_render_buffers); @@ -133,6 +134,10 @@ protected: void _pre_opaque_render(RenderDataRD *p_render_data, bool p_use_ssao, bool p_use_gi, RID p_normal_roughness_buffer, RID p_voxel_gi_buffer); + void _render_buffers_post_process_and_tonemap(const RenderDataRD *p_render_data); + void _post_process_subpass(RID p_source_texture, RID p_framebuffer, const RenderDataRD *p_render_data); + void _disable_clear_request(const RenderDataRD *p_render_data); + // needed for a single argument calls (material and uv2) PagedArrayPool<GeometryInstance *> cull_argument_pool; PagedArray<GeometryInstance *> cull_argument; //need this to exist @@ -146,7 +151,7 @@ protected: } else { return nullptr; } - } + }; //used for mobile renderer mostly @@ -524,7 +529,6 @@ private: void _allocate_luminance_textures(RenderBuffers *rb); void _render_buffers_debug_draw(RID p_render_buffers, RID p_shadow_atlas, RID p_occlusion_buffer); - void _render_buffers_post_process_and_tonemap(const RenderDataRD *p_render_data); /* Cluster */ @@ -923,6 +927,12 @@ public: virtual void camera_effects_set_dof_blur(RID p_camera_effects, bool p_far_enable, float p_far_distance, float p_far_transition, bool p_near_enable, float p_near_distance, float p_near_transition, float p_amount) override; virtual void camera_effects_set_custom_exposure(RID p_camera_effects, bool p_enable, float p_exposure) override; + bool camera_effects_uses_dof(RID p_camera_effects) { + CameraEffects *camfx = camera_effects_owner.getornull(p_camera_effects); + + return camfx && (camfx->dof_blur_near_enabled || camfx->dof_blur_far_enabled) && camfx->dof_blur_amount > 0.0; + } + virtual RID light_instance_create(RID p_light) override; virtual void light_instance_set_transform(RID p_light_instance, const Transform3D &p_transform) override; virtual void light_instance_set_aabb(RID p_light_instance, const AABB &p_aabb) override; @@ -1065,6 +1075,7 @@ public: virtual bool reflection_probe_instance_needs_redraw(RID p_instance) override; virtual bool reflection_probe_instance_has_reflection(RID p_instance) override; virtual bool reflection_probe_instance_begin_render(RID p_instance, RID p_reflection_atlas) override; + virtual RID reflection_probe_create_framebuffer(RID p_color, RID p_depth); virtual bool reflection_probe_instance_postprocess_step(RID p_instance) override; uint32_t reflection_probe_instance_get_resolution(RID p_instance); diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp index 5b4d78c15b..9331cab090 100644 --- a/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.cpp @@ -283,14 +283,17 @@ void RendererSceneSkyRD::_render_sky(RD::DrawListID p_list, float p_time, RID p_ RD::DrawListID draw_list = p_list; - RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, p_pipeline->get_render_pipeline(RD::INVALID_ID, fb_format)); + RD::get_singleton()->draw_list_bind_render_pipeline(draw_list, p_pipeline->get_render_pipeline(RD::INVALID_ID, fb_format, false, RD::get_singleton()->draw_list_get_current_pass())); - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.uniform_set, 0); - if (p_uniform_set.is_valid()) { //material may not have uniform set - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_uniform_set, 1); + // Update uniform sets. + { + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.uniform_set, 0); + if (RD::get_singleton()->uniform_set_is_valid(p_uniform_set)) { // Material may not have a uniform set. + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_uniform_set, 1); + } + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_texture_set, 2); + RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.fog_uniform_set, 3); } - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, p_texture_set, 2); - RD::get_singleton()->draw_list_bind_uniform_set(draw_list, sky_scene_state.fog_uniform_set, 3); RD::get_singleton()->draw_list_bind_index_array(draw_list, index_array); @@ -313,12 +316,16 @@ void RendererSceneSkyRD::ReflectionData::clear_reflection_data() { 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) { +void RendererSceneSkyRD::ReflectionData::update_reflection_data(RendererStorageRD *p_storage, 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, RD::DataFormat p_texture_format) { //recreate radiance and all data int mipmaps = p_mipmaps; uint32_t w = p_size, h = p_size; + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + if (p_use_array) { int num_layers = p_low_quality ? 8 : p_roughness_layers; @@ -377,9 +384,9 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int } 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::get_singleton()->set_resource_name(radiance_base_cubemap, "radiance base cubemap"); RD::TextureFormat tf; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = p_texture_format; tf.width = 64; // Always 64x64 tf.height = 64; tf.texture_type = RD::TEXTURE_TYPE_CUBE; @@ -388,6 +395,7 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int 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()); + RD::get_singleton()->set_resource_name(downsampled_radiance_cubemap, "downsampled radiance cubemap"); { uint32_t mmw = 64; uint32_t mmh = 64; @@ -397,6 +405,18 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int 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); + RD::get_singleton()->set_resource_name(mm.view, "Downsampled Radiance Cubemap Mip " + itos(j) + " "); + if (prefer_raster_effects) { + // we need a framebuffer for each side of our cubemap + + for (int k = 0; k < 6; k++) { + mm.views[k] = RD::get_singleton()->texture_create_shared_from_slice(RD::TextureView(), downsampled_radiance_cubemap, k, j); + RD::get_singleton()->set_resource_name(mm.view, "Downsampled Radiance Cubemap Mip: " + itos(j) + " Face: " + itos(k) + " "); + Vector<RID> fbtex; + fbtex.push_back(mm.views[k]); + mm.framebuffers[k] = RD::get_singleton()->framebuffer_create(fbtex); + } + } mmw = MAX(1, mmw >> 1); mmh = MAX(1, mmh >> 1); @@ -405,50 +425,128 @@ void RendererSceneSkyRD::ReflectionData::update_reflection_data(int p_size, int } 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); + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + if (prefer_raster_effects) { + RD::get_singleton()->draw_command_begin_label("Downsample radiance map"); + for (int k = 0; k < 6; k++) { + effects->cubemap_downsample_raster(radiance_base_cubemap, downsampled_layer.mipmaps[0].framebuffers[k], k, 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); - } + for (int i = 1; i < downsampled_layer.mipmaps.size(); i++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_downsample_raster(downsampled_layer.mipmaps[i - 1].view, downsampled_layer.mipmaps[i].framebuffers[k], k, downsampled_layer.mipmaps[i].size); + } + } + RD::get_singleton()->draw_command_end_label(); // Downsample Radiance - Vector<RID> views; - if (p_use_arrays) { - for (int i = 1; i < layers.size(); i++) { - views.push_back(layers[i].views[0]); + if (p_use_arrays) { + RD::get_singleton()->draw_command_begin_label("filter radiance map into array heads"); + for (int i = 0; i < layers.size(); i++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_filter_raster(downsampled_radiance_cubemap, layers[i].mipmaps[0].framebuffers[k], k, i); + } + } + } else { + RD::get_singleton()->draw_command_begin_label("filter radiance map into mipmaps directly"); + for (int j = 0; j < layers[0].mipmaps.size(); j++) { + for (int k = 0; k < 6; k++) { + effects->cubemap_filter_raster(downsampled_radiance_cubemap, layers[0].mipmaps[j].framebuffers[k], k, j); + } + } } + RD::get_singleton()->draw_command_end_label(); // Filter radiance } else { - for (int i = 1; i < layers[0].views.size(); i++) { - views.push_back(layers[0].views[i]); + 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++) { + 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); + 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); + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + if (prefer_raster_effects) { + // Need to ask clayjohn but p_cube_side is set to 10, looks like in the compute shader we're doing all 6 sides in one call + // here we need to do them one by one so ignoring p_cube_side + if (p_use_arrays) { + for (int k = 0; k < 6; k++) { + effects->cubemap_roughness_raster( + radiance_base_cubemap, + layers[p_base_layer].mipmaps[0].framebuffers[k], + k, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers.size() - 1.0), + layers[p_base_layer].mipmaps[0].size.x); + } + } else { + for (int k = 0; k < 6; k++) { + effects->cubemap_roughness_raster( + layers[0].views[p_base_layer - 1], + layers[0].mipmaps[p_base_layer].framebuffers[k], + k, + p_sky_ggx_samples_quality, + float(p_base_layer) / (layers[0].mipmaps.size() - 1.0), + layers[0].mipmaps[p_base_layer].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); + if (p_use_arrays) { + //render directly to the layers + 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 { + 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) { + EffectsRD *effects = p_storage->get_effects(); + ERR_FAIL_NULL_MSG(effects, "Effects haven't been initialised"); + bool prefer_raster_effects = effects->get_prefer_raster_effects(); + + RD::get_singleton()->draw_command_begin_label("Update Radiance Cubemap Array Mipmaps"); 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); + if (prefer_raster_effects) { + for (int k = 0; k < 6; k++) { + RID framebuffer = layers[i].mipmaps[j + 1].framebuffers[k]; + effects->cubemap_downsample_raster(view, framebuffer, k, size); + } + } else { + RID texture = layers[i].views[j + 1]; + effects->cubemap_downsample(view, texture, size); + } } } + RD::get_singleton()->draw_command_end_label(); } //////////////////////////////////////////////////////////////////////////////// @@ -902,6 +1000,10 @@ void sky() { } } +void RendererSceneSkyRD::set_texture_format(RD::DataFormat p_texture_format) { + texture_format = p_texture_format; +} + RendererSceneSkyRD::~RendererSceneSkyRD() { // TODO cleanup anything created in init... @@ -1170,6 +1272,7 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM cm = correction * cm; if (shader_data->uses_quarter_res) { + RD::get_singleton()->draw_command_begin_label("Render Sky to Quarter Res Cubemap"); PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_QUARTER_RES]; Vector<Color> clear_colors; @@ -1184,9 +1287,11 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[2].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); } if (shader_data->uses_half_res) { + RD::get_singleton()->draw_command_begin_label("Render Sky to Half Res Cubemap"); PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP_HALF_RES]; Vector<Color> clear_colors; @@ -1201,11 +1306,13 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[1].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); } RD::DrawListID cubemap_draw_list; PipelineCacheRD *pipeline = &shader_data->pipelines[SKY_VERSION_CUBEMAP]; + RD::get_singleton()->draw_command_begin_label("Render Sky Cubemap"); for (int i = 0; i < 6; i++) { Basis local_view = Basis::looking_at(view_normals[i], view_up[i]); RID texture_uniform_set = sky->get_textures(storage, SKY_TEXTURE_SET_CUBEMAP, sky_shader.default_shader_rd); @@ -1214,6 +1321,7 @@ void RendererSceneSkyRD::update(RendererSceneEnvironmentRD *p_env, const CameraM _render_sky(cubemap_draw_list, p_time, sky->reflection.layers[0].mipmaps[0].framebuffers[i], pipeline, material->uniform_set, texture_uniform_set, 1, &cm, local_view, multiplier, p_transform.origin); RD::get_singleton()->draw_list_end(); } + RD::get_singleton()->draw_command_end_label(); if (sky_mode == RS::SKY_MODE_REALTIME) { sky->reflection.create_reflection_fast_filter(storage, sky_use_cubemap_array); @@ -1359,6 +1467,179 @@ void RendererSceneSkyRD::draw(RendererSceneEnvironmentRD *p_env, bool p_can_cont RD::get_singleton()->draw_list_end(); } +void RendererSceneSkyRD::update_res_buffers(RendererSceneEnvironmentRD *p_env, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time) { + ERR_FAIL_COND(!p_env); + + ERR_FAIL_COND(p_view_count == 0); + ERR_FAIL_COND(p_view_count > RendererSceneRender::MAX_RENDER_VIEWS); + + 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; + uint32_t view_count = p_view_count; + const CameraMatrix *projections = p_projections; + + if (custom_fov) { + // With custom fov we don't support stereo... + float near_plane = p_projections[0].get_z_near(); + float far_plane = p_projections[0].get_z_far(); + float aspect = p_projections[0].get_aspect(); + + camera.set_perspective(custom_fov, aspect, near_plane, far_plane); + + view_count = 1; + projections = &camera; + } + + sky_transform = p_transform.basis * sky_transform; + + if (shader_data->uses_quarter_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_QUARTER_RES_MULTIVIEW : 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); + _render_sky(draw_list, p_time, sky->quarter_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } + + if (shader_data->uses_half_res) { + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_HALF_RES_MULTIVIEW : 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); + _render_sky(draw_list, p_time, sky->half_res_framebuffer, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); + RD::get_singleton()->draw_list_end(); + } +} + +void RendererSceneSkyRD::draw(RD::DrawListID p_draw_list, RendererSceneEnvironmentRD *p_env, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time) { + ERR_FAIL_COND(!p_env); + + ERR_FAIL_COND(p_view_count == 0); + ERR_FAIL_COND(p_view_count > RendererSceneRender::MAX_RENDER_VIEWS); + + 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; + uint32_t view_count = p_view_count; + const CameraMatrix *projections = p_projections; + + if (custom_fov) { + // With custom fov we don't support stereo... + float near_plane = p_projections[0].get_z_near(); + float far_plane = p_projections[0].get_z_far(); + float aspect = p_projections[0].get_aspect(); + + camera.set_perspective(custom_fov, aspect, near_plane, far_plane); + + view_count = 1; + projections = &camera; + } + + sky_transform = p_transform.basis * sky_transform; + + PipelineCacheRD *pipeline = &shader_data->pipelines[view_count > 1 ? SKY_VERSION_BACKGROUND_MULTIVIEW : 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; + } + + _render_sky(p_draw_list, p_time, p_fb, pipeline, material->uniform_set, texture_uniform_set, view_count, projections, sky_transform, multiplier, p_transform.origin); +} + void RendererSceneSkyRD::invalidate_sky(Sky *p_sky) { if (!p_sky->dirty) { p_sky->dirty = true; @@ -1390,7 +1671,7 @@ void RendererSceneSkyRD::update_dirty_skys() { //array (higher quality, 6 times more memory) RD::TextureFormat tf; tf.array_layers = layers * 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = texture_format; tf.texture_type = RD::TEXTURE_TYPE_CUBE_ARRAY; tf.mipmaps = mipmaps; tf.width = w; @@ -1399,13 +1680,13 @@ void RendererSceneSkyRD::update_dirty_skys() { 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); + sky->reflection.update_reflection_data(storage, sky->radiance_size, mipmaps, true, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers, texture_format); } else { //regular cubemap, lower quality (aliasing, less memory) RD::TextureFormat tf; tf.array_layers = 6; - tf.format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; + tf.format = texture_format; tf.texture_type = RD::TEXTURE_TYPE_CUBE; tf.mipmaps = MIN(mipmaps, layers); tf.width = w; @@ -1414,7 +1695,7 @@ void RendererSceneSkyRD::update_dirty_skys() { 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); + sky->reflection.update_reflection_data(storage, sky->radiance_size, MIN(mipmaps, layers), false, sky->radiance, 0, sky->mode == RS::SKY_MODE_REALTIME, roughness_layers, texture_format); } texture_set_dirty = true; } @@ -1422,7 +1703,7 @@ void RendererSceneSkyRD::update_dirty_skys() { // 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.format = texture_format; 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; @@ -1437,7 +1718,7 @@ void RendererSceneSkyRD::update_dirty_skys() { 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.format = texture_format; 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; diff --git a/servers/rendering/renderer_rd/renderer_scene_sky_rd.h b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h index 4f852e55a7..7b670bddd5 100644 --- a/servers/rendering/renderer_rd/renderer_scene_sky_rd.h +++ b/servers/rendering/renderer_rd/renderer_scene_sky_rd.h @@ -64,6 +64,7 @@ public: private: RendererStorageRD *storage; + RD::DataFormat texture_format = RD::DATA_FORMAT_R16G16B16A16_SFLOAT; RID index_buffer; RID index_array; @@ -190,6 +191,10 @@ public: struct Mipmap { RID view; Size2i size; + + // for mobile only + RID views[6]; + RID framebuffers[6]; }; Vector<Mipmap> mipmaps; }; @@ -204,7 +209,7 @@ public: 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 update_reflection_data(RendererStorageRD *p_storage, 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, RD::DataFormat p_texture_format); 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); @@ -284,11 +289,14 @@ public: RendererSceneSkyRD(); void init(RendererStorageRD *p_storage); + void set_texture_format(RD::DataFormat p_texture_format); ~RendererSceneSkyRD(); void setup(RendererSceneEnvironmentRD *p_env, RID p_render_buffers, const CameraMatrix &p_projection, const Transform3D &p_transform, const Size2i p_screen_size, RendererSceneRenderRD *p_scene_render); void update(RendererSceneEnvironmentRD *p_env, const CameraMatrix &p_projection, const Transform3D &p_transform, double p_time); void draw(RendererSceneEnvironmentRD *p_env, bool p_can_continue_color, bool p_can_continue_depth, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time); + void update_res_buffers(RendererSceneEnvironmentRD *p_env, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time); + void draw(RD::DrawListID p_draw_list, RendererSceneEnvironmentRD *p_env, RID p_fb, uint32_t p_view_count, const CameraMatrix *p_projections, const Transform3D &p_transform, double p_time); void invalidate_sky(Sky *p_sky); void update_dirty_skys(); diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.cpp b/servers/rendering/renderer_rd/renderer_storage_rd.cpp index d5c7db6fd2..5b845cbb88 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.cpp +++ b/servers/rendering/renderer_rd/renderer_storage_rd.cpp @@ -4091,7 +4091,7 @@ void RendererStorageRD::particles_set_amount(RID p_particles, int p_amount) { particles->dependency.changed_notify(DEPENDENCY_CHANGED_PARTICLES); } -void RendererStorageRD::particles_set_lifetime(RID p_particles, float p_lifetime) { +void RendererStorageRD::particles_set_lifetime(RID p_particles, double p_lifetime) { Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND(!particles); particles->lifetime = p_lifetime; @@ -4103,7 +4103,7 @@ void RendererStorageRD::particles_set_one_shot(RID p_particles, bool p_one_shot) particles->one_shot = p_one_shot; } -void RendererStorageRD::particles_set_pre_process_time(RID p_particles, float p_time) { +void RendererStorageRD::particles_set_pre_process_time(RID p_particles, double p_time) { Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND(!particles); particles->pre_process_time = p_time; @@ -4126,7 +4126,7 @@ void RendererStorageRD::particles_set_custom_aabb(RID p_particles, const AABB &p particles->dependency.changed_notify(DEPENDENCY_CHANGED_AABB); } -void RendererStorageRD::particles_set_speed_scale(RID p_particles, float p_scale) { +void RendererStorageRD::particles_set_speed_scale(RID p_particles, double p_scale) { Particles *particles = particles_owner.getornull(p_particles); ERR_FAIL_COND(!particles); @@ -4442,7 +4442,7 @@ void RendererStorageRD::particles_set_canvas_sdf_collision(RID p_particles, bool particles->sdf_collision_texture = p_texture; } -void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta) { +void RendererStorageRD::_particles_process(Particles *p_particles, double p_delta) { if (p_particles->particles_material_uniform_set.is_null() || !RD::get_singleton()->uniform_set_is_valid(p_particles->particles_material_uniform_set)) { Vector<RD::Uniform> uniforms; @@ -4491,7 +4491,7 @@ void RendererStorageRD::_particles_process(Particles *p_particles, float p_delta p_particles->particles_material_uniform_set = RD::get_singleton()->uniform_set_create(uniforms, particles_shader.default_shader_rd, 1); } - float new_phase = Math::fmod((float)p_particles->phase + (p_delta / p_particles->lifetime) * p_particles->speed_scale, (float)1.0); + double new_phase = Math::fmod((double)p_particles->phase + (p_delta / p_particles->lifetime) * p_particles->speed_scale, 1.0); //move back history (if there is any) for (uint32_t i = p_particles->frame_history.size() - 1; i > 0; i--) { @@ -5131,14 +5131,14 @@ void RendererStorageRD::update_particles() { bool zero_time_scale = Engine::get_singleton()->get_time_scale() <= 0.0; if (particles->clear && particles->pre_process_time > 0.0) { - float frame_time; + double frame_time; if (fixed_fps > 0) { frame_time = 1.0 / fixed_fps; } else { frame_time = 1.0 / 30.0; } - float todo = particles->pre_process_time; + double todo = particles->pre_process_time; while (todo >= 0) { _particles_process(particles, frame_time); @@ -5147,8 +5147,8 @@ void RendererStorageRD::update_particles() { } if (fixed_fps > 0) { - float frame_time; - float decr; + double frame_time; + double decr; if (zero_time_scale) { frame_time = 0.0; decr = 1.0 / fixed_fps; @@ -5156,13 +5156,13 @@ void RendererStorageRD::update_particles() { frame_time = 1.0 / fixed_fps; decr = frame_time; } - float delta = RendererCompositorRD::singleton->get_frame_delta_time(); + double delta = RendererCompositorRD::singleton->get_frame_delta_time(); if (delta > 0.1) { //avoid recursive stalls if fps goes below 10 delta = 0.1; } else if (delta <= 0.0) { //unlikely but.. delta = 0.001; } - float todo = particles->frame_remainder + delta; + double todo = particles->frame_remainder + delta; while (todo >= frame_time) { _particles_process(particles, frame_time); diff --git a/servers/rendering/renderer_rd/renderer_storage_rd.h b/servers/rendering/renderer_rd/renderer_storage_rd.h index b290c07705..71c3d8b09a 100644 --- a/servers/rendering/renderer_rd/renderer_storage_rd.h +++ b/servers/rendering/renderer_rd/renderer_storage_rd.h @@ -715,12 +715,12 @@ private: struct Particles { RS::ParticlesMode mode = RS::PARTICLES_MODE_3D; bool inactive = true; - float inactive_time = 0.0; + double inactive_time = 0.0; bool emitting = false; bool one_shot = false; int amount = 0; - float lifetime = 1.0; - float pre_process_time = 0.0; + double lifetime = 1.0; + double pre_process_time = 0.0; float explosiveness = 0.0; float randomness = 0.0; bool restart_request = false; @@ -773,12 +773,12 @@ private: uint32_t cycle_number = 0; - float speed_scale = 1.0; + double speed_scale = 1.0; int fixed_fps = 30; bool interpolate = true; bool fractional_delta = false; - float frame_remainder = 0; + double frame_remainder = 0; float collision_base_size = 0.01; bool clear = true; @@ -805,14 +805,14 @@ private: } }; - void _particles_process(Particles *p_particles, float p_delta); + void _particles_process(Particles *p_particles, double p_delta); void _particles_allocate_emission_buffer(Particles *particles); void _particles_free_data(Particles *particles); void _particles_update_buffers(Particles *particles); struct ParticlesShader { struct PushConstant { - float lifetime; + double lifetime; uint32_t clear; uint32_t total_particles; uint32_t trail_size; @@ -2144,13 +2144,13 @@ public: void particles_set_mode(RID p_particles, RS::ParticlesMode p_mode); void particles_set_emitting(RID p_particles, bool p_emitting); void particles_set_amount(RID p_particles, int p_amount); - void particles_set_lifetime(RID p_particles, float p_lifetime); + void particles_set_lifetime(RID p_particles, double p_lifetime); void particles_set_one_shot(RID p_particles, bool p_one_shot); - void particles_set_pre_process_time(RID p_particles, float p_time); + void particles_set_pre_process_time(RID p_particles, double p_time); void particles_set_explosiveness_ratio(RID p_particles, float p_ratio); void particles_set_randomness_ratio(RID p_particles, float p_ratio); void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb); - void particles_set_speed_scale(RID p_particles, float p_scale); + void particles_set_speed_scale(RID p_particles, double p_scale); void particles_set_use_local_coordinates(RID p_particles, bool p_enable); void particles_set_process_material(RID p_particles, RID p_material); void particles_set_fixed_fps(RID p_particles, int p_fps); diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl index 9fa84657d1..63f0ce690e 100644 --- a/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler.glsl @@ -32,53 +32,7 @@ layout(set = 0, binding = 0) uniform samplerCube source_cubemap; layout(rgba16f, set = 1, binding = 0) uniform restrict writeonly imageCube dest_cubemap; -layout(push_constant, binding = 1, std430) uniform Params { - uint face_size; -} -params; - -#define M_PI 3.14159265359 - -void get_dir_0(out vec3 dir, in float u, in float v) { - dir[0] = 1.0; - dir[1] = v; - dir[2] = -u; -} - -void get_dir_1(out vec3 dir, in float u, in float v) { - dir[0] = -1.0; - dir[1] = v; - dir[2] = u; -} - -void get_dir_2(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = 1.0; - dir[2] = -v; -} - -void get_dir_3(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = -1.0; - dir[2] = v; -} - -void get_dir_4(out vec3 dir, in float u, in float v) { - dir[0] = u; - dir[1] = v; - dir[2] = 1.0; -} - -void get_dir_5(out vec3 dir, in float u, in float v) { - dir[0] = -u; - dir[1] = v; - dir[2] = -1.0; -} - -float calcWeight(float u, float v) { - float val = u * u + v * v + 1.0; - return val * sqrt(val); -} +#include "cubemap_downsampler_inc.glsl" void main() { uvec3 id = gl_GlobalInvocationID; diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl new file mode 100644 index 0000000000..b329e67293 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_inc.glsl @@ -0,0 +1,48 @@ +layout(push_constant, binding = 1, std430) uniform Params { + uint face_size; + uint face_id; // only used in raster shader +} +params; + +#define M_PI 3.14159265359 + +void get_dir_0(out vec3 dir, in float u, in float v) { + dir[0] = 1.0; + dir[1] = v; + dir[2] = -u; +} + +void get_dir_1(out vec3 dir, in float u, in float v) { + dir[0] = -1.0; + dir[1] = v; + dir[2] = u; +} + +void get_dir_2(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = 1.0; + dir[2] = -v; +} + +void get_dir_3(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = -1.0; + dir[2] = v; +} + +void get_dir_4(out vec3 dir, in float u, in float v) { + dir[0] = u; + dir[1] = v; + dir[2] = 1.0; +} + +void get_dir_5(out vec3 dir, in float u, in float v) { + dir[0] = -u; + dir[1] = v; + dir[2] = -1.0; +} + +float calcWeight(float u, float v) { + float val = u * u + v * v + 1.0; + return val * sqrt(val); +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl new file mode 100644 index 0000000000..0828ffd921 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_downsampler_raster.glsl @@ -0,0 +1,163 @@ +// Copyright 2016 Activision Publishing, Inc. +// +// 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. + +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_downsampler_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +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] * float(params.face_size); + gl_Position = vec4(base_arr[gl_VertexIndex] * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_downsampler_inc.glsl" + +layout(set = 0, binding = 0) uniform samplerCube source_cubemap; + +layout(location = 0) in vec2 uv_interp; +layout(location = 0) out vec4 frag_color; +/* clang-format on */ + +void main() { + // Converted from compute shader which uses absolute coordinates. + // Could possibly simplify this + float face_size = float(params.face_size); + + if (uv_interp.x < face_size && uv_interp.y < face_size) { + float inv_face_size = 1.0 / face_size; + + float u0 = (uv_interp.x * 2.0 + 1.0 - 0.75) * inv_face_size - 1.0; + float u1 = (uv_interp.x * 2.0 + 1.0 + 0.75) * inv_face_size - 1.0; + + float v0 = (uv_interp.y * 2.0 + 1.0 - 0.75) * -inv_face_size + 1.0; + float v1 = (uv_interp.y * 2.0 + 1.0 + 0.75) * -inv_face_size + 1.0; + + float weights[4]; + weights[0] = calcWeight(u0, v0); + weights[1] = calcWeight(u1, v0); + weights[2] = calcWeight(u0, v1); + weights[3] = calcWeight(u1, v1); + + const float wsum = 0.5 / (weights[0] + weights[1] + weights[2] + weights[3]); + for (int i = 0; i < 4; i++) { + weights[i] = weights[i] * wsum + .125; + } + + vec3 dir; + vec4 color; + switch (params.face_id) { + case 0: + get_dir_0(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_0(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_0(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_0(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 1: + get_dir_1(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_1(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_1(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_1(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 2: + get_dir_2(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_2(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_2(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_2(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 3: + get_dir_3(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_3(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_3(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_3(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + case 4: + get_dir_4(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_4(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_4(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_4(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + default: + get_dir_5(dir, u0, v0); + color = textureLod(source_cubemap, normalize(dir), 0.0) * weights[0]; + + get_dir_5(dir, u1, v0); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[1]; + + get_dir_5(dir, u0, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[2]; + + get_dir_5(dir, u1, v1); + color += textureLod(source_cubemap, normalize(dir), 0.0) * weights[3]; + break; + } + frag_color = color; + } +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl new file mode 100644 index 0000000000..324d306218 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_filter_raster.glsl @@ -0,0 +1,256 @@ +// Copyright 2016 Activision Publishing, Inc. +// +// 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. + +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +layout(push_constant, binding = 1, std430) uniform Params { + int mip_level; + uint face_id; +} +params; + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +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]; + gl_Position = vec4(base_arr[gl_VertexIndex] * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +layout(push_constant, binding = 1, std430) uniform Params { + int mip_level; + uint face_id; +} +params; + +layout(set = 0, binding = 0) uniform samplerCube source_cubemap; + +layout(location = 0) in vec2 uv_interp; +layout(location = 0) out vec4 frag_color; + +/* clang-format on */ + +#ifdef USE_HIGH_QUALITY +#define NUM_TAPS 32 +#else +#define NUM_TAPS 8 +#endif + +#define BASE_RESOLUTION 128 + +#ifdef USE_HIGH_QUALITY +layout(set = 1, binding = 0, std430) buffer restrict readonly Data { + vec4[7][5][3][24] coeffs; +} +data; +#else +layout(set = 1, binding = 0, std430) buffer restrict readonly Data { + vec4[7][5][6] coeffs; +} +data; +#endif + +void get_dir(out vec3 dir, in vec2 uv, in uint face) { + switch (face) { + case 0: + dir = vec3(1.0, uv[1], -uv[0]); + break; + case 1: + dir = vec3(-1.0, uv[1], uv[0]); + break; + case 2: + dir = vec3(uv[0], 1.0, -uv[1]); + break; + case 3: + dir = vec3(uv[0], -1.0, uv[1]); + break; + case 4: + dir = vec3(uv[0], uv[1], 1.0); + break; + default: + dir = vec3(-uv[0], uv[1], -1.0); + break; + } +} + +void main() { + // determine dir / pos for the texel + vec3 dir, adir, frameZ; + { + vec2 uv; + uv.x = uv_interp.x; + uv.y = 1.0 - uv_interp.y; + uv = uv * 2.0 - 1.0; + + get_dir(dir, uv, params.face_id); + frameZ = normalize(dir); + + adir = abs(dir); + } + + // determine which texel this is + // NOTE (macOS/MoltenVK): Do not rename, "level" variable name conflicts with the Metal "level(float lod)" mipmap sampling function name. + int mip_level = 0; + + if (params.mip_level < 0) { + // return as is + frag_color.rgb = textureLod(source_cubemap, frameZ, 0.0).rgb; + frag_color.a = 1.0; + return; + } else if (params.mip_level > 6) { + // maximum level + mip_level = 6; + } else { + mip_level = params.mip_level; + } + + // GGX gather colors + vec4 color = vec4(0.0); + for (int axis = 0; axis < 3; axis++) { + const int otherAxis0 = 1 - (axis & 1) - (axis >> 1); + const int otherAxis1 = 2 - (axis >> 1); + + float frameweight = (max(adir[otherAxis0], adir[otherAxis1]) - .75) / .25; + if (frameweight > 0.0) { + // determine frame + vec3 UpVector; + switch (axis) { + case 0: + UpVector = vec3(1, 0, 0); + break; + case 1: + UpVector = vec3(0, 1, 0); + break; + default: + UpVector = vec3(0, 0, 1); + break; + } + + vec3 frameX = normalize(cross(UpVector, frameZ)); + vec3 frameY = cross(frameZ, frameX); + + // calculate parametrization for polynomial + float Nx = dir[otherAxis0]; + float Ny = dir[otherAxis1]; + float Nz = adir[axis]; + + float NmaxXY = max(abs(Ny), abs(Nx)); + Nx /= NmaxXY; + Ny /= NmaxXY; + + float theta; + if (Ny < Nx) { + if (Ny <= -0.999) + theta = Nx; + else + theta = Ny; + } else { + if (Ny >= 0.999) + theta = -Nx; + else + theta = -Ny; + } + + float phi; + if (Nz <= -0.999) + phi = -NmaxXY; + else if (Nz >= 0.999) + phi = NmaxXY; + else + phi = Nz; + + float theta2 = theta * theta; + float phi2 = phi * phi; + + // sample + for (int iSuperTap = 0; iSuperTap < NUM_TAPS / 4; iSuperTap++) { + const int index = (NUM_TAPS / 4) * axis + iSuperTap; + +#ifdef USE_HIGH_QUALITY + vec4 coeffsDir0[3]; + vec4 coeffsDir1[3]; + vec4 coeffsDir2[3]; + vec4 coeffsLevel[3]; + vec4 coeffsWeight[3]; + + for (int iCoeff = 0; iCoeff < 3; iCoeff++) { + coeffsDir0[iCoeff] = data.coeffs[mip_level][0][iCoeff][index]; + coeffsDir1[iCoeff] = data.coeffs[mip_level][1][iCoeff][index]; + coeffsDir2[iCoeff] = data.coeffs[mip_level][2][iCoeff][index]; + coeffsLevel[iCoeff] = data.coeffs[mip_level][3][iCoeff][index]; + coeffsWeight[iCoeff] = data.coeffs[mip_level][4][iCoeff][index]; + } + + for (int iSubTap = 0; iSubTap < 4; iSubTap++) { + // determine sample attributes (dir, weight, mip_level) + vec3 sample_dir = frameX * (coeffsDir0[0][iSubTap] + coeffsDir0[1][iSubTap] * theta2 + coeffsDir0[2][iSubTap] * phi2) + frameY * (coeffsDir1[0][iSubTap] + coeffsDir1[1][iSubTap] * theta2 + coeffsDir1[2][iSubTap] * phi2) + frameZ * (coeffsDir2[0][iSubTap] + coeffsDir2[1][iSubTap] * theta2 + coeffsDir2[2][iSubTap] * phi2); + + float sample_level = coeffsLevel[0][iSubTap] + coeffsLevel[1][iSubTap] * theta2 + coeffsLevel[2][iSubTap] * phi2; + + float sample_weight = coeffsWeight[0][iSubTap] + coeffsWeight[1][iSubTap] * theta2 + coeffsWeight[2][iSubTap] * phi2; +#else + vec4 coeffsDir0 = data.coeffs[mip_level][0][index]; + vec4 coeffsDir1 = data.coeffs[mip_level][1][index]; + vec4 coeffsDir2 = data.coeffs[mip_level][2][index]; + vec4 coeffsLevel = data.coeffs[mip_level][3][index]; + vec4 coeffsWeight = data.coeffs[mip_level][4][index]; + + for (int iSubTap = 0; iSubTap < 4; iSubTap++) { + // determine sample attributes (dir, weight, mip_level) + vec3 sample_dir = frameX * coeffsDir0[iSubTap] + frameY * coeffsDir1[iSubTap] + frameZ * coeffsDir2[iSubTap]; + + float sample_level = coeffsLevel[iSubTap]; + + float sample_weight = coeffsWeight[iSubTap]; +#endif + + sample_weight *= frameweight; + + // adjust for jacobian + sample_dir /= max(abs(sample_dir[0]), max(abs(sample_dir[1]), abs(sample_dir[2]))); + sample_level += 0.75 * log2(dot(sample_dir, sample_dir)); + // sample cubemap + color.xyz += textureLod(source_cubemap, normalize(sample_dir), sample_level).xyz * sample_weight; + color.w += sample_weight; + } + } + } + } + color /= color.w; + + // write color + color.xyz = max(vec3(0.0), color.xyz); + color.w = 1.0; + + frag_color = color; +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl index ce7c03c1d4..28f4dc59ec 100644 --- a/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness.glsl @@ -12,100 +12,7 @@ layout(set = 0, binding = 0) uniform samplerCube source_cube; layout(rgba16f, set = 1, binding = 0) uniform restrict writeonly imageCube dest_cubemap; -layout(push_constant, binding = 1, std430) uniform Params { - uint face_id; - uint sample_count; - float roughness; - bool use_direct_write; - float face_size; -} -params; - -#define M_PI 3.14159265359 - -vec3 texelCoordToVec(vec2 uv, uint faceID) { - mat3 faceUvVectors[6]; - - // -x - faceUvVectors[1][0] = vec3(0.0, 0.0, 1.0); // u -> +z - faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[1][2] = vec3(-1.0, 0.0, 0.0); // -x face - - // +x - faceUvVectors[0][0] = vec3(0.0, 0.0, -1.0); // u -> -z - faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[0][2] = vec3(1.0, 0.0, 0.0); // +x face - - // -y - faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[3][1] = vec3(0.0, 0.0, -1.0); // v -> -z - faceUvVectors[3][2] = vec3(0.0, -1.0, 0.0); // -y face - - // +y - faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[2][1] = vec3(0.0, 0.0, 1.0); // v -> +z - faceUvVectors[2][2] = vec3(0.0, 1.0, 0.0); // +y face - - // -z - faceUvVectors[5][0] = vec3(-1.0, 0.0, 0.0); // u -> -x - faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[5][2] = vec3(0.0, 0.0, -1.0); // -z face - - // +z - faceUvVectors[4][0] = vec3(1.0, 0.0, 0.0); // u -> +x - faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y - faceUvVectors[4][2] = vec3(0.0, 0.0, 1.0); // +z face - - // out = u * s_faceUv[0] + v * s_faceUv[1] + s_faceUv[2]. - vec3 result = (faceUvVectors[faceID][0] * uv.x) + (faceUvVectors[faceID][1] * uv.y) + faceUvVectors[faceID][2]; - return normalize(result); -} - -vec3 ImportanceSampleGGX(vec2 Xi, float Roughness, vec3 N) { - float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph] - - // Compute distribution direction - float Phi = 2.0 * M_PI * Xi.x; - float CosTheta = sqrt((1.0 - Xi.y) / (1.0 + (a * a - 1.0) * Xi.y)); - float SinTheta = sqrt(1.0 - CosTheta * CosTheta); - - // Convert to spherical direction - vec3 H; - H.x = SinTheta * cos(Phi); - H.y = SinTheta * sin(Phi); - H.z = CosTheta; - - vec3 UpVector = abs(N.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(1.0, 0.0, 0.0); - vec3 TangentX = normalize(cross(UpVector, N)); - vec3 TangentY = cross(N, TangentX); - - // Tangent to world space - return TangentX * H.x + TangentY * H.y + N * H.z; -} - -// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html -float GGX(float NdotV, float a) { - float k = a / 2.0; - return NdotV / (NdotV * (1.0 - k) + k); -} - -// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html -float G_Smith(float a, float nDotV, float nDotL) { - return GGX(nDotL, a * a) * GGX(nDotV, a * a); -} - -float radicalInverse_VdC(uint bits) { - bits = (bits << 16u) | (bits >> 16u); - bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u); - bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u); - bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u); - bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u); - return float(bits) * 2.3283064365386963e-10; // / 0x100000000 -} - -vec2 Hammersley(uint i, uint N) { - return vec2(float(i) / float(N), radicalInverse_VdC(i)); -} +#include "cubemap_roughness_inc.glsl" void main() { uvec3 id = gl_GlobalInvocationID; diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl new file mode 100644 index 0000000000..80c0ac4fb4 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness_inc.glsl @@ -0,0 +1,94 @@ +#define M_PI 3.14159265359 + +layout(push_constant, binding = 1, std430) uniform Params { + uint face_id; + uint sample_count; + float roughness; + bool use_direct_write; + float face_size; +} +params; + +vec3 texelCoordToVec(vec2 uv, uint faceID) { + mat3 faceUvVectors[6]; + + // -x + faceUvVectors[1][0] = vec3(0.0, 0.0, 1.0); // u -> +z + faceUvVectors[1][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[1][2] = vec3(-1.0, 0.0, 0.0); // -x face + + // +x + faceUvVectors[0][0] = vec3(0.0, 0.0, -1.0); // u -> -z + faceUvVectors[0][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[0][2] = vec3(1.0, 0.0, 0.0); // +x face + + // -y + faceUvVectors[3][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[3][1] = vec3(0.0, 0.0, -1.0); // v -> -z + faceUvVectors[3][2] = vec3(0.0, -1.0, 0.0); // -y face + + // +y + faceUvVectors[2][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[2][1] = vec3(0.0, 0.0, 1.0); // v -> +z + faceUvVectors[2][2] = vec3(0.0, 1.0, 0.0); // +y face + + // -z + faceUvVectors[5][0] = vec3(-1.0, 0.0, 0.0); // u -> -x + faceUvVectors[5][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[5][2] = vec3(0.0, 0.0, -1.0); // -z face + + // +z + faceUvVectors[4][0] = vec3(1.0, 0.0, 0.0); // u -> +x + faceUvVectors[4][1] = vec3(0.0, -1.0, 0.0); // v -> -y + faceUvVectors[4][2] = vec3(0.0, 0.0, 1.0); // +z face + + // out = u * s_faceUv[0] + v * s_faceUv[1] + s_faceUv[2]. + vec3 result = (faceUvVectors[faceID][0] * uv.x) + (faceUvVectors[faceID][1] * uv.y) + faceUvVectors[faceID][2]; + return normalize(result); +} + +vec3 ImportanceSampleGGX(vec2 Xi, float Roughness, vec3 N) { + float a = Roughness * Roughness; // DISNEY'S ROUGHNESS [see Burley'12 siggraph] + + // Compute distribution direction + float Phi = 2.0 * M_PI * Xi.x; + float CosTheta = sqrt((1.0 - Xi.y) / (1.0 + (a * a - 1.0) * Xi.y)); + float SinTheta = sqrt(1.0 - CosTheta * CosTheta); + + // Convert to spherical direction + vec3 H; + H.x = SinTheta * cos(Phi); + H.y = SinTheta * sin(Phi); + H.z = CosTheta; + + vec3 UpVector = abs(N.z) < 0.999 ? vec3(0.0, 0.0, 1.0) : vec3(1.0, 0.0, 0.0); + vec3 TangentX = normalize(cross(UpVector, N)); + vec3 TangentY = cross(N, TangentX); + + // Tangent to world space + return TangentX * H.x + TangentY * H.y + N * H.z; +} + +// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float GGX(float NdotV, float a) { + float k = a / 2.0; + return NdotV / (NdotV * (1.0 - k) + k); +} + +// http://graphicrants.blogspot.com.au/2013/08/specular-brdf-reference.html +float G_Smith(float a, float nDotV, float nDotL) { + return GGX(nDotL, a * a) * GGX(nDotV, a * a); +} + +float radicalInverse_VdC(uint bits) { + bits = (bits << 16u) | (bits >> 16u); + bits = ((bits & 0x55555555u) << 1u) | ((bits & 0xAAAAAAAAu) >> 1u); + bits = ((bits & 0x33333333u) << 2u) | ((bits & 0xCCCCCCCCu) >> 2u); + bits = ((bits & 0x0F0F0F0Fu) << 4u) | ((bits & 0xF0F0F0F0u) >> 4u); + bits = ((bits & 0x00FF00FFu) << 8u) | ((bits & 0xFF00FF00u) >> 8u); + return float(bits) * 2.3283064365386963e-10; // / 0x100000000 +} + +vec2 Hammersley(uint i, uint N) { + return vec2(float(i) / float(N), radicalInverse_VdC(i)); +} diff --git a/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl b/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl new file mode 100644 index 0000000000..2570308816 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/cubemap_roughness_raster.glsl @@ -0,0 +1,63 @@ +/* clang-format off */ +#[vertex] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_roughness_inc.glsl" + +layout(location = 0) out vec2 uv_interp; +/* clang-format on */ + +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]; + gl_Position = vec4(uv_interp * 2.0 - 1.0, 0.0, 1.0); +} + +/* clang-format off */ +#[fragment] + +#version 450 + +#VERSION_DEFINES + +#include "cubemap_roughness_inc.glsl" + +layout(location = 0) in vec2 uv_interp; + +layout(set = 0, binding = 0) uniform samplerCube source_cube; + +layout(location = 0) out vec4 frag_color; +/* clang-format on */ + +void main() { + vec3 N = texelCoordToVec(uv_interp * 2.0 - 1.0, params.face_id); + + //vec4 color = color_interp; + + if (params.use_direct_write) { + frag_color = vec4(texture(source_cube, N).rgb, 1.0); + } else { + vec4 sum = vec4(0.0, 0.0, 0.0, 0.0); + + for (uint sampleNum = 0u; sampleNum < params.sample_count; sampleNum++) { + vec2 xi = Hammersley(sampleNum, params.sample_count); + + vec3 H = ImportanceSampleGGX(xi, params.roughness, N); + vec3 V = N; + vec3 L = (2.0 * dot(V, H) * H - V); + + float ndotl = clamp(dot(N, L), 0.0, 1.0); + + if (ndotl > 0.0) { + sum.rgb += textureLod(source_cube, L, 0.0).rgb * ndotl; + sum.a += ndotl; + } + } + sum /= sum.a; + + frag_color = vec4(sum.rgb, 1.0); + } +} diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl index b3a349c948..4f140dd10d 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl @@ -466,6 +466,11 @@ layout(location = 0) out vec4 frag_color; #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) +/* Make a default specular mode SPECULAR_SCHLICK_GGX. */ +#if !defined(SPECULAR_DISABLED) && !defined(SPECULAR_SCHLICK_GGX) && !defined(SPECULAR_BLINN) && !defined(SPECULAR_PHONG) && !defined(SPECULAR_TOON) +#define SPECULAR_SCHLICK_GGX +#endif + #include "scene_forward_lights_inc.glsl" #include "scene_forward_gi_inc.glsl" @@ -907,6 +912,8 @@ void main() { specular_light = textureLod(samplerCube(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), ref_vec, roughness * MAX_ROUGHNESS_LOD).rgb; #endif //USE_RADIANCE_CUBEMAP_ARRAY + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); + specular_light *= horizon * horizon; specular_light *= scene_data.ambient_light_color_energy.a; } diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl index 70900a847c..681656e1d5 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl @@ -491,6 +491,11 @@ layout(location = 0) out vec4 frag_color; #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) +/* Make a default specular mode SPECULAR_SCHLICK_GGX. */ +#if !defined(SPECULAR_DISABLED) && !defined(SPECULAR_SCHLICK_GGX) && !defined(SPECULAR_BLINN) && !defined(SPECULAR_PHONG) && !defined(SPECULAR_TOON) +#define SPECULAR_SCHLICK_GGX +#endif + #include "scene_forward_lights_inc.glsl" #endif //!defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) @@ -843,6 +848,8 @@ void main() { specular_light = textureLod(samplerCube(radiance_cubemap, material_samplers[SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP]), ref_vec, roughness * MAX_ROUGHNESS_LOD).rgb; #endif //USE_RADIANCE_CUBEMAP_ARRAY + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); + specular_light *= horizon * horizon; specular_light *= scene_data.ambient_light_color_energy.a; } diff --git a/servers/rendering/renderer_rd/shaders/tonemap.glsl b/servers/rendering/renderer_rd/shaders/tonemap.glsl index 23f83b3b9c..3c685c25b9 100644 --- a/servers/rendering/renderer_rd/shaders/tonemap.glsl +++ b/servers/rendering/renderer_rd/shaders/tonemap.glsl @@ -35,11 +35,15 @@ void main() { layout(location = 0) in vec2 uv_interp; -#ifdef MULTIVIEW +#ifdef SUBPASS +layout(input_attachment_index = 0, set = 0, binding = 0) uniform subpassInput input_color; +#else +#if MULTIVIEW layout(set = 0, binding = 0) uniform sampler2DArray source_color; #else layout(set = 0, binding = 0) uniform sampler2D source_color; #endif +#endif layout(set = 1, binding = 0) uniform sampler2D source_auto_exposure; layout(set = 2, binding = 0) uniform sampler2D source_glow; #ifdef USE_1D_LUT @@ -180,10 +184,6 @@ vec3 tonemap_aces(vec3 color, float white) { } vec3 tonemap_reinhard(vec3 color, float white) { - // Ensure color values are positive. - // They can be negative in the case of negative lights, which leads to undesired behavior. - color = max(vec3(0.0), color); - return (white * color + color) / (color * white + white); } @@ -207,7 +207,7 @@ vec3 apply_tonemapping(vec3 color, float white) { // inputs are LINEAR, always o return tonemap_reinhard(color, white); } else if (params.tonemapper == TONEMAPPER_FILMIC) { return tonemap_filmic(color, white); - } else { //aces + } else { // TONEMAPPER_ACES return tonemap_aces(color, white); } } @@ -291,6 +291,7 @@ vec3 apply_color_correction(vec3 color) { } #endif +#ifndef SUBPASS vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { const float FXAA_REDUCE_MIN = (1.0 / 128.0); const float FXAA_REDUCE_MUL = (1.0 / 8.0); @@ -346,6 +347,7 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { return rgbB; } } +#endif // !SUBPASS // From http://alex.vlachos.com/graphics/Alex_Vlachos_Advanced_VR_Rendering_GDC2015.pdf // and https://www.shadertoy.com/view/MslGR8 (5th one starting from the bottom) @@ -360,7 +362,10 @@ vec3 screen_space_dither(vec2 frag_coord) { } void main() { -#ifdef MULTIVIEW +#ifdef SUBPASS + // SUBPASS and MULTIVIEW can be combined but in that case we're already reading from the correct layer + vec3 color = subpassLoad(input_color).rgb; +#elif MULTIVIEW vec3 color = textureLod(source_color, vec3(uv_interp, ViewIndex), 0.0f).rgb; #else vec3 color = textureLod(source_color, uv_interp, 0.0f).rgb; @@ -370,14 +375,16 @@ void main() { float exposure = params.exposure; +#ifndef SUBPASS if (params.use_auto_exposure) { exposure *= 1.0 / (texelFetch(source_auto_exposure, ivec2(0, 0), 0).r / params.auto_exposure_grey); } +#endif color *= exposure; // Early Tonemap & SRGB Conversion - +#ifndef SUBPASS if (params.use_glow && params.glow_mode == GLOW_MODE_MIX) { vec3 glow = gather_glow(source_glow, uv_interp); color.rgb = mix(color.rgb, glow, params.glow_intensity); @@ -386,15 +393,21 @@ void main() { if (params.use_fxaa) { color = do_fxaa(color, exposure, uv_interp); } +#endif + if (params.use_debanding) { // For best results, debanding should be done before tonemapping. // Otherwise, we're adding noise to an already-quantized image. color += screen_space_dither(gl_FragCoord.xy); } - color = apply_tonemapping(color, params.white); + + // Ensure color values passed to tonemappers are positive. + // They can be negative in the case of negative lights, which leads to undesired behavior. + color = apply_tonemapping(max(vec3(0.0), color), params.white); color = linear_to_srgb(color); // regular linear -> SRGB conversion +#ifndef SUBPASS // Glow if (params.use_glow && params.glow_mode != GLOW_MODE_MIX) { @@ -406,6 +419,7 @@ void main() { color = apply_glow(color, glow); } +#endif // Additional effects diff --git a/servers/rendering/renderer_scene_cull.cpp b/servers/rendering/renderer_scene_cull.cpp index 83d1b33bf2..6357a6f4c2 100644 --- a/servers/rendering/renderer_scene_cull.cpp +++ b/servers/rendering/renderer_scene_cull.cpp @@ -1866,7 +1866,7 @@ void RendererSceneCull::_update_instance_lightmap_captures(Instance *p_instance) //rotate it Basis rot = lightmap->transform.basis.orthonormalized(); for (int i = 0; i < 3; i++) { - float csh[9]; + real_t csh[9]; for (int j = 0; j < 9; j++) { csh[j] = sh[j][i]; } @@ -1878,7 +1878,7 @@ void RendererSceneCull::_update_instance_lightmap_captures(Instance *p_instance) Vector3 inner_pos = ((lm_pos - bounds.position) / bounds.size) * 2.0 - Vector3(1.0, 1.0, 1.0); - float blend = MAX(inner_pos.x, MAX(inner_pos.y, inner_pos.z)); + real_t blend = MAX(inner_pos.x, MAX(inner_pos.y, inner_pos.z)); //make blend more rounded blend = Math::lerp(inner_pos.length(), blend, blend); blend *= blend; diff --git a/servers/rendering/renderer_scene_occlusion_cull.h b/servers/rendering/renderer_scene_occlusion_cull.h index 1d0f53c0bf..e06b3ba153 100644 --- a/servers/rendering/renderer_scene_occlusion_cull.h +++ b/servers/rendering/renderer_scene_occlusion_cull.h @@ -60,7 +60,7 @@ public: void update_mips(); - _FORCE_INLINE_ bool is_occluded(const float p_bounds[6], const Vector3 &p_cam_position, const Transform3D &p_cam_inv_transform, const CameraMatrix &p_cam_projection, float p_near) const { + _FORCE_INLINE_ bool is_occluded(const real_t p_bounds[6], const Vector3 &p_cam_position, const Transform3D &p_cam_inv_transform, const CameraMatrix &p_cam_projection, real_t p_near) const { if (is_empty()) { return false; } diff --git a/servers/rendering/renderer_storage.h b/servers/rendering/renderer_storage.h index b9bc349f79..6ffbdbf685 100644 --- a/servers/rendering/renderer_storage.h +++ b/servers/rendering/renderer_storage.h @@ -468,13 +468,13 @@ public: virtual bool particles_get_emitting(RID p_particles) = 0; virtual void particles_set_amount(RID p_particles, int p_amount) = 0; - virtual void particles_set_lifetime(RID p_particles, float p_lifetime) = 0; + virtual void particles_set_lifetime(RID p_particles, double p_lifetime) = 0; virtual void particles_set_one_shot(RID p_particles, bool p_one_shot) = 0; - virtual void particles_set_pre_process_time(RID p_particles, float p_time) = 0; + virtual void particles_set_pre_process_time(RID p_particles, double p_time) = 0; virtual void particles_set_explosiveness_ratio(RID p_particles, float p_ratio) = 0; virtual void particles_set_randomness_ratio(RID p_particles, float p_ratio) = 0; virtual void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb) = 0; - virtual void particles_set_speed_scale(RID p_particles, float p_scale) = 0; + virtual void particles_set_speed_scale(RID p_particles, double p_scale) = 0; virtual void particles_set_use_local_coordinates(RID p_particles, bool p_enable) = 0; virtual void particles_set_process_material(RID p_particles, RID p_material) = 0; virtual void particles_set_fixed_fps(RID p_particles, int p_fps) = 0; diff --git a/servers/rendering/rendering_device.cpp b/servers/rendering/rendering_device.cpp index b298ad193b..0297724a22 100644 --- a/servers/rendering/rendering_device.cpp +++ b/servers/rendering/rendering_device.cpp @@ -739,7 +739,7 @@ void RenderingDevice::_bind_methods() { BIND_ENUM_CONSTANT(TEXTURE_USAGE_CAN_UPDATE_BIT); BIND_ENUM_CONSTANT(TEXTURE_USAGE_CAN_COPY_FROM_BIT); BIND_ENUM_CONSTANT(TEXTURE_USAGE_CAN_COPY_TO_BIT); - BIND_ENUM_CONSTANT(TEXTURE_USAGE_RESOLVE_ATTACHMENT_BIT); + BIND_ENUM_CONSTANT(TEXTURE_USAGE_INPUT_ATTACHMENT_BIT); BIND_ENUM_CONSTANT(TEXTURE_SWIZZLE_IDENTITY); BIND_ENUM_CONSTANT(TEXTURE_SWIZZLE_ZERO); diff --git a/servers/rendering/rendering_device.h b/servers/rendering/rendering_device.h index eaf1ace798..bb028cb84c 100644 --- a/servers/rendering/rendering_device.h +++ b/servers/rendering/rendering_device.h @@ -422,7 +422,7 @@ public: TEXTURE_USAGE_CAN_UPDATE_BIT = (1 << 6), TEXTURE_USAGE_CAN_COPY_FROM_BIT = (1 << 7), TEXTURE_USAGE_CAN_COPY_TO_BIT = (1 << 8), - TEXTURE_USAGE_RESOLVE_ATTACHMENT_BIT = (1 << 9), + TEXTURE_USAGE_INPUT_ATTACHMENT_BIT = (1 << 9), }; enum TextureSwizzle { @@ -1065,6 +1065,7 @@ 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 uint32_t draw_list_get_current_pass() = 0; virtual DrawListID draw_list_switch_to_next_pass() = 0; virtual Error draw_list_switch_to_next_pass_split(uint32_t p_splits, DrawListID *r_split_ids) = 0; diff --git a/servers/rendering/rendering_server_default.cpp b/servers/rendering/rendering_server_default.cpp index c79dfb1649..bed6ade1f6 100644 --- a/servers/rendering/rendering_server_default.cpp +++ b/servers/rendering/rendering_server_default.cpp @@ -134,8 +134,8 @@ void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { } if (RSG::storage->capturing_timestamps) { - new_profile.write[i].gpu_msec = float((time_gpu - base_gpu) / 1000) / 1000.0; - new_profile.write[i].cpu_msec = float(time_cpu - base_cpu) / 1000.0; + new_profile.write[i].gpu_msec = double((time_gpu - base_gpu) / 1000) / 1000.0; + new_profile.write[i].cpu_msec = double(time_cpu - base_cpu) / 1000.0; new_profile.write[i].name = RSG::storage->get_captured_timestamp_name(i); } } @@ -149,7 +149,7 @@ void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { if (print_frame_profile_ticks_from == 0) { print_frame_profile_ticks_from = OS::get_singleton()->get_ticks_usec(); } - float total_time = 0.0; + double total_time = 0.0; for (int i = 0; i < frame_profile.size() - 1; i++) { String name = frame_profile[i].name; @@ -157,7 +157,7 @@ void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { continue; } - float time = frame_profile[i + 1].gpu_msec - frame_profile[i].gpu_msec; + double 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)) { @@ -179,7 +179,7 @@ void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { 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); + double time = E.value() / double(print_frame_profile_frame_count); if (time > print_threshold) { print_line("\t-" + E.key() + ": " + rtos(time) + "ms"); } @@ -193,7 +193,7 @@ void RenderingServerDefault::_draw(bool p_swap_buffers, double frame_step) { RSG::storage->update_memory_info(); } -float RenderingServerDefault::get_frame_setup_time_cpu() const { +double RenderingServerDefault::get_frame_setup_time_cpu() const { return frame_setup_time; } diff --git a/servers/rendering/rendering_server_default.h b/servers/rendering/rendering_server_default.h index 282b0564da..381f38d80e 100644 --- a/servers/rendering/rendering_server_default.h +++ b/servers/rendering/rendering_server_default.h @@ -71,7 +71,7 @@ class RenderingServerDefault : public RenderingServer { uint64_t frame_profile_frame; Vector<FrameProfileArea> frame_profile; - float frame_setup_time = 0; + double frame_setup_time = 0; //for printing bool print_gpu_profile = false; @@ -440,13 +440,13 @@ public: 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_lifetime, RID, double) FUNC2(particles_set_one_shot, RID, bool) - FUNC2(particles_set_pre_process_time, RID, float) + FUNC2(particles_set_pre_process_time, RID, double) 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_speed_scale, RID, double) FUNC2(particles_set_use_local_coordinates, RID, bool) FUNC2(particles_set_process_material, RID, RID) FUNC2(particles_set_fixed_fps, RID, int) @@ -574,8 +574,8 @@ public: FUNC2(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) + FUNC1RC(double, viewport_get_measured_render_time_cpu, RID) + FUNC1RC(double, viewport_get_measured_render_time_gpu, RID) FUNC2(call_set_vsync_mode, DisplayServer::VSyncMode, DisplayServer::WindowID) @@ -890,7 +890,7 @@ public: /* TESTING */ - virtual float get_frame_setup_time_cpu() const override; + virtual double 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) override; virtual void set_default_clear_color(const Color &p_color) override; diff --git a/servers/rendering/shader_types.cpp b/servers/rendering/shader_types.cpp index 376d23ccb3..4488069698 100644 --- a/servers/rendering/shader_types.cpp +++ b/servers/rendering/shader_types.cpp @@ -31,18 +31,22 @@ #include "shader_types.h" #include "core/math/math_defs.h" -const Map<StringName, ShaderLanguage::FunctionInfo> &ShaderTypes::get_functions(RS::ShaderMode p_mode) { +const Map<StringName, ShaderLanguage::FunctionInfo> &ShaderTypes::get_functions(RS::ShaderMode p_mode) const { return shader_modes[p_mode].functions; } -const Vector<StringName> &ShaderTypes::get_modes(RS::ShaderMode p_mode) { +const Vector<StringName> &ShaderTypes::get_modes(RS::ShaderMode p_mode) const { return shader_modes[p_mode].modes; } -const Set<String> &ShaderTypes::get_types() { +const Set<String> &ShaderTypes::get_types() const { return shader_types; } +const List<String> &ShaderTypes::get_types_list() const { + return shader_types_list; +} + ShaderTypes *ShaderTypes::singleton = nullptr; static ShaderLanguage::BuiltInInfo constt(ShaderLanguage::DataType p_type) { @@ -440,8 +444,12 @@ ShaderTypes::ShaderTypes() { shader_modes[RS::SHADER_SKY].modes.push_back("use_quarter_res_pass"); shader_modes[RS::SHADER_SKY].modes.push_back("disable_fog"); - shader_types.insert("spatial"); - shader_types.insert("canvas_item"); - shader_types.insert("particles"); - shader_types.insert("sky"); + shader_types_list.push_back("spatial"); + shader_types_list.push_back("canvas_item"); + shader_types_list.push_back("particles"); + shader_types_list.push_back("sky"); + + for (int i = 0; i < shader_types_list.size(); i++) { + shader_types.insert(shader_types_list[i]); + } } diff --git a/servers/rendering/shader_types.h b/servers/rendering/shader_types.h index e59cef6b79..75a310a1b1 100644 --- a/servers/rendering/shader_types.h +++ b/servers/rendering/shader_types.h @@ -46,13 +46,15 @@ class ShaderTypes { static ShaderTypes *singleton; Set<String> shader_types; + List<String> shader_types_list; public: static ShaderTypes *get_singleton() { return singleton; } - const Map<StringName, ShaderLanguage::FunctionInfo> &get_functions(RS::ShaderMode p_mode); - const Vector<StringName> &get_modes(RS::ShaderMode p_mode); - const Set<String> &get_types(); + const Map<StringName, ShaderLanguage::FunctionInfo> &get_functions(RS::ShaderMode p_mode) const; + const Vector<StringName> &get_modes(RS::ShaderMode p_mode) const; + const Set<String> &get_types() const; + const List<String> &get_types_list() const; ShaderTypes(); }; diff --git a/servers/rendering_server.cpp b/servers/rendering_server.cpp index 314b59869d..2da8e0f297 100644 --- a/servers/rendering_server.cpp +++ b/servers/rendering_server.cpp @@ -242,7 +242,7 @@ RID RenderingServer::_make_test_cube() { return test_cube; } -RID RenderingServer::make_sphere_mesh(int p_lats, int p_lons, float p_radius) { +RID RenderingServer::make_sphere_mesh(int p_lats, int p_lons, real_t p_radius) { Vector<Vector3> vertices; Vector<Vector3> normals; const double lat_step = Math_TAU / p_lats; @@ -412,26 +412,40 @@ Error RenderingServer::_surface_set_data(Array p_arrays, uint32_t p_format, uint } break; case RS::ARRAY_TANGENT: { - ERR_FAIL_COND_V(p_arrays[ai].get_type() != Variant::PACKED_FLOAT32_ARRAY, ERR_INVALID_PARAMETER); + Variant::Type type = p_arrays[ai].get_type(); + ERR_FAIL_COND_V(type != Variant::PACKED_FLOAT32_ARRAY && type != Variant::PACKED_FLOAT64_ARRAY, ERR_INVALID_PARAMETER); + if (type == Variant::PACKED_FLOAT32_ARRAY) { + Vector<float> array = p_arrays[ai]; + ERR_FAIL_COND_V(array.size() != p_vertex_array_len * 4, ERR_INVALID_PARAMETER); + const float *src = array.ptr(); - Vector<real_t> array = p_arrays[ai]; - - ERR_FAIL_COND_V(array.size() != p_vertex_array_len * 4, ERR_INVALID_PARAMETER); - - const real_t *src = array.ptr(); + for (int i = 0; i < p_vertex_array_len; i++) { + uint32_t value = 0; + value |= CLAMP(int((src[i * 4 + 0] * 0.5 + 0.5) * 1023.0), 0, 1023); + value |= CLAMP(int((src[i * 4 + 1] * 0.5 + 0.5) * 1023.0), 0, 1023) << 10; + value |= CLAMP(int((src[i * 4 + 2] * 0.5 + 0.5) * 1023.0), 0, 1023) << 20; + if (src[i * 4 + 3] > 0) { + value |= 3 << 30; + } - for (int i = 0; i < p_vertex_array_len; i++) { - uint32_t value = 0; - value |= CLAMP(int((src[i * 4 + 0] * 0.5 + 0.5) * 1023.0), 0, 1023); - value |= CLAMP(int((src[i * 4 + 1] * 0.5 + 0.5) * 1023.0), 0, 1023) << 10; - value |= CLAMP(int((src[i * 4 + 2] * 0.5 + 0.5) * 1023.0), 0, 1023) << 20; - if (src[i * 4 + 3] > 0) { - value |= 3 << 30; + memcpy(&vw[p_offsets[ai] + i * p_vertex_stride], &value, 4); } + } else { // if (type == Variant::PACKED_FLOAT64_ARRAY) + Vector<double> array = p_arrays[ai]; + ERR_FAIL_COND_V(array.size() != p_vertex_array_len * 4, ERR_INVALID_PARAMETER); + const double *src = array.ptr(); - memcpy(&vw[p_offsets[ai] + i * p_vertex_stride], &value, 4); + for (int i = 0; i < p_vertex_array_len; i++) { + uint32_t value = 0; + value |= CLAMP(int((src[i * 4 + 0] * 0.5 + 0.5) * 1023.0), 0, 1023); + value |= CLAMP(int((src[i * 4 + 1] * 0.5 + 0.5) * 1023.0), 0, 1023) << 10; + value |= CLAMP(int((src[i * 4 + 2] * 0.5 + 0.5) * 1023.0), 0, 1023) << 20; + if (src[i * 4 + 3] > 0) { + value |= 3 << 30; + } + memcpy(&vw[p_offsets[ai] + i * p_vertex_stride], &value, 4); + } } - } break; case RS::ARRAY_COLOR: { ERR_FAIL_COND_V(p_arrays[ai].get_type() != Variant::PACKED_COLOR_ARRAY, ERR_INVALID_PARAMETER); @@ -543,27 +557,38 @@ Error RenderingServer::_surface_set_data(Array p_arrays, uint32_t p_format, uint } break; case RS::ARRAY_WEIGHTS: { - ERR_FAIL_COND_V(p_arrays[ai].get_type() != Variant::PACKED_FLOAT32_ARRAY, ERR_INVALID_PARAMETER); - + Variant::Type type = p_arrays[ai].get_type(); + ERR_FAIL_COND_V(type != Variant::PACKED_FLOAT32_ARRAY && type != Variant::PACKED_FLOAT64_ARRAY, ERR_INVALID_PARAMETER); uint32_t bone_count = (p_format & ARRAY_FLAG_USE_8_BONE_WEIGHTS) ? 8 : 4; + if (type == Variant::PACKED_FLOAT32_ARRAY) { + Vector<float> array = p_arrays[ai]; + ERR_FAIL_COND_V(array.size() != (int32_t)(p_vertex_array_len * bone_count), ERR_INVALID_PARAMETER); + const float *src = array.ptr(); + { + uint16_t data[8]; + for (int i = 0; i < p_vertex_array_len; i++) { + for (uint32_t j = 0; j < bone_count; j++) { + data[j] = CLAMP(src[i * bone_count + j] * 65535, 0, 65535); + } - Vector<real_t> array = p_arrays[ai]; - - ERR_FAIL_COND_V(array.size() != (int32_t)(p_vertex_array_len * bone_count), ERR_INVALID_PARAMETER); - - const real_t *src = array.ptr(); - - { - uint16_t data[8]; - for (int i = 0; i < p_vertex_array_len; i++) { - for (uint32_t j = 0; j < bone_count; j++) { - data[j] = CLAMP(src[i * bone_count + j] * 65535, 0, 65535); + memcpy(&sw[p_offsets[ai] + i * p_skin_stride], data, 2 * bone_count); } + } + } else { // if (type == Variant::PACKED_FLOAT64_ARRAY) + Vector<double> array = p_arrays[ai]; + ERR_FAIL_COND_V(array.size() != (int32_t)(p_vertex_array_len * bone_count), ERR_INVALID_PARAMETER); + const double *src = array.ptr(); + { + uint16_t data[8]; + for (int i = 0; i < p_vertex_array_len; i++) { + for (uint32_t j = 0; j < bone_count; j++) { + data[j] = CLAMP(src[i * bone_count + j] * 65535, 0, 65535); + } - memcpy(&sw[p_offsets[ai] + i * p_skin_stride], data, 2 * bone_count); + memcpy(&sw[p_offsets[ai] + i * p_skin_stride], data, 2 * bone_count); + } } } - } break; case RS::ARRAY_BONES: { ERR_FAIL_COND_V(p_arrays[ai].get_type() != Variant::PACKED_INT32_ARRAY && p_arrays[ai].get_type() != Variant::PACKED_FLOAT32_ARRAY, ERR_INVALID_PARAMETER); diff --git a/servers/rendering_server.h b/servers/rendering_server.h index 28aee1b575..6b789a6121 100644 --- a/servers/rendering_server.h +++ b/servers/rendering_server.h @@ -613,13 +613,13 @@ public: virtual void particles_set_emitting(RID p_particles, bool p_enable) = 0; virtual bool particles_get_emitting(RID p_particles) = 0; virtual void particles_set_amount(RID p_particles, int p_amount) = 0; - virtual void particles_set_lifetime(RID p_particles, float p_lifetime) = 0; + virtual void particles_set_lifetime(RID p_particles, double p_lifetime) = 0; virtual void particles_set_one_shot(RID p_particles, bool p_one_shot) = 0; - virtual void particles_set_pre_process_time(RID p_particles, float p_time) = 0; + virtual void particles_set_pre_process_time(RID p_particles, double p_time) = 0; virtual void particles_set_explosiveness_ratio(RID p_particles, float p_ratio) = 0; virtual void particles_set_randomness_ratio(RID p_particles, float p_ratio) = 0; virtual void particles_set_custom_aabb(RID p_particles, const AABB &p_aabb) = 0; - virtual void particles_set_speed_scale(RID p_particles, float p_scale) = 0; + virtual void particles_set_speed_scale(RID p_particles, double p_scale) = 0; virtual void particles_set_use_local_coordinates(RID p_particles, bool p_enable) = 0; virtual void particles_set_process_material(RID p_particles, RID p_material) = 0; virtual void particles_set_fixed_fps(RID p_particles, int p_fps) = 0; @@ -900,8 +900,8 @@ public: virtual void viewport_set_debug_draw(RID p_viewport, ViewportDebugDraw p_draw) = 0; virtual void viewport_set_measure_render_time(RID p_viewport, bool p_enable) = 0; - 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 double viewport_get_measured_render_time_cpu(RID p_viewport) const = 0; + virtual double viewport_get_measured_render_time_gpu(RID p_viewport) const = 0; /* SKY API */ @@ -1436,15 +1436,15 @@ public: struct FrameProfileArea { String name; - float gpu_msec; - float cpu_msec; + double gpu_msec; + double cpu_msec; }; virtual void set_frame_profiling_enabled(bool p_enable) = 0; virtual Vector<FrameProfileArea> get_frame_profile() = 0; virtual uint64_t get_frame_profile_frame() = 0; - virtual float get_frame_setup_time_cpu() const = 0; + virtual double get_frame_setup_time_cpu() const = 0; virtual void gi_set_use_half_resolution(bool p_enable) = 0; @@ -1457,7 +1457,7 @@ public: virtual void sdfgi_set_debug_probe_select(const Vector3 &p_position, const Vector3 &p_dir) = 0; - virtual RID make_sphere_mesh(int p_lats, int p_lons, float p_radius); + virtual RID make_sphere_mesh(int p_lats, int p_lons, real_t p_radius); virtual void mesh_add_surface_from_mesh_data(RID p_mesh, const Geometry3D::MeshData &p_mesh_data); virtual void mesh_add_surface_from_planes(RID p_mesh, const Vector<Plane> &p_planes); diff --git a/servers/text_server.cpp b/servers/text_server.cpp index 1491368109..e32ea9cdea 100644 --- a/servers/text_server.cpp +++ b/servers/text_server.cpp @@ -314,8 +314,8 @@ void TextServer::_bind_methods() { ClassDB::bind_method(D_METHOD("shaped_text_get_preserve_control", "shaped"), &TextServer::shaped_text_get_preserve_control); ClassDB::bind_method(D_METHOD("shaped_text_add_string", "shaped", "text", "fonts", "size", "opentype_features", "language"), &TextServer::shaped_text_add_string, DEFVAL(Dictionary()), DEFVAL("")); - ClassDB::bind_method(D_METHOD("shaped_text_add_object", "shaped", "key", "size", "inline_align", "length"), &TextServer::shaped_text_add_object, DEFVAL(VALIGN_CENTER), DEFVAL(1)); - ClassDB::bind_method(D_METHOD("shaped_text_resize_object", "shaped", "key", "size", "inline_align"), &TextServer::shaped_text_resize_object, DEFVAL(VALIGN_CENTER)); + ClassDB::bind_method(D_METHOD("shaped_text_add_object", "shaped", "key", "size", "inline_align", "length"), &TextServer::shaped_text_add_object, DEFVAL(INLINE_ALIGN_CENTER), DEFVAL(1)); + ClassDB::bind_method(D_METHOD("shaped_text_resize_object", "shaped", "key", "size", "inline_align"), &TextServer::shaped_text_resize_object, DEFVAL(INLINE_ALIGN_CENTER)); ClassDB::bind_method(D_METHOD("shaped_text_substr", "shaped", "start", "length"), &TextServer::shaped_text_substr); ClassDB::bind_method(D_METHOD("shaped_text_get_parent", "shaped"), &TextServer::shaped_text_get_parent); diff --git a/servers/text_server.h b/servers/text_server.h index 4c2ada7fc9..d824504bf4 100644 --- a/servers/text_server.h +++ b/servers/text_server.h @@ -181,7 +181,7 @@ public: struct EmbeddedObject { int pos = 0; - VAlign inline_align = VALIGN_TOP; + InlineAlign inline_align = INLINE_ALIGN_CENTER; Rect2 rect; }; Map<Variant, EmbeddedObject> objects; @@ -332,8 +332,8 @@ public: virtual bool shaped_text_get_preserve_control(RID p_shaped) const = 0; virtual bool shaped_text_add_string(RID p_shaped, const String &p_text, const Vector<RID> &p_fonts, int p_size, const Dictionary &p_opentype_features = Dictionary(), const String &p_language = "") = 0; - virtual bool shaped_text_add_object(RID p_shaped, Variant p_key, const Size2 &p_size, VAlign p_inline_align = VALIGN_CENTER, int p_length = 1) = 0; - virtual bool shaped_text_resize_object(RID p_shaped, Variant p_key, const Size2 &p_size, VAlign p_inline_align = VALIGN_CENTER) = 0; + virtual bool shaped_text_add_object(RID p_shaped, Variant p_key, const Size2 &p_size, InlineAlign p_inline_align = INLINE_ALIGN_CENTER, int p_length = 1) = 0; + virtual bool shaped_text_resize_object(RID p_shaped, Variant p_key, const Size2 &p_size, InlineAlign p_inline_align = INLINE_ALIGN_CENTER) = 0; virtual RID shaped_text_substr(RID p_shaped, int p_start, int p_length) const = 0; // Copy shaped substring (e.g. line break) without reshaping, but correctly reordered, preservers range. virtual RID shaped_text_get_parent(RID p_shaped) const = 0; |