diff options
Diffstat (limited to 'servers/rendering/renderer_rd/shaders')
14 files changed, 1318 insertions, 796 deletions
diff --git a/servers/rendering/renderer_rd/shaders/blit.glsl b/servers/rendering/renderer_rd/shaders/blit.glsl index 967da1e6e4..8051f96738 100644 --- a/servers/rendering/renderer_rd/shaders/blit.glsl +++ b/servers/rendering/renderer_rd/shaders/blit.glsl @@ -5,6 +5,7 @@ #VERSION_DEFINES layout(push_constant, binding = 0, std140) uniform Pos { + vec4 src_rect; vec4 dst_rect; vec2 eye_center; @@ -22,8 +23,8 @@ layout(location = 0) out vec2 uv; 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 = base_arr[gl_VertexIndex]; - vec2 vtx = data.dst_rect.xy + uv * data.dst_rect.zw; + uv = data.src_rect.xy + base_arr[gl_VertexIndex] * data.src_rect.zw; + vec2 vtx = data.dst_rect.xy + base_arr[gl_VertexIndex] * data.dst_rect.zw; gl_Position = vec4(vtx * 2.0 - 1.0, 0.0, 1.0); } @@ -34,6 +35,7 @@ void main() { #VERSION_DEFINES layout(push_constant, binding = 0, std140) uniform Pos { + vec4 src_rect; vec4 dst_rect; vec2 eye_center; diff --git a/servers/rendering/renderer_rd/shaders/canvas.glsl b/servers/rendering/renderer_rd/shaders/canvas.glsl index 2911e8b731..65a621b203 100644 --- a/servers/rendering/renderer_rd/shaders/canvas.glsl +++ b/servers/rendering/renderer_rd/shaders/canvas.glsl @@ -91,7 +91,6 @@ void main() { uint instancing = draw_data.flags & FLAGS_INSTANCING_MASK; #ifdef USE_ATTRIBUTES - if (instancing > 1) { // trails @@ -128,37 +127,37 @@ void main() { vertex = new_vertex; color *= pcolor; - } else #endif // USE_ATTRIBUTES + { + if (instancing == 1) { + uint stride = 2; + { + if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_COLORS)) { + stride += 1; + } + if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_CUSTOM_DATA)) { + stride += 1; + } + } + + uint offset = stride * gl_InstanceIndex; + + mat4 matrix = mat4(transforms.data[offset + 0], transforms.data[offset + 1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0)); + offset += 2; - if (instancing == 1) { - uint stride = 2; - { if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_COLORS)) { - stride += 1; + color *= transforms.data[offset]; + offset += 1; } + if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_CUSTOM_DATA)) { - stride += 1; + instance_custom = transforms.data[offset]; } - } - - uint offset = stride * gl_InstanceIndex; - mat4 matrix = mat4(transforms.data[offset + 0], transforms.data[offset + 1], vec4(0.0, 0.0, 1.0, 0.0), vec4(0.0, 0.0, 0.0, 1.0)); - offset += 2; - - if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_COLORS)) { - color *= transforms.data[offset]; - offset += 1; - } - - if (bool(draw_data.flags & FLAGS_INSTANCING_HAS_CUSTOM_DATA)) { - instance_custom = transforms.data[offset]; + matrix = transpose(matrix); + world_matrix = world_matrix * matrix; } - - matrix = transpose(matrix); - world_matrix = world_matrix * matrix; } #if !defined(USE_ATTRIBUTES) && !defined(USE_PRIMITIVE) diff --git a/servers/rendering/renderer_rd/shaders/fsr_upscale.glsl b/servers/rendering/renderer_rd/shaders/fsr_upscale.glsl new file mode 100644 index 0000000000..4e2ba84033 --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/fsr_upscale.glsl @@ -0,0 +1,173 @@ +/*************************************************************************/ +/* fsr_upscale.glsl */ +/*************************************************************************/ +/* This file is part of: */ +/* GODOT ENGINE */ +/* https://godotengine.org */ +/*************************************************************************/ +/* Copyright (c) 2007-2021 Juan Linietsky, Ariel Manzur. */ +/* Copyright (c) 2014-2021 Godot Engine contributors (cf. AUTHORS.md). */ +/* */ +/* Permission is hereby granted, free of charge, to any person obtaining */ +/* a copy of this software and associated documentation files (the */ +/* "Software"), to deal in the Software without restriction, including */ +/* without limitation the rights to use, copy, modify, merge, publish, */ +/* distribute, sublicense, and/or sell copies of the Software, and to */ +/* permit persons to whom the Software is furnished to do so, subject to */ +/* the following conditions: */ +/* */ +/* The above copyright notice and this permission notice shall be */ +/* included in all copies or substantial portions of the Software. */ +/* */ +/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */ +/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */ +/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/ +/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */ +/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */ +/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */ +/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/*************************************************************************/ + +#[compute] + +#version 450 + +#VERSION_DEFINES + +#define A_GPU +#define A_GLSL + +#ifdef MODE_FSR_UPSCALE_NORMAL + +#define A_HALF + +#endif + +#include "thirdparty/amd-fsr/ffx_a.h" + +layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in; + +layout(rgba16f, set = 1, binding = 0) uniform restrict writeonly image2D fsr_image; +layout(set = 0, binding = 0) uniform sampler2D source_image; + +#define FSR_UPSCALE_PASS_TYPE_EASU 0 +#define FSR_UPSCALE_PASS_TYPE_RCAS 1 + +layout(push_constant, binding = 1, std430) uniform Params { + float resolution_width; + float resolution_height; + float upscaled_width; + float upscaled_height; + float sharpness; + int pass; +} +params; + +AU4 Const0, Const1, Const2, Const3; + +#ifdef MODE_FSR_UPSCALE_FALLBACK + +#define FSR_EASU_F +AF4 FsrEasuRF(AF2 p) { + AF4 res = textureGather(source_image, p, 0); + return res; +} +AF4 FsrEasuGF(AF2 p) { + AF4 res = textureGather(source_image, p, 1); + return res; +} +AF4 FsrEasuBF(AF2 p) { + AF4 res = textureGather(source_image, p, 2); + return res; +} + +#define FSR_RCAS_F +AF4 FsrRcasLoadF(ASU2 p) { + return AF4(texelFetch(source_image, ASU2(p), 0)); +} +void FsrRcasInputF(inout AF1 r, inout AF1 g, inout AF1 b) {} + +#else + +#define FSR_EASU_H +AH4 FsrEasuRH(AF2 p) { + AH4 res = AH4(textureGather(source_image, p, 0)); + return res; +} +AH4 FsrEasuGH(AF2 p) { + AH4 res = AH4(textureGather(source_image, p, 1)); + return res; +} +AH4 FsrEasuBH(AF2 p) { + AH4 res = AH4(textureGather(source_image, p, 2)); + return res; +} + +#define FSR_RCAS_H +AH4 FsrRcasLoadH(ASW2 p) { + return AH4(texelFetch(source_image, ASU2(p), 0)); +} +void FsrRcasInputH(inout AH1 r, inout AH1 g, inout AH1 b) {} + +#endif + +#include "thirdparty/amd-fsr/ffx_fsr1.h" + +void fsr_easu_pass(AU2 pos) { +#ifdef MODE_FSR_UPSCALE_NORMAL + + AH3 Gamma2Color = AH3(0, 0, 0); + FsrEasuH(Gamma2Color, pos, Const0, Const1, Const2, Const3); + imageStore(fsr_image, ASU2(pos), AH4(Gamma2Color, 1)); + +#else + + AF3 Gamma2Color = AF3(0, 0, 0); + FsrEasuF(Gamma2Color, pos, Const0, Const1, Const2, Const3); + imageStore(fsr_image, ASU2(pos), AF4(Gamma2Color, 1)); + +#endif +} + +void fsr_rcas_pass(AU2 pos) { +#ifdef MODE_FSR_UPSCALE_NORMAL + + AH3 Gamma2Color = AH3(0, 0, 0); + FsrRcasH(Gamma2Color.r, Gamma2Color.g, Gamma2Color.b, pos, Const0); + imageStore(fsr_image, ASU2(pos), AH4(Gamma2Color, 1)); + +#else + + AF3 Gamma2Color = AF3(0, 0, 0); + FsrRcasF(Gamma2Color.r, Gamma2Color.g, Gamma2Color.b, pos, Const0); + imageStore(fsr_image, ASU2(pos), AF4(Gamma2Color, 1)); + +#endif +} + +void fsr_pass(AU2 pos) { + if (params.pass == FSR_UPSCALE_PASS_TYPE_EASU) { + fsr_easu_pass(pos); + } else if (params.pass == FSR_UPSCALE_PASS_TYPE_RCAS) { + fsr_rcas_pass(pos); + } +} + +void main() { + // Clang does not like unused functions. If ffx_a.h is included in the binary, clang will throw a fit and not compile so we must configure FSR in this shader + if (params.pass == FSR_UPSCALE_PASS_TYPE_EASU) { + FsrEasuCon(Const0, Const1, Const2, Const3, params.resolution_width, params.resolution_height, params.resolution_width, params.resolution_height, params.upscaled_width, params.upscaled_height); + } else if (params.pass == FSR_UPSCALE_PASS_TYPE_RCAS) { + FsrRcasCon(Const0, params.sharpness); + } + + AU2 gxy = ARmp8x8(gl_LocalInvocationID.x) + AU2(gl_WorkGroupID.x << 4u, gl_WorkGroupID.y << 4u); + + fsr_pass(gxy); + gxy.x += 8u; + fsr_pass(gxy); + gxy.y += 8u; + fsr_pass(gxy); + gxy.x -= 8u; + fsr_pass(gxy); +} diff --git a/servers/rendering/renderer_rd/shaders/gi.glsl b/servers/rendering/renderer_rd/shaders/gi.glsl index 60c881881d..5528ea3659 100644 --- a/servers/rendering/renderer_rd/shaders/gi.glsl +++ b/servers/rendering/renderer_rd/shaders/gi.glsl @@ -97,12 +97,10 @@ layout(push_constant, binding = 0, std430) uniform Params { vec4 proj_info; - vec3 ao_color; uint max_voxel_gi_instances; - bool high_quality_vct; bool orthogonal; - uint pad[2]; + uint pad; mat3x4 cam_rotation; } diff --git a/servers/rendering/renderer_rd/shaders/particles.glsl b/servers/rendering/renderer_rd/shaders/particles.glsl index 9f8410fd8a..328becbc20 100644 --- a/servers/rendering/renderer_rd/shaders/particles.glsl +++ b/servers/rendering/renderer_rd/shaders/particles.glsl @@ -567,11 +567,11 @@ void main() { depth = particle_size - s; const float EPSILON = 0.001; normal = mat3(FRAME.colliders[i].transform) * - normalize( - vec3( - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(EPSILON, 0.0, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(EPSILON, 0.0, 0.0)).r, - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, EPSILON, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, EPSILON, 0.0)).r, - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, 0.0, EPSILON)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, 0.0, EPSILON)).r)); + normalize( + vec3( + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(EPSILON, 0.0, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(EPSILON, 0.0, 0.0)).r, + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, EPSILON, 0.0)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, EPSILON, 0.0)).r, + texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos + vec3(0.0, 0.0, EPSILON)).r - texture(sampler3D(sdf_vec_textures[FRAME.colliders[i].texture_index], material_samplers[SAMPLER_LINEAR_CLAMP]), uvw_pos - vec3(0.0, 0.0, EPSILON)).r)); } } break; diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_aa_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_aa_inc.glsl index 99714b4504..97c913d489 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_aa_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_aa_inc.glsl @@ -2,7 +2,7 @@ float hash_2d(vec2 p) { return fract(1.0e4 * sin(17.0 * p.x + 0.1 * p.y) * - (0.1 + abs(sin(13.0 * p.y + p.x)))); + (0.1 + abs(sin(13.0 * p.y + p.x)))); } float hash_3d(vec3 p) { @@ -29,8 +29,7 @@ float compute_alpha_hash_threshold(vec3 pos, float hash_scale) { vec3 cases = vec3(a_interp * a_interp / (2.0 * min_lerp * (1.0 - min_lerp)), (a_interp - 0.5 * min_lerp) / (1.0 - min_lerp), - 1.0 - ((1.0 - a_interp) * (1.0 - a_interp) / - (2.0 * min_lerp * (1.0 - min_lerp)))); + 1.0 - ((1.0 - a_interp) * (1.0 - a_interp) / (2.0 * min_lerp * (1.0 - min_lerp)))); float alpha_hash_threshold = (lerp_factor < (1.0 - min_lerp)) ? ((lerp_factor < min_lerp) ? cases.x : cases.y) : cases.z; diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl index edbe1031b7..e4628b2d5a 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered.glsl @@ -6,6 +6,8 @@ #include "scene_forward_clustered_inc.glsl" +#define SHADER_IS_SRGB false + /* INPUT ATTRIBS */ layout(location = 0) in vec3 vertex_attrib; @@ -95,7 +97,7 @@ layout(location = 8) out float dp_clip; #endif -layout(location = 9) out flat uint instance_index; +layout(location = 9) out flat uint instance_index_interp; invariant gl_Position; @@ -107,13 +109,15 @@ void main() { color_interp = color_attrib; #endif - instance_index = draw_call.instance_index; + uint instance_index = draw_call.instance_index; bool is_multimesh = bool(instances.data[instance_index].flags & INSTANCE_FLAGS_MULTIMESH); if (!is_multimesh) { instance_index += gl_InstanceIndex; } + instance_index_interp = instance_index; + mat4 world_matrix = instances.data[instance_index].transform; mat3 world_normal_matrix; @@ -356,6 +360,8 @@ void main() { #VERSION_DEFINES +#define SHADER_IS_SRGB false + /* Specialization Constants (Toggles) */ layout(constant_id = 0) const bool sc_use_forward_gi = false; @@ -410,7 +416,7 @@ layout(location = 8) in float dp_clip; #endif -layout(location = 9) in flat uint instance_index; +layout(location = 9) in flat uint instance_index_interp; //defines to keep compatibility with vertex @@ -524,14 +530,14 @@ vec4 fog_process(vec3 vertex) { } } - float fog_amount = 1.0 - exp(min(0.0, vertex.z * scene_data.fog_density)); + float fog_amount = 1.0 - exp(min(0.0, -length(vertex) * scene_data.fog_density)); - if (abs(scene_data.fog_height_density) > 0.001) { + if (abs(scene_data.fog_height_density) >= 0.0001) { float y = (scene_data.camera_matrix * vec4(vertex, 1.0)).y; - float y_dist = scene_data.fog_height - y; + float y_dist = y - scene_data.fog_height; - float vfog_amount = clamp(exp(y_dist * scene_data.fog_height_density), 0.0, 1.0); + float vfog_amount = 1.0 - exp(min(0.0, y_dist * scene_data.fog_height_density)); fog_amount = max(vfog_amount, fog_amount); } @@ -564,6 +570,8 @@ void main() { discard; #endif + uint instance_index = instance_index_interp; + //lay out everything, whathever is unused is optimized away anyway vec3 vertex = vertex_interp; vec3 view = -normalize(vertex_interp); @@ -593,7 +601,7 @@ void main() { float ao = 1.0; float ao_light_affect = 0.0; - float alpha = 1.0; + float alpha = float(instances.data[instance_index].flags >> INSTANCE_FLAGS_FADE_SHIFT) / float(255.0); #if defined(TANGENT_USED) || defined(NORMAL_MAP_USED) || defined(LIGHT_ANISOTROPY_USED) vec3 binormal = normalize(binormal_interp); @@ -687,7 +695,7 @@ void main() { #endif // ALPHA_ANTIALIASING_EDGE_USED #ifdef USE_OPAQUE_PREPASS - if (alpha < opaque_prepass_threshold) { + if (alpha < scene_data.opaque_prepass_threshold) { discard; } #endif // USE_OPAQUE_PREPASS @@ -903,6 +911,7 @@ void main() { if (scene_data.use_reflection_cubemap) { vec3 ref_vec = reflect(-view, normal); + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); ref_vec = scene_data.radiance_inverse_xform * ref_vec; #ifdef USE_RADIANCE_CUBEMAP_ARRAY @@ -915,7 +924,6 @@ 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; } @@ -964,15 +972,15 @@ void main() { const float c4 = 0.886227; const float c5 = 0.247708; ambient_light += (c1 * lightmap_captures.data[index].sh[8].rgb * (wnormal.x * wnormal.x - wnormal.y * wnormal.y) + - c3 * lightmap_captures.data[index].sh[6].rgb * wnormal.z * wnormal.z + - c4 * lightmap_captures.data[index].sh[0].rgb - - c5 * lightmap_captures.data[index].sh[6].rgb + - 2.0 * c1 * lightmap_captures.data[index].sh[4].rgb * wnormal.x * wnormal.y + - 2.0 * c1 * lightmap_captures.data[index].sh[7].rgb * wnormal.x * wnormal.z + - 2.0 * c1 * lightmap_captures.data[index].sh[5].rgb * wnormal.y * wnormal.z + - 2.0 * c2 * lightmap_captures.data[index].sh[3].rgb * wnormal.x + - 2.0 * c2 * lightmap_captures.data[index].sh[1].rgb * wnormal.y + - 2.0 * c2 * lightmap_captures.data[index].sh[2].rgb * wnormal.z); + c3 * lightmap_captures.data[index].sh[6].rgb * wnormal.z * wnormal.z + + c4 * lightmap_captures.data[index].sh[0].rgb - + c5 * lightmap_captures.data[index].sh[6].rgb + + 2.0 * c1 * lightmap_captures.data[index].sh[4].rgb * wnormal.x * wnormal.y + + 2.0 * c1 * lightmap_captures.data[index].sh[7].rgb * wnormal.x * wnormal.z + + 2.0 * c1 * lightmap_captures.data[index].sh[5].rgb * wnormal.y * wnormal.z + + 2.0 * c2 * lightmap_captures.data[index].sh[3].rgb * wnormal.x + + 2.0 * c2 * lightmap_captures.data[index].sh[1].rgb * wnormal.y + + 2.0 * c2 * lightmap_captures.data[index].sh[2].rgb * wnormal.z); } else if (bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_LIGHTMAP)) { // has actual lightmap bool uses_sh = bool(instances.data[instance_index].flags & INSTANCE_FLAGS_USE_SH_LIGHTMAP); @@ -1249,9 +1257,10 @@ void main() { // LIGHTING #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) - { //directional light + { // Directional light. - // Do shadow and lighting in two passes to reduce register pressure + // Do shadow and lighting in two passes to reduce register pressure. +#ifndef SHADOWS_DISABLED uint shadow0 = 0; uint shadow1 = 0; @@ -1270,21 +1279,21 @@ void main() { float shadow = 1.0; - //version with soft shadows, more expensive if (directional_lights.data[i].shadow_enabled) { - if (sc_use_directional_soft_shadows && directional_lights.data[i].softshadow_angle > 0) { - float depth_z = -vertex.z; - - vec3 shadow_color = vec3(0.0); - vec3 light_dir = directional_lights.data[i].direction; + float depth_z = -vertex.z; + vec3 light_dir = directional_lights.data[i].direction; + vec3 base_normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))); -#define BIAS_FUNC(m_var, m_idx) \ - m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ - vec3 normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))) * directional_lights.data[i].shadow_normal_bias[m_idx]; \ - normal_bias -= light_dir * dot(light_dir, normal_bias); \ +#define BIAS_FUNC(m_var, m_idx) \ + m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ + vec3 normal_bias = base_normal_bias * directional_lights.data[i].shadow_normal_bias[m_idx]; \ + normal_bias -= light_dir * dot(light_dir, normal_bias); \ m_var.xyz += normal_bias; - uint blend_index = 0; + //version with soft shadows, more expensive + if (sc_use_directional_soft_shadows && directional_lights.data[i].softshadow_angle > 0) { + uint blend_count = 0; + const uint blend_max = directional_lights.data[i].blend_splits ? 2 : 1; if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { vec4 v = vec4(vertex, 1.0); @@ -1299,10 +1308,10 @@ void main() { float test_radius = (range_pos - range_begin) * directional_lights.data[i].softshadow_angle; vec2 tex_scale = directional_lights.data[i].uv_scale1 * test_radius; shadow = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - blend_index++; + blend_count++; } - if (blend_index < 2 && depth_z < directional_lights.data[i].shadow_split_offsets.y) { + if (blend_count < blend_max && depth_z < directional_lights.data[i].shadow_split_offsets.y) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 1) @@ -1316,7 +1325,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale2 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1324,10 +1333,10 @@ void main() { shadow = mix(shadow, s, blend); } - blend_index++; + blend_count++; } - if (blend_index < 2 && depth_z < directional_lights.data[i].shadow_split_offsets.z) { + if (blend_count < blend_max && depth_z < directional_lights.data[i].shadow_split_offsets.z) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 2) @@ -1341,7 +1350,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale3 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1349,10 +1358,10 @@ void main() { shadow = mix(shadow, s, blend); } - blend_index++; + blend_count++; } - if (blend_index < 2) { + if (blend_count < blend_max) { vec4 v = vec4(vertex, 1.0); BIAS_FUNC(v, 3) @@ -1366,7 +1375,7 @@ void main() { vec2 tex_scale = directional_lights.data[i].uv_scale4 * test_radius; float s = sample_directional_soft_shadow(directional_shadow_atlas, pssm_coord.xyz, tex_scale * directional_lights.data[i].soft_shadow_scale); - if (blend_index == 0) { + if (blend_count == 0) { shadow = s; } else { //blend @@ -1375,21 +1384,9 @@ void main() { } } -#undef BIAS_FUNC } else { //no soft shadows - float depth_z = -vertex.z; - vec4 pssm_coord; - vec3 light_dir = directional_lights.data[i].direction; - vec3 base_normal_bias = normalize(normal_interp) * (1.0 - max(0.0, dot(light_dir, -normalize(normal_interp)))); - -#define BIAS_FUNC(m_var, m_idx) \ - m_var.xyz += light_dir * directional_lights.data[i].shadow_bias[m_idx]; \ - vec3 normal_bias = base_normal_bias * directional_lights.data[i].shadow_normal_bias[m_idx]; \ - normal_bias -= light_dir * dot(light_dir, normal_bias); \ - m_var.xyz += normal_bias; - if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { vec4 v = vec4(vertex, 1.0); @@ -1448,11 +1445,11 @@ void main() { float shadow2 = sample_directional_pcf_shadow(directional_shadow_atlas, scene_data.directional_shadow_pixel_size * directional_lights.data[i].soft_shadow_scale, pssm_coord); shadow = mix(shadow, shadow2, pssm_blend); } + } - shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance + shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, vertex.z)); //done with negative values for performance #undef BIAS_FUNC - } } // shadows if (i < 4) { @@ -1461,6 +1458,7 @@ void main() { shadow1 |= uint(clamp(shadow * 255.0, 0.0, 255.0)) << ((i - 4) * 8); } } +#endif // SHADOWS_DISABLED for (uint i = 0; i < 8; i++) { if (i >= scene_data.directional_light_count) { @@ -1523,18 +1521,19 @@ void main() { #endif float shadow = 1.0; - +#ifndef SHADOWS_DISABLED if (i < 4) { shadow = float(shadow0 >> (i * 8) & 0xFF) / 255.0; } else { shadow = float(shadow1 >> ((i - 4) * 8) & 0xFF) / 255.0; } +#endif blur_shadow(shadow); float size_A = sc_use_light_soft_shadows ? directional_lights.data[i].size : 0.0; - light_compute(normal, directional_lights.data[i].direction, normalize(view), size_A, directional_lights.data[i].color * directional_lights.data[i].energy, shadow, f0, orms, 1.0, + light_compute(normal, directional_lights.data[i].direction, normalize(view), size_A, directional_lights.data[i].color * directional_lights.data[i].energy, shadow, f0, orms, 1.0, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1545,7 +1544,7 @@ void main() { transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim, rim_tint, albedo, + rim, rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1553,9 +1552,6 @@ void main() { #ifdef LIGHT_ANISOTROPY_USED binormal, tangent, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } @@ -1608,7 +1604,7 @@ void main() { shadow = blur_shadow(shadow); - light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, + light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1620,7 +1616,6 @@ void main() { #ifdef LIGHT_RIM_USED rim, rim_tint, - albedo, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1628,9 +1623,6 @@ void main() { #ifdef LIGHT_ANISOTROPY_USED tangent, binormal, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } } @@ -1684,7 +1676,7 @@ void main() { shadow = blur_shadow(shadow); - light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, + light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1696,7 +1688,6 @@ void main() { #ifdef LIGHT_RIM_USED rim, rim_tint, - albedo, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1704,9 +1695,6 @@ void main() { #ifdef LIGHT_ANISOTROPY_USED tangent, binormal, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } } @@ -1723,7 +1711,7 @@ void main() { #ifdef USE_OPAQUE_PREPASS - if (alpha < opaque_prepass_threshold) { + if (alpha < scene_data.opaque_prepass_threshold) { discard; } @@ -1769,7 +1757,11 @@ void main() { } } +#ifdef MOLTENVK_USED + imageStore(geom_facing_grid, grid_pos, uvec4(imageLoad(geom_facing_grid, grid_pos).r | facing_bits)); //store facing bits +#else imageAtomicOr(geom_facing_grid, grid_pos, facing_bits); //store facing bits +#endif if (length(emission) > 0.001) { float lumas[6]; @@ -1909,7 +1901,6 @@ void main() { // Draw "fixed" fog before volumetric fog to ensure volumetric fog can appear in front of the sky. frag_color.rgb = mix(frag_color.rgb, fog.rgb, fog.a); - ; #endif //MODE_MULTIPLE_RENDER_TARGETS diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl index b53bf6a6d4..be29cf4f58 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_clustered_inc.glsl @@ -68,6 +68,7 @@ layout(set = 0, binding = 4) uniform sampler light_projector_sampler; #define INSTANCE_FLAGS_MULTIMESH_HAS_COLOR (1 << 14) #define INSTANCE_FLAGS_MULTIMESH_HAS_CUSTOM_DATA (1 << 15) #define INSTANCE_FLAGS_PARTICLE_TRAIL_SHIFT 16 +#define INSTANCE_FLAGS_FADE_SHIFT 24 //3 bits of stride #define INSTANCE_FLAGS_PARTICLE_TRAIL_MASK 0xFF @@ -207,9 +208,8 @@ layout(set = 1, binding = 0, std140) uniform SceneData { float roughness_limiter_amount; float roughness_limiter_limit; - uvec2 roughness_limiter_pad; - - vec4 ao_color; + float opaque_prepass_threshold; + uint roughness_limiter_pad; mat4 sdf_to_bounds; diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl index f3db4abe3b..d22f936a35 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_lights_inc.glsl @@ -73,7 +73,7 @@ vec3 F0(float metallic, float specular, vec3 albedo) { return mix(vec3(dielectric), albedo, vec3(metallic)); } -void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float attenuation, vec3 f0, uint orms, float specular_amount, +void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float attenuation, vec3 f0, uint orms, float specular_amount, vec3 albedo, inout float alpha, #ifdef LIGHT_BACKLIGHT_USED vec3 backlight, #endif @@ -84,7 +84,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float transmittance_z, #endif #ifdef LIGHT_RIM_USED - float rim, float rim_tint, vec3 rim_color, + float rim, float rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED float clearcoat, float clearcoat_gloss, @@ -92,11 +92,13 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #ifdef LIGHT_ANISOTROPY_USED vec3 B, vec3 T, float anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - inout float alpha, -#endif inout vec3 diffuse_light, inout vec3 specular_light) { + vec4 orms_unpacked = unpackUnorm4x8(orms); + + float roughness = orms_unpacked.y; + float metallic = orms_unpacked.z; + #if defined(LIGHT_CODE_USED) // light is written by the light shader @@ -125,9 +127,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float cLdotH = clamp(A + dot(L, H), 0.0, 1.0); #endif - float metallic = unpackUnorm4x8(orms).z; if (metallic < 1.0) { - float roughness = unpackUnorm4x8(orms).y; float diffuse_brdf_NL; // BRDF times N.L for calculating diffuse radiance #if defined(DIFFUSE_LAMBERT_WRAP) @@ -168,7 +168,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #if defined(LIGHT_RIM_USED) float rim_light = pow(max(0.0, 1.0 - cNdotV), max(0.0, (1.0 - roughness) * 16.0)); - diffuse_light += rim_light * rim * mix(vec3(1.0), rim_color, rim_tint) * light_color; + diffuse_light += rim_light * rim * mix(vec3(1.0), albedo, rim_tint) * light_color; #endif #ifdef LIGHT_TRANSMITTANCE_USED @@ -179,11 +179,11 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float d = scale * abs(transmittance_z); float dd = -d * d; vec3 profile = vec3(0.233, 0.455, 0.649) * exp(dd / 0.0064) + - vec3(0.1, 0.336, 0.344) * exp(dd / 0.0484) + - vec3(0.118, 0.198, 0.0) * exp(dd / 0.187) + - vec3(0.113, 0.007, 0.007) * exp(dd / 0.567) + - vec3(0.358, 0.004, 0.0) * exp(dd / 1.99) + - vec3(0.078, 0.0, 0.0) * exp(dd / 7.41); + vec3(0.1, 0.336, 0.344) * exp(dd / 0.0484) + + vec3(0.118, 0.198, 0.0) * exp(dd / 0.187) + + vec3(0.113, 0.007, 0.007) * exp(dd / 0.567) + + vec3(0.358, 0.004, 0.0) * exp(dd / 1.99) + + vec3(0.078, 0.0, 0.0) * exp(dd / 7.41); diffuse_light += profile * transmittance_color.a * light_color * clamp(transmittance_boost - NdotL, 0.0, 1.0) * (1.0 / M_PI); #else @@ -199,7 +199,6 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #endif //LIGHT_TRANSMITTANCE_USED } - float roughness = unpackUnorm4x8(orms).y; if (roughness > 0.0) { // FIXME: roughness == 0 should not disable specular light entirely // D @@ -211,7 +210,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float blinn = pow(cNdotH, shininess); blinn *= (shininess + 2.0) * (1.0 / (8.0 * M_PI)); - specular_light += light_color * attenuation * specular_amount * blinn * f0 * unpackUnorm4x8(orms).w; + specular_light += light_color * attenuation * specular_amount * blinn * f0 * orms_unpacked.w; #elif defined(SPECULAR_PHONG) @@ -221,7 +220,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte float phong = pow(cRdotV, shininess); phong *= (shininess + 1.0) * (1.0 / (8.0 * M_PI)); - specular_light += light_color * attenuation * specular_amount * phong * f0 * unpackUnorm4x8(orms).w; + specular_light += light_color * attenuation * specular_amount * phong * f0 * orms_unpacked.w; #elif defined(SPECULAR_TOON) @@ -285,7 +284,7 @@ void light_compute(vec3 N, vec3 L, vec3 V, float A, vec3 light_color, float atte #endif //defined(LIGHT_CODE_USED) } -#ifndef USE_NO_SHADOWS +#ifndef SHADOWS_DISABLED // Interleaved Gradient Noise // https://www.iryoku.com/next-generation-post-processing-in-call-of-duty-advanced-warfare @@ -299,7 +298,7 @@ float sample_directional_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, ve float depth = coord.z; //if only one sample is taken, take it from the center - if (sc_directional_soft_shadow_samples == 1) { + if (sc_directional_soft_shadow_samples == 0) { return textureProj(sampler2DShadow(shadow, shadow_sampler), vec4(pos, depth, 1.0)); } @@ -325,7 +324,7 @@ float sample_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec3 coord) { float depth = coord.z; //if only one sample is taken, take it from the center - if (sc_soft_shadow_samples == 1) { + if (sc_soft_shadow_samples == 0) { return textureProj(sampler2DShadow(shadow, shadow_sampler), vec4(pos, depth, 1.0)); } @@ -348,7 +347,7 @@ float sample_pcf_shadow(texture2D shadow, vec2 shadow_pixel_size, vec3 coord) { float sample_omni_pcf_shadow(texture2D shadow, float blur_scale, vec2 coord, vec4 uv_rect, vec2 flip_offset, float depth) { //if only one sample is taken, take it from the center - if (sc_soft_shadow_samples == 1) { + if (sc_soft_shadow_samples == 0) { vec2 pos = coord * 0.5 + 0.5; pos = uv_rect.xy + pos * uv_rect.zw; return textureProj(sampler2DShadow(shadow, shadow_sampler), vec4(pos, depth, 1.0)); @@ -431,7 +430,7 @@ float sample_directional_soft_shadow(texture2D shadow, vec3 pssm_coord, vec2 tex } } -#endif //USE_NO_SHADOWS +#endif // SHADOWS_DISABLED float get_omni_attenuation(float distance, float inv_range, float decay) { float nd = distance * inv_range; @@ -443,7 +442,7 @@ float get_omni_attenuation(float distance, float inv_range, float decay) { } float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { -#ifndef USE_NO_SHADOWS +#ifndef SHADOWS_DISABLED if (omni_lights.data[idx].shadow_enabled) { // there is a shadowmap vec2 texel_size = scene_data.shadow_atlas_pixel_size; @@ -575,7 +574,7 @@ float light_process_omni_shadow(uint idx, vec3 vertex, vec3 normal) { return 1.0; } -void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, +void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, vec3 albedo, inout float alpha, #ifdef LIGHT_BACKLIGHT_USED vec3 backlight, #endif @@ -585,7 +584,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v float transmittance_boost, #endif #ifdef LIGHT_RIM_USED - float rim, float rim_tint, vec3 rim_color, + float rim, float rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED float clearcoat, float clearcoat_gloss, @@ -593,9 +592,6 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED vec3 binormal, vec3 tangent, float anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - inout float alpha, -#endif inout vec3 diffuse_light, inout vec3 specular_light) { vec3 light_rel_vec = omni_lights.data[idx].position - vertex; float light_length = length(light_rel_vec); @@ -701,7 +697,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v light_attenuation *= shadow; - light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color, light_attenuation, f0, orms, omni_lights.data[idx].specular_amount, + light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color, light_attenuation, f0, orms, omni_lights.data[idx].specular_amount, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -712,7 +708,7 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim * omni_attenuation, rim_tint, rim_color, + rim * omni_attenuation, rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -720,15 +716,12 @@ void light_process_omni(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED binormal, tangent, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } float light_process_spot_shadow(uint idx, vec3 vertex, vec3 normal) { -#ifndef USE_NO_SHADOWS +#ifndef SHADOWS_DISABLED if (spot_lights.data[idx].shadow_enabled) { vec3 light_rel_vec = spot_lights.data[idx].position - vertex; float light_length = length(light_rel_vec); @@ -804,7 +797,7 @@ float light_process_spot_shadow(uint idx, vec3 vertex, vec3 normal) { return shadow; } -#endif //USE_NO_SHADOWS +#endif // SHADOWS_DISABLED return 1.0; } @@ -821,7 +814,7 @@ vec2 normal_to_panorama(vec3 n) { return panorama_coords; } -void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, +void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 vertex_ddx, vec3 vertex_ddy, vec3 f0, uint orms, float shadow, vec3 albedo, inout float alpha, #ifdef LIGHT_BACKLIGHT_USED vec3 backlight, #endif @@ -831,7 +824,7 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v float transmittance_boost, #endif #ifdef LIGHT_RIM_USED - float rim, float rim_tint, vec3 rim_color, + float rim, float rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED float clearcoat, float clearcoat_gloss, @@ -839,9 +832,6 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED vec3 binormal, vec3 tangent, float anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - inout float alpha, -#endif inout vec3 diffuse_light, inout vec3 specular_light) { vec3 light_rel_vec = spot_lights.data[idx].position - vertex; @@ -908,7 +898,7 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v } light_attenuation *= shadow; - light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color, light_attenuation, f0, orms, spot_lights.data[idx].specular_amount, + light_compute(normal, normalize(light_rel_vec), eye_vec, size_A, color, light_attenuation, f0, orms, spot_lights.data[idx].specular_amount, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -919,7 +909,7 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v transmittance_z, #endif #ifdef LIGHT_RIM_USED - rim * spot_attenuation, rim_tint, rim_color, + rim * spot_attenuation, rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -927,9 +917,6 @@ void light_process_spot(uint idx, vec3 vertex, vec3 eye_vec, vec3 normal, vec3 v #ifdef LIGHT_ANISOTROPY_USED binormal, tangent, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl index 518b0a6c7f..e92fbecfd0 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_mobile.glsl @@ -7,6 +7,8 @@ /* Include our forward mobile UBOs definitions etc. */ #include "scene_forward_mobile_inc.glsl" +#define SHADER_IS_SRGB false + /* INPUT ATTRIBS */ layout(location = 0) in vec3 vertex_attrib; @@ -370,6 +372,8 @@ void main() { #VERSION_DEFINES +#define SHADER_IS_SRGB false + /* Specialization Constants */ #if !defined(MODE_RENDER_DEPTH) @@ -550,14 +554,14 @@ vec4 fog_process(vec3 vertex) { } } - float fog_amount = 1.0 - exp(min(0.0, vertex.z * scene_data.fog_density)); + float fog_amount = 1.0 - exp(min(0.0, -length(vertex) * scene_data.fog_density)); - if (abs(scene_data.fog_height_density) > 0.001) { + if (abs(scene_data.fog_height_density) >= 0.0001) { float y = (scene_data.camera_matrix * vec4(vertex, 1.0)).y; - float y_dist = scene_data.fog_height - y; + float y_dist = y - scene_data.fog_height; - float vfog_amount = clamp(exp(y_dist * scene_data.fog_height_density), 0.0, 1.0); + float vfog_amount = 1.0 - exp(min(0.0, y_dist * scene_data.fog_height_density)); fog_amount = max(vfog_amount, fog_amount); } @@ -701,7 +705,7 @@ void main() { #endif // ALPHA_ANTIALIASING_EDGE_USED #ifdef USE_OPAQUE_PREPASS - if (alpha < opaque_prepass_threshold) { + if (alpha < scene_data.opaque_prepass_threshold) { discard; } #endif // USE_OPAQUE_PREPASS @@ -868,6 +872,7 @@ void main() { if (scene_data.use_reflection_cubemap) { vec3 ref_vec = reflect(-view, normal); + float horizon = min(1.0 + dot(ref_vec, normal), 1.0); ref_vec = scene_data.radiance_inverse_xform * ref_vec; #ifdef USE_RADIANCE_CUBEMAP_ARRAY @@ -880,7 +885,6 @@ 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; } @@ -930,15 +934,15 @@ void main() { const float c4 = 0.886227; const float c5 = 0.247708; ambient_light += (c1 * lightmap_captures.data[index].sh[8].rgb * (wnormal.x * wnormal.x - wnormal.y * wnormal.y) + - c3 * lightmap_captures.data[index].sh[6].rgb * wnormal.z * wnormal.z + - c4 * lightmap_captures.data[index].sh[0].rgb - - c5 * lightmap_captures.data[index].sh[6].rgb + - 2.0 * c1 * lightmap_captures.data[index].sh[4].rgb * wnormal.x * wnormal.y + - 2.0 * c1 * lightmap_captures.data[index].sh[7].rgb * wnormal.x * wnormal.z + - 2.0 * c1 * lightmap_captures.data[index].sh[5].rgb * wnormal.y * wnormal.z + - 2.0 * c2 * lightmap_captures.data[index].sh[3].rgb * wnormal.x + - 2.0 * c2 * lightmap_captures.data[index].sh[1].rgb * wnormal.y + - 2.0 * c2 * lightmap_captures.data[index].sh[2].rgb * wnormal.z); + c3 * lightmap_captures.data[index].sh[6].rgb * wnormal.z * wnormal.z + + c4 * lightmap_captures.data[index].sh[0].rgb - + c5 * lightmap_captures.data[index].sh[6].rgb + + 2.0 * c1 * lightmap_captures.data[index].sh[4].rgb * wnormal.x * wnormal.y + + 2.0 * c1 * lightmap_captures.data[index].sh[7].rgb * wnormal.x * wnormal.z + + 2.0 * c1 * lightmap_captures.data[index].sh[5].rgb * wnormal.y * wnormal.z + + 2.0 * c2 * lightmap_captures.data[index].sh[3].rgb * wnormal.x + + 2.0 * c2 * lightmap_captures.data[index].sh[1].rgb * wnormal.y + + 2.0 * c2 * lightmap_captures.data[index].sh[2].rgb * wnormal.z); } else if (bool(draw_call.flags & INSTANCE_FLAGS_USE_LIGHTMAP)) { // has actual lightmap bool uses_sh = bool(draw_call.flags & INSTANCE_FLAGS_USE_SH_LIGHTMAP); @@ -1046,7 +1050,7 @@ void main() { #if !defined(MODE_RENDER_DEPTH) && !defined(MODE_UNSHADED) if (!sc_disable_directional_lights) { //directional light - +#ifndef SHADOWS_DISABLED // Do shadow and lighting in two passes to reduce register pressure uint shadow0 = 0; uint shadow1 = 0; @@ -1322,6 +1326,8 @@ void main() { } } +#endif // SHADOWS_DISABLED + for (uint i = 0; i < 8; i++) { if (i >= scene_data.directional_light_count) { break; @@ -1334,16 +1340,16 @@ void main() { // We're not doing light transmittence float shadow = 1.0; - +#ifndef SHADOWS_DISABLED if (i < 4) { shadow = float(shadow0 >> (i * 8) & 0xFF) / 255.0; } else { shadow = float(shadow1 >> ((i - 4) * 8) & 0xFF) / 255.0; } - +#endif blur_shadow(shadow); - light_compute(normal, directional_lights.data[i].direction, normalize(view), 0.0, directional_lights.data[i].color * directional_lights.data[i].energy, shadow, f0, orms, 1.0, + light_compute(normal, directional_lights.data[i].direction, normalize(view), 0.0, directional_lights.data[i].color * directional_lights.data[i].energy, shadow, f0, orms, 1.0, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1356,7 +1362,7 @@ void main() { #endif */ #ifdef LIGHT_RIM_USED - rim, rim_tint, albedo, + rim, rim_tint, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1367,9 +1373,6 @@ void main() { #ifdef USE_SOFT_SHADOW directional_lights.data[i].size, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } @@ -1393,7 +1396,7 @@ void main() { shadow = blur_shadow(shadow); - light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, + light_process_omni(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1407,7 +1410,6 @@ void main() { #ifdef LIGHT_RIM_USED rim, rim_tint, - albedo, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1415,9 +1417,6 @@ void main() { #ifdef LIGHT_ANISOTROPY_USED tangent, binormal, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } } //omni lights @@ -1441,7 +1440,7 @@ void main() { shadow = blur_shadow(shadow); - light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, + light_process_spot(light_index, vertex, view, normal, vertex_ddx, vertex_ddy, f0, orms, shadow, albedo, alpha, #ifdef LIGHT_BACKLIGHT_USED backlight, #endif @@ -1455,7 +1454,6 @@ void main() { #ifdef LIGHT_RIM_USED rim, rim_tint, - albedo, #endif #ifdef LIGHT_CLEARCOAT_USED clearcoat, clearcoat_gloss, @@ -1463,9 +1461,6 @@ void main() { #ifdef LIGHT_ANISOTROPY_USED tangent, binormal, anisotropy, #endif -#ifdef USE_SHADOW_TO_OPACITY - alpha, -#endif diffuse_light, specular_light); } } //spot lights @@ -1481,7 +1476,7 @@ void main() { #ifdef USE_OPAQUE_PREPASS - if (alpha < opaque_prepass_threshold) { + if (alpha < scene_data.opaque_prepass_threshold) { discard; } diff --git a/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl b/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl index dd8879acb4..a9a4fce82a 100644 --- a/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl +++ b/servers/rendering/renderer_rd/shaders/scene_forward_mobile_inc.glsl @@ -168,9 +168,8 @@ layout(set = 1, binding = 0, std140) uniform SceneData { mediump float roughness_limiter_amount; mediump float roughness_limiter_limit; - uvec2 roughness_limiter_pad; - - mediump vec4 ao_color; + mediump float opaque_prepass_threshold; + uint roughness_limiter_pad; bool fog_enabled; highp float fog_density; diff --git a/servers/rendering/renderer_rd/shaders/tonemap.glsl b/servers/rendering/renderer_rd/shaders/tonemap.glsl index 4411587116..948c6e1e39 100644 --- a/servers/rendering/renderer_rd/shaders/tonemap.glsl +++ b/servers/rendering/renderer_rd/shaders/tonemap.glsl @@ -140,7 +140,7 @@ vec4 texture2D_bicubic(sampler2D tex, vec2 uv, int p_lod) { vec2 p3 = (vec2(iuv.x + h1x, iuv.y + h1y) - vec2(0.5f)) * pixel_size; return (g0(fuv.y) * (g0x * textureLod(tex, p0, lod) + g1x * textureLod(tex, p1, lod))) + - (g1(fuv.y) * (g0x * textureLod(tex, p2, lod) + g1x * textureLod(tex, p3, lod))); + (g1(fuv.y) * (g0x * textureLod(tex, p2, lod) + g1x * textureLod(tex, p3, lod))); } #define GLOW_TEXTURE_SAMPLE(m_tex, m_uv, m_lod) texture2D_bicubic(m_tex, m_uv, m_lod) @@ -169,16 +169,33 @@ vec3 tonemap_filmic(vec3 color, float white) { return color_tonemapped / white_tonemapped; } +// Adapted from https://github.com/TheRealMJP/BakingLab/blob/master/BakingLab/ACES.hlsl +// (MIT License). vec3 tonemap_aces(vec3 color, float white) { - const float exposure_bias = 0.85f; - const float A = 2.51f * exposure_bias * exposure_bias; - const float B = 0.03f * exposure_bias; - const float C = 2.43f * exposure_bias * exposure_bias; - const float D = 0.59f * exposure_bias; - const float E = 0.14f; - - vec3 color_tonemapped = (color * (A * color + B)) / (color * (C * color + D) + E); - float white_tonemapped = (white * (A * white + B)) / (white * (C * white + D) + E); + const float exposure_bias = 1.8f; + const float A = 0.0245786f; + const float B = 0.000090537f; + const float C = 0.983729f; + const float D = 0.432951f; + const float E = 0.238081f; + + // Exposure bias baked into transform to save shader instructions. Equivalent to `color *= exposure_bias` + const mat3 rgb_to_rrt = mat3( + vec3(0.59719f * exposure_bias, 0.35458f * exposure_bias, 0.04823f * exposure_bias), + vec3(0.07600f * exposure_bias, 0.90834f * exposure_bias, 0.01566f * exposure_bias), + vec3(0.02840f * exposure_bias, 0.13383f * exposure_bias, 0.83777f * exposure_bias)); + + const mat3 odt_to_rgb = mat3( + vec3(1.60475f, -0.53108f, -0.07367f), + vec3(-0.10208f, 1.10813f, -0.00605f), + vec3(-0.00327f, -0.07276f, 1.07602f)); + + color *= rgb_to_rrt; + vec3 color_tonemapped = (color * (color + A) - B) / (color * (C * color + D) + E); + color_tonemapped *= odt_to_rgb; + + white *= exposure_bias; + float white_tonemapped = (white * (white + A) - B) / (white * (C * white + D) + E); return color_tonemapped / white_tonemapped; } @@ -200,15 +217,16 @@ vec3 linear_to_srgb(vec3 color) { #define TONEMAPPER_ACES 3 vec3 apply_tonemapping(vec3 color, float white) { // inputs are LINEAR, always outputs clamped [0;1] color - + // Ensure color values passed to tonemappers are positive. + // They can be negative in the case of negative lights, which leads to undesired behavior. if (params.tonemapper == TONEMAPPER_LINEAR) { return color; } else if (params.tonemapper == TONEMAPPER_REINHARD) { - return tonemap_reinhard(color, white); + return tonemap_reinhard(max(vec3(0.0f), color), white); } else if (params.tonemapper == TONEMAPPER_FILMIC) { - return tonemap_filmic(color, white); + return tonemap_filmic(max(vec3(0.0f), color), white); } else { // TONEMAPPER_ACES - return tonemap_aces(color, white); + return tonemap_aces(max(vec3(0.0f), color), white); } } @@ -323,14 +341,14 @@ vec3 do_fxaa(vec3 color, float exposure, vec2 uv_interp) { dir.y = ((lumaNW + lumaSW) - (lumaNE + lumaSE)); float dirReduce = max((lumaNW + lumaNE + lumaSW + lumaSE) * - (0.25 * FXAA_REDUCE_MUL), + (0.25 * FXAA_REDUCE_MUL), FXAA_REDUCE_MIN); float rcpDirMin = 1.0 / (min(abs(dir.x), abs(dir.y)) + dirReduce); dir = min(vec2(FXAA_SPAN_MAX, FXAA_SPAN_MAX), max(vec2(-FXAA_SPAN_MAX, -FXAA_SPAN_MAX), dir * rcpDirMin)) * - params.pixel_size; + params.pixel_size; #ifdef MULTIVIEW vec3 rgbA = 0.5 * exposure * (textureLod(source_color, vec3(uv_interp + dir * (1.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz + textureLod(source_color, vec3(uv_interp + dir * (2.0 / 3.0 - 0.5), ViewIndex), 0.0).xyz) * params.luminance_multiplier; @@ -401,9 +419,7 @@ void main() { color += screen_space_dither(gl_FragCoord.xy); } - // 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 = apply_tonemapping(color, params.white); color = linear_to_srgb(color); // regular linear -> SRGB conversion diff --git a/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl b/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl index f2010222e5..181d3b272f 100644 --- a/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl +++ b/servers/rendering/renderer_rd/shaders/volumetric_fog.glsl @@ -4,219 +4,103 @@ #VERSION_DEFINES -/* Do not use subgroups here, seems there is not much advantage and causes glitches -#if defined(has_GL_KHR_shader_subgroup_ballot) && defined(has_GL_KHR_shader_subgroup_arithmetic) -#extension GL_KHR_shader_subgroup_ballot: enable -#extension GL_KHR_shader_subgroup_arithmetic: enable - -#define USE_SUBGROUPS -#endif -*/ - -#if defined(MODE_FOG) || defined(MODE_FILTER) - -layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; - -#endif - -#if defined(MODE_DENSITY) - layout(local_size_x = 4, local_size_y = 4, local_size_z = 4) in; -#endif +#define SAMPLER_NEAREST_CLAMP 0 +#define SAMPLER_LINEAR_CLAMP 1 +#define SAMPLER_NEAREST_WITH_MIPMAPS_CLAMP 2 +#define SAMPLER_LINEAR_WITH_MIPMAPS_CLAMP 3 +#define SAMPLER_NEAREST_WITH_MIPMAPS_ANISOTROPIC_CLAMP 4 +#define SAMPLER_LINEAR_WITH_MIPMAPS_ANISOTROPIC_CLAMP 5 +#define SAMPLER_NEAREST_REPEAT 6 +#define SAMPLER_LINEAR_REPEAT 7 +#define SAMPLER_NEAREST_WITH_MIPMAPS_REPEAT 8 +#define SAMPLER_LINEAR_WITH_MIPMAPS_REPEAT 9 +#define SAMPLER_NEAREST_WITH_MIPMAPS_ANISOTROPIC_REPEAT 10 +#define SAMPLER_LINEAR_WITH_MIPMAPS_ANISOTROPIC_REPEAT 11 + +#define DENSITY_SCALE 1024.0 #include "cluster_data_inc.glsl" #include "light_data_inc.glsl" #define M_PI 3.14159265359 -layout(set = 0, binding = 1) uniform texture2D shadow_atlas; -layout(set = 0, binding = 2) uniform texture2D directional_shadow_atlas; - -layout(set = 0, binding = 3, std430) restrict readonly buffer OmniLights { - LightData data[]; -} -omni_lights; - -layout(set = 0, binding = 4, std430) restrict readonly buffer SpotLights { - LightData data[]; -} -spot_lights; +layout(set = 0, binding = 1) uniform sampler material_samplers[12]; -layout(set = 0, binding = 5, std140) uniform DirectionalLights { - DirectionalLightData data[MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS]; +layout(set = 0, binding = 2, std430) restrict readonly buffer GlobalVariableData { + vec4 data[]; } -directional_lights; - -layout(set = 0, binding = 6, std430) buffer restrict readonly ClusterBuffer { - uint data[]; -} -cluster_buffer; - -layout(set = 0, binding = 7) uniform sampler linear_sampler; - -#ifdef MODE_DENSITY -layout(rgba16f, set = 0, binding = 8) uniform restrict writeonly image3D density_map; -layout(rgba16f, set = 0, binding = 9) uniform restrict readonly image3D fog_map; //unused -#endif - -#ifdef MODE_FOG -layout(rgba16f, set = 0, binding = 8) uniform restrict readonly image3D density_map; -layout(rgba16f, set = 0, binding = 9) uniform restrict writeonly image3D fog_map; -#endif - -#ifdef MODE_FILTER -layout(rgba16f, set = 0, binding = 8) uniform restrict readonly image3D source_map; -layout(rgba16f, set = 0, binding = 9) uniform restrict writeonly image3D dest_map; -#endif - -layout(set = 0, binding = 10) uniform sampler shadow_sampler; - -#define MAX_VOXEL_GI_INSTANCES 8 - -struct VoxelGIData { - mat4 xform; - vec3 bounds; - float dynamic_range; +global_variables; - float bias; - float normal_bias; - bool blend_ambient; - uint texture_slot; - - float anisotropy_strength; - float ambient_occlusion; - float ambient_occlusion_size; - uint mipmaps; -}; - -layout(set = 0, binding = 11, std140) uniform VoxelGIs { - VoxelGIData data[MAX_VOXEL_GI_INSTANCES]; -} -voxel_gi_instances; - -layout(set = 0, binding = 12) uniform texture3D voxel_gi_textures[MAX_VOXEL_GI_INSTANCES]; - -layout(set = 0, binding = 13) uniform sampler linear_sampler_with_mipmaps; - -#ifdef ENABLE_SDFGI - -// SDFGI Integration on set 1 -#define SDFGI_MAX_CASCADES 8 - -struct SDFVoxelGICascadeData { +layout(push_constant, binding = 0, std430) uniform Params { vec3 position; - float to_probe; - ivec3 probe_world_offset; - float to_cell; // 1/bounds * grid_size -}; - -layout(set = 1, binding = 0, std140) uniform SDFGI { - vec3 grid_size; - uint max_cascades; - - bool use_occlusion; - int probe_axis_size; - float probe_to_uvw; - float normal_bias; + float pad; - vec3 lightprobe_tex_pixel_size; - float energy; + vec3 extents; + float pad2; - vec3 lightprobe_uv_offset; - float y_mult; + ivec3 corner; + uint shape; - vec3 occlusion_clamp; - uint pad3; - - vec3 occlusion_renormalize; - uint pad4; - - vec3 cascade_probe_size; - uint pad5; - - SDFVoxelGICascadeData cascades[SDFGI_MAX_CASCADES]; + mat4 transform; } -sdfgi; - -layout(set = 1, binding = 1) uniform texture2DArray sdfgi_ambient_texture; - -layout(set = 1, binding = 2) uniform texture3D sdfgi_occlusion_texture; +params; -#endif //SDFGI +#ifdef MOLTENVK_USED +layout(set = 1, binding = 1) volatile buffer emissive_only_map_buffer { + uint emissive_only_map[]; +}; +#else +layout(r32ui, set = 1, binding = 1) uniform volatile uimage3D emissive_only_map; +#endif -layout(set = 0, binding = 14, std140) uniform Params { +layout(set = 1, binding = 2, std140) uniform SceneParams { vec2 fog_frustum_size_begin; vec2 fog_frustum_size_end; float fog_frustum_end; - float z_near; - float z_far; - int filter_axis; + float z_near; // + float z_far; // + float time; ivec3 fog_volume_size; - uint directional_light_count; - - vec3 light_color; - float base_density; - - float detail_spread; - float gi_inject; - uint max_voxel_gi_instances; - uint cluster_type_size; + uint directional_light_count; // - vec2 screen_size; - uint cluster_shift; - uint cluster_width; - - uint max_cluster_element_count_div_32; bool use_temporal_reprojection; uint temporal_frame; + float detail_spread; float temporal_blend; - mat3x4 cam_rotation; mat4 to_prev_view; + mat4 transform; } -params; - -layout(set = 0, binding = 15) uniform texture3D prev_density_texture; +scene_params; -float get_depth_at_pos(float cell_depth_size, int z) { - float d = float(z) * cell_depth_size + cell_depth_size * 0.5; //center of voxels - d = pow(d, params.detail_spread); - return params.fog_frustum_end * d; -} - -vec3 hash3f(uvec3 x) { - x = ((x >> 16) ^ x) * 0x45d9f3b; - x = ((x >> 16) ^ x) * 0x45d9f3b; - x = (x >> 16) ^ x; - return vec3(x & 0xFFFFF) / vec3(float(0xFFFFF)); -} - -float get_omni_attenuation(float distance, float inv_range, float decay) { - float nd = distance * inv_range; - nd *= nd; - nd *= nd; // nd^4 - nd = max(1.0 - nd, 0.0); - nd *= nd; // nd^2 - return nd * pow(max(distance, 0.0001), -decay); -} +#ifdef MOLTENVK_USED +layout(set = 1, binding = 3) volatile buffer density_only_map_buffer { + uint density_only_map[]; +}; +layout(set = 1, binding = 4) volatile buffer light_only_map_buffer { + uint light_only_map[]; +}; +#else +layout(r32ui, set = 1, binding = 3) uniform volatile uimage3D density_only_map; +layout(r32ui, set = 1, binding = 4) uniform volatile uimage3D light_only_map; +#endif -void cluster_get_item_range(uint p_offset, out uint item_min, out uint item_max, out uint item_from, out uint item_to) { - uint item_min_max = cluster_buffer.data[p_offset]; - item_min = item_min_max & 0xFFFF; - item_max = item_min_max >> 16; - ; +#ifdef MATERIAL_UNIFORMS_USED +layout(set = 2, binding = 0, std140) uniform MaterialUniforms{ +#MATERIAL_UNIFORMS +} material; +#endif - item_from = item_min >> 5; - item_to = (item_max == 0) ? 0 : ((item_max - 1) >> 5) + 1; //side effect of how it is stored, as item_max 0 means no elements -} +#GLOBALS -uint cluster_get_range_clip_mask(uint i, uint z_min, uint z_max) { - int local_min = clamp(int(z_min) - int(i) * 32, 0, 31); - int mask_width = min(int(z_max) - int(z_min), 32 - local_min); - return bitfieldInsert(uint(0), uint(0xFFFFFFFF), local_min, mask_width); +float get_depth_at_pos(float cell_depth_size, int z) { + float d = float(z) * cell_depth_size + cell_depth_size * 0.5; //center of voxels + d = pow(d, scene_params.detail_spread); + return scene_params.fog_frustum_end * d; } #define TEMPORAL_FRAMES 16 @@ -240,464 +124,167 @@ const vec3 halton_map[TEMPORAL_FRAMES] = vec3[]( vec3(0.03125, 0.59259259, 0.32)); void main() { - vec3 fog_cell_size = 1.0 / vec3(params.fog_volume_size); + vec3 fog_cell_size = 1.0 / vec3(scene_params.fog_volume_size); -#ifdef MODE_DENSITY - - ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); - if (any(greaterThanEqual(pos, params.fog_volume_size))) { + ivec3 pos = ivec3(gl_GlobalInvocationID.xyz) + params.corner; + if (any(greaterThanEqual(pos, scene_params.fog_volume_size))) { return; //do not compute } +#ifdef MOLTENVK_USED + uint lpos = pos.z * scene_params.fog_volume_size.x * scene_params.fog_volume_size.y + pos.y * scene_params.fog_volume_size.x + pos.x; +#endif vec3 posf = vec3(pos); - //posf += mix(vec3(0.0),vec3(1.0),0.3) * hash3f(uvec3(pos)) * 2.0 - 1.0; - vec3 fog_unit_pos = posf * fog_cell_size + fog_cell_size * 0.5; //center of voxels - - uvec2 screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); - uvec2 cluster_pos = screen_pos >> params.cluster_shift; - uint cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); - //positions in screen are too spread apart, no hopes for optimizing with subgroups - - fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); + fog_unit_pos.z = pow(fog_unit_pos.z, scene_params.detail_spread); vec3 view_pos; - view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(fog_unit_pos.z)); - view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; + view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(scene_params.fog_frustum_size_begin, scene_params.fog_frustum_size_end, vec2(fog_unit_pos.z)); + view_pos.z = -scene_params.fog_frustum_end * fog_unit_pos.z; view_pos.y = -view_pos.y; - vec4 reprojected_density = vec4(0.0); - float reproject_amount = 0.0; - - if (params.use_temporal_reprojection) { - vec3 prev_view = (params.to_prev_view * vec4(view_pos, 1.0)).xyz; + if (scene_params.use_temporal_reprojection) { + vec3 prev_view = (scene_params.to_prev_view * vec4(view_pos, 1.0)).xyz; //undo transform into prev view prev_view.y = -prev_view.y; //z back to unit size - prev_view.z /= -params.fog_frustum_end; + prev_view.z /= -scene_params.fog_frustum_end; //xy back to unit size - prev_view.xy /= mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(prev_view.z)); + prev_view.xy /= mix(scene_params.fog_frustum_size_begin, scene_params.fog_frustum_size_end, vec2(prev_view.z)); prev_view.xy = prev_view.xy * 0.5 + 0.5; //z back to unspread value - prev_view.z = pow(prev_view.z, 1.0 / params.detail_spread); + prev_view.z = pow(prev_view.z, 1.0 / scene_params.detail_spread); if (all(greaterThan(prev_view, vec3(0.0))) && all(lessThan(prev_view, vec3(1.0)))) { //reprojectinon fits - - reprojected_density = textureLod(sampler3D(prev_density_texture, linear_sampler), prev_view, 0.0); - reproject_amount = params.temporal_blend; - // Since we can reproject, now we must jitter the current view pos. // This is done here because cells that can't reproject should not jitter. - fog_unit_pos = posf * fog_cell_size + fog_cell_size * halton_map[params.temporal_frame]; //center of voxels, offset by halton table - - screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); - cluster_pos = screen_pos >> params.cluster_shift; - cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); - //positions in screen are too spread apart, no hopes for optimizing with subgroups + fog_unit_pos = posf * fog_cell_size + fog_cell_size * halton_map[scene_params.temporal_frame]; //center of voxels, offset by halton table + fog_unit_pos.z = pow(fog_unit_pos.z, scene_params.detail_spread); - fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); - - view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(fog_unit_pos.z)); - view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; + view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(scene_params.fog_frustum_size_begin, scene_params.fog_frustum_size_end, vec2(fog_unit_pos.z)); + view_pos.z = -scene_params.fog_frustum_end * fog_unit_pos.z; view_pos.y = -view_pos.y; } } - uint cluster_z = uint(clamp((abs(view_pos.z) / params.z_far) * 32.0, 0.0, 31.0)); - - vec3 total_light = params.light_color; + float density = 0.0; + vec3 emission = vec3(0.0); + vec3 albedo = vec3(0.0); - float total_density = params.base_density; float cell_depth_size = abs(view_pos.z - get_depth_at_pos(fog_cell_size.z, pos.z + 1)); - //compute directional lights - - for (uint i = 0; i < params.directional_light_count; i++) { - vec3 shadow_attenuation = vec3(1.0); - - if (directional_lights.data[i].shadow_enabled) { - float depth_z = -view_pos.z; - - vec4 pssm_coord; - vec3 shadow_color = directional_lights.data[i].shadow_color1.rgb; - vec3 light_dir = directional_lights.data[i].direction; - vec4 v = vec4(view_pos, 1.0); - float z_range; - - if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { - pssm_coord = (directional_lights.data[i].shadow_matrix1 * v); - pssm_coord /= pssm_coord.w; - z_range = directional_lights.data[i].shadow_z_range.x; - - } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { - pssm_coord = (directional_lights.data[i].shadow_matrix2 * v); - pssm_coord /= pssm_coord.w; - z_range = directional_lights.data[i].shadow_z_range.y; - - } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { - pssm_coord = (directional_lights.data[i].shadow_matrix3 * v); - pssm_coord /= pssm_coord.w; - z_range = directional_lights.data[i].shadow_z_range.z; - - } else { - pssm_coord = (directional_lights.data[i].shadow_matrix4 * v); - pssm_coord /= pssm_coord.w; - z_range = directional_lights.data[i].shadow_z_range.w; - } - - float depth = texture(sampler2D(directional_shadow_atlas, linear_sampler), pssm_coord.xy).r; - float shadow = exp(min(0.0, (depth - pssm_coord.z)) * z_range * directional_lights.data[i].shadow_volumetric_fog_fade); - - /* - //float shadow = textureProj(sampler2DShadow(directional_shadow_atlas,shadow_sampler),pssm_coord); - float shadow = 0.0; - for(float xi=-1;xi<=1;xi++) { - for(float yi=-1;yi<=1;yi++) { - vec2 ofs = vec2(xi,yi) * 1.5 * params.directional_shadow_pixel_size; - shadow += textureProj(sampler2DShadow(directional_shadow_atlas,shadow_sampler),pssm_coord + vec4(ofs,0.0,0.0)); - } - - } - - shadow /= 3.0 * 3.0; -*/ - shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, view_pos.z)); //done with negative values for performance - - shadow_attenuation = mix(shadow_color, vec3(1.0), shadow); - } - - total_light += shadow_attenuation * directional_lights.data[i].color * directional_lights.data[i].energy / M_PI; + vec4 world = scene_params.transform * vec4(view_pos, 1.0); + world.xyz /= world.w; + + vec3 uvw = fog_unit_pos; + + vec4 local_pos = params.transform * world; + local_pos.xyz /= local_pos.w; + + float sdf = -1.0; + if (params.shape == 0) { + //Ellipsoid + // https://www.shadertoy.com/view/tdS3DG + float k0 = length(local_pos.xyz / params.extents); + float k1 = length(local_pos.xyz / (params.extents * params.extents)); + sdf = k0 * (k0 - 1.0) / k1; + } else if (params.shape == 1) { + // Box + // https://iquilezles.org/www/articles/distfunctions/distfunctions.htm + vec3 q = abs(local_pos.xyz) - params.extents; + sdf = length(max(q, 0.0)) + min(max(q.x, max(q.y, q.z)), 0.0); } - //compute lights from cluster - - { //omni lights - - uint cluster_omni_offset = cluster_offset; - - uint item_min; - uint item_max; - uint item_from; - uint item_to; - - cluster_get_item_range(cluster_omni_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); - -#ifdef USE_SUBGROUPS - item_from = subgroupBroadcastFirst(subgroupMin(item_from)); - item_to = subgroupBroadcastFirst(subgroupMax(item_to)); + float cull_mask = 1.0; //used to cull cells that do not contribute + if (params.shape <= 1) { +#ifndef SDF_USED + cull_mask = 1.0 - smoothstep(-0.1, 0.0, sdf); #endif + uvw = clamp((local_pos.xyz + params.extents) / (2.0 * params.extents), 0.0, 1.0); + } - for (uint i = item_from; i < item_to; i++) { - uint mask = cluster_buffer.data[cluster_omni_offset + i]; - mask &= cluster_get_range_clip_mask(i, item_min, item_max); -#ifdef USE_SUBGROUPS - uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); + if (cull_mask > 0.0) { + { +#CODE : FOG + } + +#ifdef DENSITY_USED + density *= cull_mask; + if (abs(density) > 0.001) { + int final_density = int(density * DENSITY_SCALE); +#ifdef MOLTENVK_USED + atomicAdd(density_only_map[lpos], uint(final_density)); #else - uint merged_mask = mask; + imageAtomicAdd(density_only_map, pos, uint(final_density)); #endif - while (merged_mask != 0) { - uint bit = findMSB(merged_mask); - merged_mask &= ~(1 << bit); -#ifdef USE_SUBGROUPS - if (((1 << bit) & mask) == 0) { //do not process if not originally here - continue; - } +#ifdef EMISSION_USED + { + emission *= clamp(density, 0.0, 1.0); + emission = clamp(emission, vec3(0.0), vec3(4.0)); + // Scale to fit into R11G11B10 with a range of 0-4 + uvec3 emission_u = uvec3(emission.r * 511.0, emission.g * 511.0, emission.b * 255.0); + // R and G have 11 bits each and B has 10. Then pack them into a 32 bit uint + uint final_emission = emission_u.r << 21 | emission_u.g << 10 | emission_u.b; +#ifdef MOLTENVK_USED + uint prev_emission = atomicAdd(emissive_only_map[lpos], final_emission); +#else + uint prev_emission = imageAtomicAdd(emissive_only_map, pos, final_emission); #endif - uint light_index = 32 * i + bit; - - //if (!bool(omni_omni_lights.data[light_index].mask & draw_call.layer_mask)) { - // continue; //not masked - //} - - vec3 light_pos = omni_lights.data[light_index].position; - float d = distance(omni_lights.data[light_index].position, view_pos); - float shadow_attenuation = 1.0; - - if (d * omni_lights.data[light_index].inv_radius < 1.0) { - float attenuation = get_omni_attenuation(d, omni_lights.data[light_index].inv_radius, omni_lights.data[light_index].attenuation); - - vec3 light = omni_lights.data[light_index].color / M_PI; - - if (omni_lights.data[light_index].shadow_enabled) { - //has shadow - vec4 v = vec4(view_pos, 1.0); - - vec4 splane = (omni_lights.data[light_index].shadow_matrix * v); - float shadow_len = length(splane.xyz); //need to remember shadow len from here - splane.xyz = normalize(splane.xyz); - vec4 clamp_rect = omni_lights.data[light_index].atlas_rect; + // Adding can lead to colors overflowing, so validate + uvec3 prev_emission_u = uvec3(prev_emission >> 21, (prev_emission << 11) >> 21, prev_emission % 1024); + uint add_emission = final_emission + prev_emission; + uvec3 add_emission_u = uvec3(add_emission >> 21, (add_emission << 11) >> 21, add_emission % 1024); - if (splane.z >= 0.0) { - splane.z += 1.0; + bvec3 overflowing = lessThan(add_emission_u, prev_emission_u + emission_u); - clamp_rect.y += clamp_rect.w; - - } else { - splane.z = 1.0 - splane.z; - } - - splane.xy /= splane.z; - - splane.xy = splane.xy * 0.5 + 0.5; - splane.z = shadow_len * omni_lights.data[light_index].inv_radius; - splane.xy = clamp_rect.xy + splane.xy * clamp_rect.zw; - splane.w = 1.0; //needed? i think it should be 1 already - - float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; - - shadow_attenuation = exp(min(0.0, (depth - splane.z)) / omni_lights.data[light_index].inv_radius * omni_lights.data[light_index].shadow_volumetric_fog_fade); - } - total_light += light * attenuation * shadow_attenuation; + if (any(overflowing)) { + uvec3 overflow_factor = mix(uvec3(0), uvec3(2047 << 21, 2047 << 10, 1023), overflowing); + uint force_max = overflow_factor.r | overflow_factor.g | overflow_factor.b; +#ifdef MOLTENVK_USED + atomicOr(emissive_only_map[lpos], force_max); +#else + imageAtomicOr(emissive_only_map, pos, force_max); +#endif } } - } - } - - { //spot lights - - uint cluster_spot_offset = cluster_offset + params.cluster_type_size; - - uint item_min; - uint item_max; - uint item_from; - uint item_to; - - cluster_get_item_range(cluster_spot_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); - -#ifdef USE_SUBGROUPS - item_from = subgroupBroadcastFirst(subgroupMin(item_from)); - item_to = subgroupBroadcastFirst(subgroupMax(item_to)); #endif - - for (uint i = item_from; i < item_to; i++) { - uint mask = cluster_buffer.data[cluster_spot_offset + i]; - mask &= cluster_get_range_clip_mask(i, item_min, item_max); -#ifdef USE_SUBGROUPS - uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#ifdef ALBEDO_USED + { + vec3 scattering = albedo * clamp(density, 0.0, 1.0); + scattering = clamp(scattering, vec3(0.0), vec3(1.0)); + uvec3 scattering_u = uvec3(scattering.r * 2047.0, scattering.g * 2047.0, scattering.b * 1023.0); + // R and G have 11 bits each and B has 10. Then pack them into a 32 bit uint + uint final_scattering = scattering_u.r << 21 | scattering_u.g << 10 | scattering_u.b; +#ifdef MOLTENVK_USED + uint prev_scattering = atomicAdd(light_only_map[lpos], final_scattering); #else - uint merged_mask = mask; -#endif - - while (merged_mask != 0) { - uint bit = findMSB(merged_mask); - merged_mask &= ~(1 << bit); -#ifdef USE_SUBGROUPS - if (((1 << bit) & mask) == 0) { //do not process if not originally here - continue; - } + uint prev_scattering = imageAtomicAdd(light_only_map, pos, final_scattering); #endif - //if (!bool(omni_lights.data[light_index].mask & draw_call.layer_mask)) { - // continue; //not masked - //} - - uint light_index = 32 * i + bit; - - vec3 light_pos = spot_lights.data[light_index].position; - vec3 light_rel_vec = spot_lights.data[light_index].position - view_pos; - float d = length(light_rel_vec); - float shadow_attenuation = 1.0; - - if (d * spot_lights.data[light_index].inv_radius < 1.0) { - float attenuation = get_omni_attenuation(d, spot_lights.data[light_index].inv_radius, spot_lights.data[light_index].attenuation); - - vec3 spot_dir = spot_lights.data[light_index].direction; - float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[light_index].cone_angle); - float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[light_index].cone_angle)); - attenuation *= 1.0 - pow(spot_rim, spot_lights.data[light_index].cone_attenuation); - - vec3 light = spot_lights.data[light_index].color / M_PI; - - if (spot_lights.data[light_index].shadow_enabled) { - //has shadow - vec4 v = vec4(view_pos, 1.0); - - vec4 splane = (spot_lights.data[light_index].shadow_matrix * v); - splane /= splane.w; - - float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; - - shadow_attenuation = exp(min(0.0, (depth - splane.z)) / spot_lights.data[light_index].inv_radius * spot_lights.data[light_index].shadow_volumetric_fog_fade); - } - - total_light += light * attenuation * shadow_attenuation; - } - } - } - } - - vec3 world_pos = mat3(params.cam_rotation) * view_pos; - - for (uint i = 0; i < params.max_voxel_gi_instances; i++) { - vec3 position = (voxel_gi_instances.data[i].xform * vec4(world_pos, 1.0)).xyz; - - //this causes corrupted pixels, i have no idea why.. - if (all(bvec2(all(greaterThanEqual(position, vec3(0.0))), all(lessThan(position, voxel_gi_instances.data[i].bounds))))) { - position /= voxel_gi_instances.data[i].bounds; - - vec4 light = vec4(0.0); - for (uint j = 0; j < voxel_gi_instances.data[i].mipmaps; j++) { - vec4 slight = textureLod(sampler3D(voxel_gi_textures[i], linear_sampler_with_mipmaps), position, float(j)); - float a = (1.0 - light.a); - light += a * slight; - } - - light.rgb *= voxel_gi_instances.data[i].dynamic_range * params.gi_inject; - - total_light += light.rgb; - } - } - - //sdfgi -#ifdef ENABLE_SDFGI + // Adding can lead to colors overflowing, so validate + uvec3 prev_scattering_u = uvec3(prev_scattering >> 21, (prev_scattering << 11) >> 21, prev_scattering % 1024); + uint add_scattering = final_scattering + prev_scattering; + uvec3 add_scattering_u = uvec3(add_scattering >> 21, (add_scattering << 11) >> 21, add_scattering % 1024); - { - float blend = -1.0; - vec3 ambient_total = vec3(0.0); + bvec3 overflowing = lessThan(add_scattering_u, prev_scattering_u + scattering_u); - for (uint i = 0; i < sdfgi.max_cascades; i++) { - vec3 cascade_pos = (world_pos - sdfgi.cascades[i].position) * sdfgi.cascades[i].to_probe; - - if (any(lessThan(cascade_pos, vec3(0.0))) || any(greaterThanEqual(cascade_pos, sdfgi.cascade_probe_size))) { - continue; //skip cascade - } - - vec3 base_pos = floor(cascade_pos); - ivec3 probe_base_pos = ivec3(base_pos); - - vec4 ambient_accum = vec4(0.0); - - ivec3 tex_pos = ivec3(probe_base_pos.xy, int(i)); - tex_pos.x += probe_base_pos.z * sdfgi.probe_axis_size; - - for (uint j = 0; j < 8; j++) { - ivec3 offset = (ivec3(j) >> ivec3(0, 1, 2)) & ivec3(1, 1, 1); - ivec3 probe_posi = probe_base_pos; - probe_posi += offset; - - // Compute weight - - vec3 probe_pos = vec3(probe_posi); - vec3 probe_to_pos = cascade_pos - probe_pos; - - vec3 trilinear = vec3(1.0) - abs(probe_to_pos); - float weight = trilinear.x * trilinear.y * trilinear.z; - - // Compute lightprobe occlusion - - if (sdfgi.use_occlusion) { - ivec3 occ_indexv = abs((sdfgi.cascades[i].probe_world_offset + probe_posi) & ivec3(1, 1, 1)) * ivec3(1, 2, 4); - vec4 occ_mask = mix(vec4(0.0), vec4(1.0), equal(ivec4(occ_indexv.x | occ_indexv.y), ivec4(0, 1, 2, 3))); - - vec3 occ_pos = clamp(cascade_pos, probe_pos - sdfgi.occlusion_clamp, probe_pos + sdfgi.occlusion_clamp) * sdfgi.probe_to_uvw; - occ_pos.z += float(i); - if (occ_indexv.z != 0) { //z bit is on, means index is >=4, so make it switch to the other half of textures - occ_pos.x += 1.0; - } - - occ_pos *= sdfgi.occlusion_renormalize; - float occlusion = dot(textureLod(sampler3D(sdfgi_occlusion_texture, linear_sampler), occ_pos, 0.0), occ_mask); - - weight *= max(occlusion, 0.01); + if (any(overflowing)) { + uvec3 overflow_factor = mix(uvec3(0), uvec3(2047 << 21, 2047 << 10, 1023), overflowing); + uint force_max = overflow_factor.r | overflow_factor.g | overflow_factor.b; +#ifdef MOLTENVK_USED + atomicOr(light_only_map[lpos], force_max); +#else + imageAtomicOr(light_only_map, pos, force_max); +#endif } - - // Compute ambient texture position - - ivec3 uvw = tex_pos; - uvw.xy += offset.xy; - uvw.x += offset.z * sdfgi.probe_axis_size; - - vec3 ambient = texelFetch(sampler2DArray(sdfgi_ambient_texture, linear_sampler), uvw, 0).rgb; - - ambient_accum.rgb += ambient * weight; - ambient_accum.a += weight; - } - - if (ambient_accum.a > 0) { - ambient_accum.rgb /= ambient_accum.a; } - ambient_total = ambient_accum.rgb; - break; +#endif // ALBEDO_USED } - - total_light += ambient_total * params.gi_inject; +#endif // DENSITY_USED } - -#endif - - vec4 final_density = vec4(total_light, total_density); - - final_density = mix(final_density, reprojected_density, reproject_amount); - - imageStore(density_map, pos, final_density); -#endif - -#ifdef MODE_FOG - - ivec3 pos = ivec3(gl_GlobalInvocationID.xy, 0); - - if (any(greaterThanEqual(pos, params.fog_volume_size))) { - return; //do not compute - } - - vec4 fog_accum = vec4(0.0); - float prev_z = 0.0; - - float t = 1.0; - - for (int i = 0; i < params.fog_volume_size.z; i++) { - //compute fog position - ivec3 fog_pos = pos + ivec3(0, 0, i); - //get fog value - vec4 fog = imageLoad(density_map, fog_pos); - - //get depth at cell pos - float z = get_depth_at_pos(fog_cell_size.z, i); - //get distance from previous pos - float d = abs(prev_z - z); - //compute exinction based on beer's - float extinction = t * exp(-d * fog.a); - //compute alpha based on different of extinctions - float alpha = t - extinction; - //update extinction - t = extinction; - - fog_accum += vec4(fog.rgb * alpha, alpha); - prev_z = z; - - vec4 fog_value; - - if (fog_accum.a > 0.0) { - fog_value = vec4(fog_accum.rgb / fog_accum.a, 1.0 - t); - } else { - fog_value = vec4(0.0); - } - - imageStore(fog_map, fog_pos, fog_value); - } - -#endif - -#ifdef MODE_FILTER - - ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); - - const float gauss[7] = float[](0.071303, 0.131514, 0.189879, 0.214607, 0.189879, 0.131514, 0.071303); - - const ivec3 filter_dir[3] = ivec3[](ivec3(1, 0, 0), ivec3(0, 1, 0), ivec3(0, 0, 1)); - ivec3 offset = filter_dir[params.filter_axis]; - - vec4 accum = vec4(0.0); - for (int i = -3; i <= 3; i++) { - accum += imageLoad(source_map, clamp(pos + offset * i, ivec3(0), params.fog_volume_size - ivec3(1))) * gauss[i + 3]; - } - - imageStore(dest_map, pos, accum); - -#endif } diff --git a/servers/rendering/renderer_rd/shaders/volumetric_fog_process.glsl b/servers/rendering/renderer_rd/shaders/volumetric_fog_process.glsl new file mode 100644 index 0000000000..747f88960c --- /dev/null +++ b/servers/rendering/renderer_rd/shaders/volumetric_fog_process.glsl @@ -0,0 +1,776 @@ +#[compute] + +#version 450 + +#VERSION_DEFINES + +/* Do not use subgroups here, seems there is not much advantage and causes glitches +#if defined(has_GL_KHR_shader_subgroup_ballot) && defined(has_GL_KHR_shader_subgroup_arithmetic) +#extension GL_KHR_shader_subgroup_ballot: enable +#extension GL_KHR_shader_subgroup_arithmetic: enable + +#define USE_SUBGROUPS +#endif +*/ + +#ifdef MODE_DENSITY +layout(local_size_x = 4, local_size_y = 4, local_size_z = 4) in; +#else +layout(local_size_x = 8, local_size_y = 8, local_size_z = 1) in; +#endif + +#include "cluster_data_inc.glsl" +#include "light_data_inc.glsl" + +#define M_PI 3.14159265359 + +#define DENSITY_SCALE 1024.0 + +layout(set = 0, binding = 1) uniform texture2D shadow_atlas; +layout(set = 0, binding = 2) uniform texture2D directional_shadow_atlas; + +layout(set = 0, binding = 3, std430) restrict readonly buffer OmniLights { + LightData data[]; +} +omni_lights; + +layout(set = 0, binding = 4, std430) restrict readonly buffer SpotLights { + LightData data[]; +} +spot_lights; + +layout(set = 0, binding = 5, std140) uniform DirectionalLights { + DirectionalLightData data[MAX_DIRECTIONAL_LIGHT_DATA_STRUCTS]; +} +directional_lights; + +layout(set = 0, binding = 6, std430) buffer restrict readonly ClusterBuffer { + uint data[]; +} +cluster_buffer; + +layout(set = 0, binding = 7) uniform sampler linear_sampler; + +#ifdef MODE_DENSITY +layout(rgba16f, set = 0, binding = 8) uniform restrict writeonly image3D density_map; +layout(rgba16f, set = 0, binding = 9) uniform restrict readonly image3D fog_map; //unused +#endif + +#ifdef MODE_FOG +layout(rgba16f, set = 0, binding = 8) uniform restrict readonly image3D density_map; +layout(rgba16f, set = 0, binding = 9) uniform restrict writeonly image3D fog_map; +#endif + +#ifdef MODE_COPY +layout(rgba16f, set = 0, binding = 8) uniform restrict readonly image3D source_map; +layout(rgba16f, set = 0, binding = 9) uniform restrict writeonly image3D dest_map; +#endif + +#ifdef MODE_FILTER +layout(rgba16f, set = 0, binding = 8) uniform restrict readonly image3D source_map; +layout(rgba16f, set = 0, binding = 9) uniform restrict writeonly image3D dest_map; +#endif + +layout(set = 0, binding = 10) uniform sampler shadow_sampler; + +#define MAX_VOXEL_GI_INSTANCES 8 + +struct VoxelGIData { + mat4 xform; + vec3 bounds; + float dynamic_range; + + float bias; + float normal_bias; + bool blend_ambient; + uint texture_slot; + + float anisotropy_strength; + float ambient_occlusion; + float ambient_occlusion_size; + uint mipmaps; +}; + +layout(set = 0, binding = 11, std140) uniform VoxelGIs { + VoxelGIData data[MAX_VOXEL_GI_INSTANCES]; +} +voxel_gi_instances; + +layout(set = 0, binding = 12) uniform texture3D voxel_gi_textures[MAX_VOXEL_GI_INSTANCES]; + +layout(set = 0, binding = 13) uniform sampler linear_sampler_with_mipmaps; + +#ifdef ENABLE_SDFGI + +// SDFGI Integration on set 1 +#define SDFGI_MAX_CASCADES 8 + +struct SDFVoxelGICascadeData { + vec3 position; + float to_probe; + ivec3 probe_world_offset; + float to_cell; // 1/bounds * grid_size +}; + +layout(set = 1, binding = 0, std140) uniform SDFGI { + vec3 grid_size; + uint max_cascades; + + bool use_occlusion; + int probe_axis_size; + float probe_to_uvw; + float normal_bias; + + vec3 lightprobe_tex_pixel_size; + float energy; + + vec3 lightprobe_uv_offset; + float y_mult; + + vec3 occlusion_clamp; + uint pad3; + + vec3 occlusion_renormalize; + uint pad4; + + vec3 cascade_probe_size; + uint pad5; + + SDFVoxelGICascadeData cascades[SDFGI_MAX_CASCADES]; +} +sdfgi; + +layout(set = 1, binding = 1) uniform texture2DArray sdfgi_ambient_texture; + +layout(set = 1, binding = 2) uniform texture3D sdfgi_occlusion_texture; + +#endif //SDFGI + +layout(set = 0, binding = 14, std140) uniform Params { + vec2 fog_frustum_size_begin; + vec2 fog_frustum_size_end; + + float fog_frustum_end; + float ambient_inject; + float z_far; + int filter_axis; + + vec3 ambient_color; + float sky_contribution; + + ivec3 fog_volume_size; + uint directional_light_count; + + vec3 base_emission; + float base_density; + + vec3 base_scattering; + float phase_g; + + float detail_spread; + float gi_inject; + uint max_voxel_gi_instances; + uint cluster_type_size; + + vec2 screen_size; + uint cluster_shift; + uint cluster_width; + + uint max_cluster_element_count_div_32; + bool use_temporal_reprojection; + uint temporal_frame; + float temporal_blend; + + mat3x4 cam_rotation; + mat4 to_prev_view; + + mat3 radiance_inverse_xform; +} +params; +#ifndef MODE_COPY +layout(set = 0, binding = 15) uniform texture3D prev_density_texture; + +#ifdef MOLTENVK_USED +layout(set = 0, binding = 16) buffer density_only_map_buffer { + uint density_only_map[]; +}; +layout(set = 0, binding = 17) buffer light_only_map_buffer { + uint light_only_map[]; +}; +layout(set = 0, binding = 18) buffer emissive_only_map_buffer { + uint emissive_only_map[]; +}; +#else +layout(r32ui, set = 0, binding = 16) uniform uimage3D density_only_map; +layout(r32ui, set = 0, binding = 17) uniform uimage3D light_only_map; +layout(r32ui, set = 0, binding = 18) uniform uimage3D emissive_only_map; +#endif + +#ifdef USE_RADIANCE_CUBEMAP_ARRAY +layout(set = 0, binding = 19) uniform textureCubeArray sky_texture; +#else +layout(set = 0, binding = 19) uniform textureCube sky_texture; +#endif +#endif // MODE_COPY + +float get_depth_at_pos(float cell_depth_size, int z) { + float d = float(z) * cell_depth_size + cell_depth_size * 0.5; //center of voxels + d = pow(d, params.detail_spread); + return params.fog_frustum_end * d; +} + +vec3 hash3f(uvec3 x) { + x = ((x >> 16) ^ x) * 0x45d9f3b; + x = ((x >> 16) ^ x) * 0x45d9f3b; + x = (x >> 16) ^ x; + return vec3(x & 0xFFFFF) / vec3(float(0xFFFFF)); +} + +float get_omni_attenuation(float dist, float inv_range, float decay) { + float nd = dist * inv_range; + nd *= nd; + nd *= nd; // nd^4 + nd = max(1.0 - nd, 0.0); + nd *= nd; // nd^2 + return nd * pow(max(dist, 0.0001), -decay); +} + +void cluster_get_item_range(uint p_offset, out uint item_min, out uint item_max, out uint item_from, out uint item_to) { + uint item_min_max = cluster_buffer.data[p_offset]; + item_min = item_min_max & 0xFFFF; + item_max = item_min_max >> 16; + ; + + item_from = item_min >> 5; + item_to = (item_max == 0) ? 0 : ((item_max - 1) >> 5) + 1; //side effect of how it is stored, as item_max 0 means no elements +} + +uint cluster_get_range_clip_mask(uint i, uint z_min, uint z_max) { + int local_min = clamp(int(z_min) - int(i) * 32, 0, 31); + int mask_width = min(int(z_max) - int(z_min), 32 - local_min); + return bitfieldInsert(uint(0), uint(0xFFFFFFFF), local_min, mask_width); +} + +float henyey_greenstein(float cos_theta, float g) { + const float k = 0.0795774715459; // 1 / (4 * PI) + return k * (1.0 - g * g) / (pow(1.0 + g * g - 2.0 * g * cos_theta, 1.5)); +} + +#define TEMPORAL_FRAMES 16 + +const vec3 halton_map[TEMPORAL_FRAMES] = vec3[]( + vec3(0.5, 0.33333333, 0.2), + vec3(0.25, 0.66666667, 0.4), + vec3(0.75, 0.11111111, 0.6), + vec3(0.125, 0.44444444, 0.8), + vec3(0.625, 0.77777778, 0.04), + vec3(0.375, 0.22222222, 0.24), + vec3(0.875, 0.55555556, 0.44), + vec3(0.0625, 0.88888889, 0.64), + vec3(0.5625, 0.03703704, 0.84), + vec3(0.3125, 0.37037037, 0.08), + vec3(0.8125, 0.7037037, 0.28), + vec3(0.1875, 0.14814815, 0.48), + vec3(0.6875, 0.48148148, 0.68), + vec3(0.4375, 0.81481481, 0.88), + vec3(0.9375, 0.25925926, 0.12), + vec3(0.03125, 0.59259259, 0.32)); + +void main() { + vec3 fog_cell_size = 1.0 / vec3(params.fog_volume_size); + +#ifdef MODE_DENSITY + + ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); + if (any(greaterThanEqual(pos, params.fog_volume_size))) { + return; //do not compute + } +#ifdef MOLTENVK_USED + uint lpos = pos.z * params.fog_volume_size.x * params.fog_volume_size.y + pos.y * params.fog_volume_size.x + pos.x; +#endif + + vec3 posf = vec3(pos); + + //posf += mix(vec3(0.0),vec3(1.0),0.3) * hash3f(uvec3(pos)) * 2.0 - 1.0; + + vec3 fog_unit_pos = posf * fog_cell_size + fog_cell_size * 0.5; //center of voxels + + uvec2 screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); + uvec2 cluster_pos = screen_pos >> params.cluster_shift; + uint cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); + //positions in screen are too spread apart, no hopes for optimizing with subgroups + + fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); + + vec3 view_pos; + view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(fog_unit_pos.z)); + view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; + view_pos.y = -view_pos.y; + + vec4 reprojected_density = vec4(0.0); + float reproject_amount = 0.0; + + if (params.use_temporal_reprojection) { + vec3 prev_view = (params.to_prev_view * vec4(view_pos, 1.0)).xyz; + //undo transform into prev view + prev_view.y = -prev_view.y; + //z back to unit size + prev_view.z /= -params.fog_frustum_end; + //xy back to unit size + prev_view.xy /= mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(prev_view.z)); + prev_view.xy = prev_view.xy * 0.5 + 0.5; + //z back to unspread value + prev_view.z = pow(prev_view.z, 1.0 / params.detail_spread); + + if (all(greaterThan(prev_view, vec3(0.0))) && all(lessThan(prev_view, vec3(1.0)))) { + //reprojectinon fits + + reprojected_density = textureLod(sampler3D(prev_density_texture, linear_sampler), prev_view, 0.0); + reproject_amount = params.temporal_blend; + + // Since we can reproject, now we must jitter the current view pos. + // This is done here because cells that can't reproject should not jitter. + + fog_unit_pos = posf * fog_cell_size + fog_cell_size * halton_map[params.temporal_frame]; //center of voxels, offset by halton table + + screen_pos = uvec2(fog_unit_pos.xy * params.screen_size); + cluster_pos = screen_pos >> params.cluster_shift; + cluster_offset = (params.cluster_width * cluster_pos.y + cluster_pos.x) * (params.max_cluster_element_count_div_32 + 32); + //positions in screen are too spread apart, no hopes for optimizing with subgroups + + fog_unit_pos.z = pow(fog_unit_pos.z, params.detail_spread); + + view_pos.xy = (fog_unit_pos.xy * 2.0 - 1.0) * mix(params.fog_frustum_size_begin, params.fog_frustum_size_end, vec2(fog_unit_pos.z)); + view_pos.z = -params.fog_frustum_end * fog_unit_pos.z; + view_pos.y = -view_pos.y; + } + } + + uint cluster_z = uint(clamp((abs(view_pos.z) / params.z_far) * 32.0, 0.0, 31.0)); + + vec3 total_light = vec3(0.0); + + float total_density = params.base_density; +#ifdef MOLTENVK_USED + uint local_density = density_only_map[lpos]; +#else + uint local_density = imageLoad(density_only_map, pos).x; +#endif + + total_density += float(int(local_density)) / DENSITY_SCALE; + total_density = max(0.0, total_density); + +#ifdef MOLTENVK_USED + uint scattering_u = light_only_map[lpos]; +#else + uint scattering_u = imageLoad(light_only_map, pos).x; +#endif + vec3 scattering = vec3(scattering_u >> 21, (scattering_u << 11) >> 21, scattering_u % 1024) / vec3(2047.0, 2047.0, 1023.0); + scattering += params.base_scattering * params.base_density; + +#ifdef MOLTENVK_USED + uint emission_u = emissive_only_map[lpos]; +#else + uint emission_u = imageLoad(emissive_only_map, pos).x; +#endif + vec3 emission = vec3(emission_u >> 21, (emission_u << 11) >> 21, emission_u % 1024) / vec3(511.0, 511.0, 255.0); + emission += params.base_emission * params.base_density; + + float cell_depth_size = abs(view_pos.z - get_depth_at_pos(fog_cell_size.z, pos.z + 1)); + //compute directional lights + + if (total_density > 0.001) { + for (uint i = 0; i < params.directional_light_count; i++) { + vec3 shadow_attenuation = vec3(1.0); + + if (directional_lights.data[i].shadow_enabled) { + float depth_z = -view_pos.z; + + vec4 pssm_coord; + vec3 shadow_color = directional_lights.data[i].shadow_color1.rgb; + vec3 light_dir = directional_lights.data[i].direction; + vec4 v = vec4(view_pos, 1.0); + float z_range; + + if (depth_z < directional_lights.data[i].shadow_split_offsets.x) { + pssm_coord = (directional_lights.data[i].shadow_matrix1 * v); + pssm_coord /= pssm_coord.w; + z_range = directional_lights.data[i].shadow_z_range.x; + + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.y) { + pssm_coord = (directional_lights.data[i].shadow_matrix2 * v); + pssm_coord /= pssm_coord.w; + z_range = directional_lights.data[i].shadow_z_range.y; + + } else if (depth_z < directional_lights.data[i].shadow_split_offsets.z) { + pssm_coord = (directional_lights.data[i].shadow_matrix3 * v); + pssm_coord /= pssm_coord.w; + z_range = directional_lights.data[i].shadow_z_range.z; + + } else { + pssm_coord = (directional_lights.data[i].shadow_matrix4 * v); + pssm_coord /= pssm_coord.w; + z_range = directional_lights.data[i].shadow_z_range.w; + } + + float depth = texture(sampler2D(directional_shadow_atlas, linear_sampler), pssm_coord.xy).r; + float shadow = exp(min(0.0, (depth - pssm_coord.z)) * z_range * directional_lights.data[i].shadow_volumetric_fog_fade); + + shadow = mix(shadow, 1.0, smoothstep(directional_lights.data[i].fade_from, directional_lights.data[i].fade_to, view_pos.z)); //done with negative values for performance + + shadow_attenuation = mix(shadow_color, vec3(1.0), shadow); + } + + total_light += shadow_attenuation * directional_lights.data[i].color * directional_lights.data[i].energy * henyey_greenstein(dot(normalize(view_pos), normalize(directional_lights.data[i].direction)), params.phase_g); + } + + // Compute light from sky + if (params.ambient_inject > 0.0) { + vec3 isotropic = vec3(0.0); + vec3 anisotropic = vec3(0.0); + if (params.sky_contribution > 0.0) { + float mip_bias = 2.0 + total_density * (MAX_SKY_LOD - 2.0); // Not physically based, but looks nice + vec3 scatter_direction = (params.radiance_inverse_xform * normalize(view_pos)) * sign(params.phase_g); +#ifdef USE_RADIANCE_CUBEMAP_ARRAY + isotropic = texture(samplerCubeArray(sky_texture, linear_sampler_with_mipmaps), vec4(0.0, 1.0, 0.0, mip_bias)).rgb; + anisotropic = texture(samplerCubeArray(sky_texture, linear_sampler_with_mipmaps), vec4(scatter_direction, mip_bias)).rgb; +#else + isotropic = textureLod(samplerCube(sky_texture, linear_sampler_with_mipmaps), vec3(0.0, 1.0, 0.0), mip_bias).rgb; + anisotropic = textureLod(samplerCube(sky_texture, linear_sampler_with_mipmaps), vec3(scatter_direction), mip_bias).rgb; +#endif //USE_RADIANCE_CUBEMAP_ARRAY + } + + total_light += mix(params.ambient_color, mix(isotropic, anisotropic, abs(params.phase_g)), params.sky_contribution) * params.ambient_inject; + } + + //compute lights from cluster + + { //omni lights + + uint cluster_omni_offset = cluster_offset; + + uint item_min; + uint item_max; + uint item_from; + uint item_to; + + cluster_get_item_range(cluster_omni_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); + +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_omni_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + uint light_index = 32 * i + bit; + + //if (!bool(omni_omni_lights.data[light_index].mask & draw_call.layer_mask)) { + // continue; //not masked + //} + + vec3 light_pos = omni_lights.data[light_index].position; + float d = distance(omni_lights.data[light_index].position, view_pos); + float shadow_attenuation = 1.0; + + if (d * omni_lights.data[light_index].inv_radius < 1.0) { + float attenuation = get_omni_attenuation(d, omni_lights.data[light_index].inv_radius, omni_lights.data[light_index].attenuation); + + vec3 light = omni_lights.data[light_index].color; + + if (omni_lights.data[light_index].shadow_enabled) { + //has shadow + vec4 uv_rect = omni_lights.data[light_index].atlas_rect; + vec2 flip_offset = omni_lights.data[light_index].direction.xy; + + vec3 local_vert = (omni_lights.data[light_index].shadow_matrix * vec4(view_pos, 1.0)).xyz; + + float shadow_len = length(local_vert); //need to remember shadow len from here + vec3 shadow_sample = normalize(local_vert); + + if (shadow_sample.z >= 0.0) { + uv_rect.xy += flip_offset; + } + + shadow_sample.z = 1.0 + abs(shadow_sample.z); + vec3 pos = vec3(shadow_sample.xy / shadow_sample.z, shadow_len - omni_lights.data[light_index].shadow_bias); + pos.z *= omni_lights.data[light_index].inv_radius; + + pos.xy = pos.xy * 0.5 + 0.5; + pos.xy = uv_rect.xy + pos.xy * uv_rect.zw; + + float depth = texture(sampler2D(shadow_atlas, linear_sampler), pos.xy).r; + + shadow_attenuation = exp(min(0.0, (depth - pos.z)) / omni_lights.data[light_index].inv_radius * omni_lights.data[light_index].shadow_volumetric_fog_fade); + } + total_light += light * attenuation * shadow_attenuation * henyey_greenstein(dot(normalize(light_pos - view_pos), normalize(view_pos)), params.phase_g); + } + } + } + } + + { //spot lights + + uint cluster_spot_offset = cluster_offset + params.cluster_type_size; + + uint item_min; + uint item_max; + uint item_from; + uint item_to; + + cluster_get_item_range(cluster_spot_offset + params.max_cluster_element_count_div_32 + cluster_z, item_min, item_max, item_from, item_to); + +#ifdef USE_SUBGROUPS + item_from = subgroupBroadcastFirst(subgroupMin(item_from)); + item_to = subgroupBroadcastFirst(subgroupMax(item_to)); +#endif + + for (uint i = item_from; i < item_to; i++) { + uint mask = cluster_buffer.data[cluster_spot_offset + i]; + mask &= cluster_get_range_clip_mask(i, item_min, item_max); +#ifdef USE_SUBGROUPS + uint merged_mask = subgroupBroadcastFirst(subgroupOr(mask)); +#else + uint merged_mask = mask; +#endif + + while (merged_mask != 0) { + uint bit = findMSB(merged_mask); + merged_mask &= ~(1 << bit); +#ifdef USE_SUBGROUPS + if (((1 << bit) & mask) == 0) { //do not process if not originally here + continue; + } +#endif + + //if (!bool(omni_lights.data[light_index].mask & draw_call.layer_mask)) { + // continue; //not masked + //} + + uint light_index = 32 * i + bit; + + vec3 light_pos = spot_lights.data[light_index].position; + vec3 light_rel_vec = spot_lights.data[light_index].position - view_pos; + float d = length(light_rel_vec); + float shadow_attenuation = 1.0; + + if (d * spot_lights.data[light_index].inv_radius < 1.0) { + float attenuation = get_omni_attenuation(d, spot_lights.data[light_index].inv_radius, spot_lights.data[light_index].attenuation); + + vec3 spot_dir = spot_lights.data[light_index].direction; + float scos = max(dot(-normalize(light_rel_vec), spot_dir), spot_lights.data[light_index].cone_angle); + float spot_rim = max(0.0001, (1.0 - scos) / (1.0 - spot_lights.data[light_index].cone_angle)); + attenuation *= 1.0 - pow(spot_rim, spot_lights.data[light_index].cone_attenuation); + + vec3 light = spot_lights.data[light_index].color; + + if (spot_lights.data[light_index].shadow_enabled) { + //has shadow + vec4 v = vec4(view_pos, 1.0); + + vec4 splane = (spot_lights.data[light_index].shadow_matrix * v); + splane /= splane.w; + + float depth = texture(sampler2D(shadow_atlas, linear_sampler), splane.xy).r; + + shadow_attenuation = exp(min(0.0, (depth - splane.z)) / spot_lights.data[light_index].inv_radius * spot_lights.data[light_index].shadow_volumetric_fog_fade); + } + + total_light += light * attenuation * shadow_attenuation * henyey_greenstein(dot(normalize(light_rel_vec), normalize(view_pos)), params.phase_g); + } + } + } + } + + vec3 world_pos = mat3(params.cam_rotation) * view_pos; + + for (uint i = 0; i < params.max_voxel_gi_instances; i++) { + vec3 position = (voxel_gi_instances.data[i].xform * vec4(world_pos, 1.0)).xyz; + + //this causes corrupted pixels, i have no idea why.. + if (all(bvec2(all(greaterThanEqual(position, vec3(0.0))), all(lessThan(position, voxel_gi_instances.data[i].bounds))))) { + position /= voxel_gi_instances.data[i].bounds; + + vec4 light = vec4(0.0); + for (uint j = 0; j < voxel_gi_instances.data[i].mipmaps; j++) { + vec4 slight = textureLod(sampler3D(voxel_gi_textures[i], linear_sampler_with_mipmaps), position, float(j)); + float a = (1.0 - light.a); + light += a * slight; + } + + light.rgb *= voxel_gi_instances.data[i].dynamic_range * params.gi_inject; + + total_light += light.rgb; + } + } + + //sdfgi +#ifdef ENABLE_SDFGI + + { + float blend = -1.0; + vec3 ambient_total = vec3(0.0); + + for (uint i = 0; i < sdfgi.max_cascades; i++) { + vec3 cascade_pos = (world_pos - sdfgi.cascades[i].position) * sdfgi.cascades[i].to_probe; + + if (any(lessThan(cascade_pos, vec3(0.0))) || any(greaterThanEqual(cascade_pos, sdfgi.cascade_probe_size))) { + continue; //skip cascade + } + + vec3 base_pos = floor(cascade_pos); + ivec3 probe_base_pos = ivec3(base_pos); + + vec4 ambient_accum = vec4(0.0); + + ivec3 tex_pos = ivec3(probe_base_pos.xy, int(i)); + tex_pos.x += probe_base_pos.z * sdfgi.probe_axis_size; + + for (uint j = 0; j < 8; j++) { + ivec3 offset = (ivec3(j) >> ivec3(0, 1, 2)) & ivec3(1, 1, 1); + ivec3 probe_posi = probe_base_pos; + probe_posi += offset; + + // Compute weight + + vec3 probe_pos = vec3(probe_posi); + vec3 probe_to_pos = cascade_pos - probe_pos; + + vec3 trilinear = vec3(1.0) - abs(probe_to_pos); + float weight = trilinear.x * trilinear.y * trilinear.z; + + // Compute lightprobe occlusion + + if (sdfgi.use_occlusion) { + ivec3 occ_indexv = abs((sdfgi.cascades[i].probe_world_offset + probe_posi) & ivec3(1, 1, 1)) * ivec3(1, 2, 4); + vec4 occ_mask = mix(vec4(0.0), vec4(1.0), equal(ivec4(occ_indexv.x | occ_indexv.y), ivec4(0, 1, 2, 3))); + + vec3 occ_pos = clamp(cascade_pos, probe_pos - sdfgi.occlusion_clamp, probe_pos + sdfgi.occlusion_clamp) * sdfgi.probe_to_uvw; + occ_pos.z += float(i); + if (occ_indexv.z != 0) { //z bit is on, means index is >=4, so make it switch to the other half of textures + occ_pos.x += 1.0; + } + + occ_pos *= sdfgi.occlusion_renormalize; + float occlusion = dot(textureLod(sampler3D(sdfgi_occlusion_texture, linear_sampler), occ_pos, 0.0), occ_mask); + + weight *= max(occlusion, 0.01); + } + + // Compute ambient texture position + + ivec3 uvw = tex_pos; + uvw.xy += offset.xy; + uvw.x += offset.z * sdfgi.probe_axis_size; + + vec3 ambient = texelFetch(sampler2DArray(sdfgi_ambient_texture, linear_sampler), uvw, 0).rgb; + + ambient_accum.rgb += ambient * weight; + ambient_accum.a += weight; + } + + if (ambient_accum.a > 0) { + ambient_accum.rgb /= ambient_accum.a; + } + ambient_total = ambient_accum.rgb; + break; + } + + total_light += ambient_total * params.gi_inject; + } + +#endif + } + + vec4 final_density = vec4(total_light * scattering + emission, total_density); + + final_density = mix(final_density, reprojected_density, reproject_amount); + + imageStore(density_map, pos, final_density); +#ifdef MOLTENVK_USED + density_only_map[lpos] = 0; + light_only_map[lpos] = 0; + emissive_only_map[lpos] = 0; +#else + imageStore(density_only_map, pos, uvec4(0)); + imageStore(light_only_map, pos, uvec4(0)); + imageStore(emissive_only_map, pos, uvec4(0)); +#endif +#endif + +#ifdef MODE_FOG + + ivec3 pos = ivec3(gl_GlobalInvocationID.xy, 0); + + if (any(greaterThanEqual(pos, params.fog_volume_size))) { + return; //do not compute + } + + vec4 fog_accum = vec4(0.0, 0.0, 0.0, 1.0); + float prev_z = 0.0; + + for (int i = 0; i < params.fog_volume_size.z; i++) { + //compute fog position + ivec3 fog_pos = pos + ivec3(0, 0, i); + //get fog value + vec4 fog = imageLoad(density_map, fog_pos); + + //get depth at cell pos + float z = get_depth_at_pos(fog_cell_size.z, i); + //get distance from previous pos + float d = abs(prev_z - z); + //compute transmittance using beer's law + float transmittance = exp(-d * fog.a); + + fog_accum.rgb += ((fog.rgb - fog.rgb * transmittance) / max(fog.a, 0.00001)) * fog_accum.a; + fog_accum.a *= transmittance; + + prev_z = z; + + imageStore(fog_map, fog_pos, vec4(fog_accum.rgb, 1.0 - fog_accum.a)); + } + +#endif + +#ifdef MODE_FILTER + + ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); + + const float gauss[7] = float[](0.071303, 0.131514, 0.189879, 0.214607, 0.189879, 0.131514, 0.071303); + + const ivec3 filter_dir[3] = ivec3[](ivec3(1, 0, 0), ivec3(0, 1, 0), ivec3(0, 0, 1)); + ivec3 offset = filter_dir[params.filter_axis]; + + vec4 accum = vec4(0.0); + for (int i = -3; i <= 3; i++) { + accum += imageLoad(source_map, clamp(pos + offset * i, ivec3(0), params.fog_volume_size - ivec3(1))) * gauss[i + 3]; + } + + imageStore(dest_map, pos, accum); + +#endif +#ifdef MODE_COPY + ivec3 pos = ivec3(gl_GlobalInvocationID.xyz); + if (any(greaterThanEqual(pos, params.fog_volume_size))) { + return; //do not compute + } + + imageStore(dest_map, pos, imageLoad(source_map, pos)); + +#endif +} |