Shader helper improvements;

etc/shaders/animator.comp

@@ -1,4 +1,7 @@
 #version 460
+#extension GL_GOOGLE_include_directive : require
+
+#include "lovr.glsl"
 
 layout(local_size_x = 32, local_size_x_id = 0) in;
 
@@ -25,9 +28,9 @@ layout(set = 0, binding = 1) buffer restrict writeonly VertexOut { ModelVertex v
 layout(set = 0, binding = 2) buffer restrict readonly VertexWeights { SkinVertex skin[]; };
 layout(set = 0, binding = 3) uniform JointTransforms { mat4 joints[256]; };
 
-void main() {
+void lovrmain() {
-  if (gl_GlobalInvocationID.x >= vertexCount) return;
+  if (GlobalThreadID.x >= vertexCount) return;
-  uint vertexIndex = baseVertex + gl_GlobalInvocationID.x;
+  uint vertexIndex = baseVertex + GlobalThreadID.x;
 
   uint indices = skin[vertexIndex].indices;
   uint i0 = (indices >> 0) & 0xff;

etc/shaders/equirect.frag

@@ -4,8 +4,6 @@
 
 #include "lovr.glsl"
 
-#define PI 3.141592653589793238462643383
-
 layout(location = 0) in vec3 Direction;
 
 vec4 lovrmain() {

etc/shaders/lovr.glsl

@@ -33,7 +33,6 @@ struct MaterialData {
   float clearcoat;
   float clearcoatRoughness;
   float occlusionStrength;
-  float glowStrength;
   float normalScale;
   float alphaCutoff;
   float pointSize;
@@ -81,7 +80,13 @@ layout(location = 12) in vec2 UV;
 
 // Macros
 #ifdef GL_COMPUTE_SHADER
-//
+#define SubgroupCount gl_NumSubgroups
+#define WorkgroupCount gl_NumWorkGroups
+#define WorkgroupSize gl_WorkGroupSize
+#define WorkgroupID gl_WorkGroupID
+#define GlobalThreadID gl_GlobalInvocationID
+#define LocalThreadID gl_LocalInvocationID
+#define LocalThreadIndex gl_LocalInvocationIndex
 #else
 #define BaseInstance gl_BaseInstance
 #define BaseVertex gl_BaseVertex
@@ -95,36 +100,46 @@ layout(location = 12) in vec2 UV;
 #define PointCoord gl_PointCoord
 #define PointSize gl_PointSize
 #define Position gl_Position
-#define PrimitiveId gl_PrimitiveID
+#define PrimitiveID gl_PrimitiveID
-#define SampleId gl_SampleID
+#define SampleID gl_SampleID
 #define SampleMaskIn gl_SampleMaskIn
 #define SampleMask gl_SampleMask
 #define SamplePosition gl_SamplePosition
 #define VertexIndex gl_VertexIndex
 #define ViewIndex gl_ViewIndex
 
-#define DrawId gl_BaseInstance
+#define DrawID gl_BaseInstance
 #define Projection Cameras[ViewIndex].projection
 #define View Cameras[ViewIndex].view
 #define ViewProjection Cameras[ViewIndex].viewProjection
 #define InverseProjection Cameras[ViewIndex].inverseProjection
-#define Transform Draws[DrawId].transform
+#define Transform Draws[DrawID].transform
-#define NormalMatrix mat3(Draws[DrawId].normalMatrix)
+#define NormalMatrix mat3(Draws[DrawID].normalMatrix)
-#define PassColor Draws[DrawId].color
+#define PassColor Draws[DrawID].color
 
 #define ClipFromLocal (ViewProjection * Transform)
 #define ClipFromWorld (ViewProjection)
 #define ClipFromView (Projection)
 #define ViewFromLocal (View * Transform)
 #define ViewFromWorld (View)
+#define ViewFromClip (InverseProjection)
 #define WorldFromLocal (Transform)
+#define WorldFromView (inverse(View))
+#define WorldFromClip (inverse(ViewProjection))
+
+#define CameraPositionWorld (-View[3].xyz * mat3(View))
 
 #define DefaultPosition (ClipFromLocal * VertexPosition)
 #define DefaultColor (Color * getPixel(ColorTexture, UV))
 #endif
 
+// Constants
+#define PI 3.141592653589793238462643383279502f
+#define TAU (2.f * PI)
+#define PI_2 (.5f * PI)
+
 // Helpers
-#ifdef GL_FRAGMENT_SHADER
+#ifndef GL_COMPUTE_SHADER
 vec4 getPixel(texture2D t, vec2 uv) { return texture(sampler2D(t, Sampler), uv); }
 vec4 getPixel(texture3D t, vec3 uvw) { return texture(sampler3D(t, Sampler), uvw); }
 vec4 getPixel(textureCube t, vec3 dir) { return texture(samplerCube(t, Sampler), dir); }
@@ -156,7 +171,7 @@ void main() {
   PixelColors[0] = lovrmain();
 
   if (applyGlow) {
-    PixelColors[0].rgb += getPixel(GlowTexture, UV).rgb * Material.glow.rgb;
+    PixelColors[0].rgb += getPixel(GlowTexture, UV).rgb * Material.glow.rgb * Material.glow.a;
   }
 
   if (applyAlphaCutoff && PixelColors[0].a <= Material.alphaCutoff) {
@@ -166,7 +181,7 @@ void main() {
 #endif
 
 #ifdef GL_COMPUTE_SHADER
-vec4 lovrmain();
+void lovrmain();
 void main() {
   lovrmain();
 }

src/api/l_data_modelData.c

@@ -607,7 +607,6 @@ static int l_lovrModelDataGetMaterial(lua_State* L) {
   lua_pushnumber(L, material->clearcoat), lua_setfield(L, -2, "clearcoat");
   lua_pushnumber(L, material->clearcoatRoughness), lua_setfield(L, -2, "clearcoatRoughness");
   lua_pushnumber(L, material->occlusionStrength), lua_setfield(L, -2, "occlusionStrength");
-  lua_pushnumber(L, material->glowStrength), lua_setfield(L, -2, "glowStrength");
   lua_pushnumber(L, material->normalScale), lua_setfield(L, -2, "normalScale");
   lua_pushnumber(L, material->alphaCutoff), lua_setfield(L, -2, "alphaCutoff");
   lua_pushnumber(L, material->pointSize), lua_setfield(L, -2, "pointSize");

src/api/l_graphics.c

@@ -771,18 +771,18 @@ static int l_lovrGraphicsGetLimits(lua_State* L) {
   lua_pushinteger(L, limits.clipAndCullDistances), lua_setfield(L, -2, "clipAndCullDistances");
 
   lua_createtable(L, 3, 0);
-  lua_pushinteger(L, limits.computeDispatchCount[0]), lua_rawseti(L, -2, 1);
+  lua_pushinteger(L, limits.workgroupCount[0]), lua_rawseti(L, -2, 1);
-  lua_pushinteger(L, limits.computeDispatchCount[1]), lua_rawseti(L, -2, 2);
+  lua_pushinteger(L, limits.workgroupCount[1]), lua_rawseti(L, -2, 2);
-  lua_pushinteger(L, limits.computeDispatchCount[2]), lua_rawseti(L, -2, 3);
+  lua_pushinteger(L, limits.workgroupCount[2]), lua_rawseti(L, -2, 3);
-  lua_setfield(L, -2, "computeDispatchCount");
+  lua_setfield(L, -2, "workgroupCount");
 
   lua_createtable(L, 3, 0);
-  lua_pushinteger(L, limits.computeWorkgroupSize[0]), lua_rawseti(L, -2, 1);
+  lua_pushinteger(L, limits.workgroupSize[0]), lua_rawseti(L, -2, 1);
-  lua_pushinteger(L, limits.computeWorkgroupSize[1]), lua_rawseti(L, -2, 2);
+  lua_pushinteger(L, limits.workgroupSize[1]), lua_rawseti(L, -2, 2);
-  lua_pushinteger(L, limits.computeWorkgroupSize[2]), lua_rawseti(L, -2, 3);
+  lua_pushinteger(L, limits.workgroupSize[2]), lua_rawseti(L, -2, 3);
-  lua_setfield(L, -2, "computeWorkgroupSize");
+  lua_setfield(L, -2, "workgroupSize");
 
-  lua_pushinteger(L, limits.computeWorkgroupVolume), lua_setfield(L, -2, "computeWorkgroupVolume");
+  lua_pushinteger(L, limits.totalWorkgroupSize), lua_setfield(L, -2, "totalWorkgroupSize");
   lua_pushinteger(L, limits.computeSharedMemory), lua_setfield(L, -2, "computeSharedMemory");
   lua_pushinteger(L, limits.shaderConstantSize), lua_setfield(L, -2, "shaderConstantSize");
   lua_pushinteger(L, limits.indirectDrawCount), lua_setfield(L, -2, "indirectDrawCount");
@@ -1346,10 +1346,6 @@ static int l_lovrGraphicsNewMaterial(lua_State* L) {
     info.data.occlusionStrength = luax_optfloat(L, -1, 1.f);
     lua_pop(L, 1);
 
-    lua_getfield(L, 1, "glowStrength");
-    info.data.glowStrength = luax_optfloat(L, -1, 1.f);
-    lua_pop(L, 1);
-
     lua_getfield(L, 1, "normalScale");
     info.data.normalScale = luax_optfloat(L, -1, 1.f);
     lua_pop(L, 1);

src/api/l_graphics_material.c

@@ -50,7 +50,6 @@ static int l_lovrMaterialGetProperties(lua_State* L) {
   lua_pushnumber(L, info->data.clearcoat), lua_setfield(L, -2, "clearcoat");
   lua_pushnumber(L, info->data.clearcoatRoughness), lua_setfield(L, -2, "clearcoatRoughness");
   lua_pushnumber(L, info->data.occlusionStrength), lua_setfield(L, -2, "occlusionStrength");
-  lua_pushnumber(L, info->data.glowStrength), lua_setfield(L, -2, "glowStrength");
   lua_pushnumber(L, info->data.normalScale), lua_setfield(L, -2, "normalScale");
   lua_pushnumber(L, info->data.alphaCutoff), lua_setfield(L, -2, "alphaCutoff");
   lua_pushnumber(L, info->data.pointSize), lua_setfield(L, -2, "pointSize");

src/api/l_graphics_shader.c

@@ -64,7 +64,7 @@ static int l_lovrShaderHasAttribute(lua_State* L) {
   return 1;
 }
 
-static int l_lovrShaderGetLocalWorkgroupSize(lua_State* L) {
+static int l_lovrShaderGetWorkgroupSize(lua_State* L) {
   Shader* shader = luax_checktype(L, 1, Shader);
 
   if (!lovrShaderHasStage(shader, STAGE_COMPUTE)) {
@@ -73,7 +73,7 @@ static int l_lovrShaderGetLocalWorkgroupSize(lua_State* L) {
   }
 
   uint32_t size[3];
-  lovrShaderGetLocalWorkgroupSize(shader, size);
+  lovrShaderGetWorkgroupSize(shader, size);
   lua_pushinteger(L, size[0]);
   lua_pushinteger(L, size[1]);
   lua_pushinteger(L, size[2]);
@@ -85,6 +85,6 @@ const luaL_Reg lovrShader[] = {
   { "getType", l_lovrShaderGetType },
   { "hasStage", l_lovrShaderHasStage },
   { "hasAttribute", l_lovrShaderHasAttribute },
-  { "getLocalWorkgroupSize", l_lovrShaderGetLocalWorkgroupSize },
+  { "getWorkgroupSize", l_lovrShaderGetWorkgroupSize },
   { NULL, NULL }
 };

src/core/gpu.h

@@ -665,9 +665,9 @@ typedef struct {
   uint32_t clipDistances;
   uint32_t cullDistances;
   uint32_t clipAndCullDistances;
-  uint32_t computeDispatchCount[3];
+  uint32_t workgroupCount[3];
-  uint32_t computeWorkgroupSize[3];
+  uint32_t workgroupSize[3];
-  uint32_t computeWorkgroupVolume;
+  uint32_t totalWorkgroupSize;
   uint32_t computeSharedMemory;
   uint32_t pushConstantSize;
   uint32_t indirectDrawCount;

src/core/gpu_vk.c

@@ -1928,13 +1928,13 @@ bool gpu_init(gpu_config* config) {
       config->limits->clipDistances = limits->maxClipDistances;
       config->limits->cullDistances = limits->maxCullDistances;
       config->limits->clipAndCullDistances = limits->maxCombinedClipAndCullDistances;
-      config->limits->computeDispatchCount[0] = limits->maxComputeWorkGroupCount[0];
+      config->limits->workgroupCount[0] = limits->maxComputeWorkGroupCount[0];
-      config->limits->computeDispatchCount[1] = limits->maxComputeWorkGroupCount[1];
+      config->limits->workgroupCount[1] = limits->maxComputeWorkGroupCount[1];
-      config->limits->computeDispatchCount[2] = limits->maxComputeWorkGroupCount[2];
+      config->limits->workgroupCount[2] = limits->maxComputeWorkGroupCount[2];
-      config->limits->computeWorkgroupSize[0] = limits->maxComputeWorkGroupSize[0];
+      config->limits->workgroupSize[0] = limits->maxComputeWorkGroupSize[0];
-      config->limits->computeWorkgroupSize[1] = limits->maxComputeWorkGroupSize[1];
+      config->limits->workgroupSize[1] = limits->maxComputeWorkGroupSize[1];
-      config->limits->computeWorkgroupSize[2] = limits->maxComputeWorkGroupSize[2];
+      config->limits->workgroupSize[2] = limits->maxComputeWorkGroupSize[2];
-      config->limits->computeWorkgroupVolume = limits->maxComputeWorkGroupInvocations;
+      config->limits->totalWorkgroupSize = limits->maxComputeWorkGroupInvocations;
       config->limits->computeSharedMemory = limits->maxComputeSharedMemorySize;
       config->limits->pushConstantSize = limits->maxPushConstantsSize;
       config->limits->indirectDrawCount = limits->maxDrawIndirectCount;

src/core/spv.c

@@ -187,9 +187,9 @@ static spv_result spv_parse_execution_mode(spv_context* spv, const uint32_t* op,
     return SPV_OK;
   }
 
-  info->localWorkgroupSize[0] = op[3];
+  info->workgroupSize[0] = op[3];
-  info->localWorkgroupSize[1] = op[4];
+  info->workgroupSize[1] = op[4];
-  info->localWorkgroupSize[2] = op[5];
+  info->workgroupSize[2] = op[5];
   return SPV_OK;
 }
 

src/core/spv.h

@@ -56,7 +56,7 @@ typedef struct {
 
 typedef struct {
   uint32_t version;
-  uint32_t localWorkgroupSize[3];
+  uint32_t workgroupSize[3];
   uint32_t featureCount;
   uint32_t specConstantCount;
   uint32_t pushConstantCount;

src/modules/data/modelData.h

@@ -84,7 +84,6 @@ typedef struct {
   float clearcoat;
   float clearcoatRoughness;
   float occlusionStrength;
-  float glowStrength;
   float normalScale;
   float alphaCutoff;
   float pointSize;

src/modules/data/modelData_gltf.c

@@ -695,7 +695,7 @@ ModelData* lovrModelDataInitGltf(ModelData* model, Blob* source, ModelDataIO* io
     ModelMaterial* material = model->materials;
     for (int i = (token++)->size; i > 0; i--, material++) {
       memcpy(material->color, (float[4]) { 1.f, 1.f, 1.f, 1.f }, 16);
-      memcpy(material->glow, (float[4]) { 0.f, 0.f, 0.f, 0.f }, 16);
+      memcpy(material->glow, (float[4]) { 0.f, 0.f, 0.f, 1.f }, 16);
       material->uvShift[0] = 0.f;
       material->uvShift[1] = 0.f;
       material->uvScale[0] = 1.f;
@@ -705,7 +705,6 @@ ModelData* lovrModelDataInitGltf(ModelData* model, Blob* source, ModelDataIO* io
       material->clearcoat = 0.f;
       material->clearcoatRoughness = 0.f;
       material->occlusionStrength = 1.f;
-      material->glowStrength = 1.f;
       material->normalScale = 1.f;
       material->alphaCutoff = 0.f;
       material->pointSize = 1.f;

src/modules/data/modelData_obj.c

@@ -48,7 +48,7 @@ static void parseMtl(char* path, char* base, ModelDataIO* io, arr_image_t* image
       map_set(names, hash64(line + 7, length - 7), materials->length);
       arr_push(materials, ((ModelMaterial) {
         .color = { 1.f, 1.f, 1.f, 1.f },
-        .glow = { 0.f, 0.f, 0.f, 0.f },
+        .glow = { 0.f, 0.f, 0.f, 1.f },
         .uvShift = { 0.f, 0.f },
         .uvScale = { 1.f, 1.f },
         .metalness = 1.f,
@@ -56,7 +56,6 @@ static void parseMtl(char* path, char* base, ModelDataIO* io, arr_image_t* image
         .clearcoat = 0.f,
         .clearcoatRoughness = 0.f,
         .occlusionStrength = 1.f,
-        .glowStrength = 1.f,
         .normalScale = 1.f,
         .alphaCutoff = 0.f,
         .pointSize = 1.f,

src/modules/graphics/graphics.c

@@ -101,7 +101,7 @@ struct Shader {
   ShaderInfo info;
   uint32_t layout;
   uint32_t computePipeline;
-  uint32_t localWorkgroupSize[3];
+  uint32_t workgroupSize[3];
   uint32_t bufferMask;
   uint32_t textureMask;
   uint32_t samplerMask;
@@ -728,9 +728,9 @@ void lovrGraphicsGetLimits(GraphicsLimits* limits) {
   limits->clipDistances = state.limits.clipDistances;
   limits->cullDistances = state.limits.cullDistances;
   limits->clipAndCullDistances = state.limits.clipAndCullDistances;
-  memcpy(limits->computeDispatchCount, state.limits.computeDispatchCount, 3 * sizeof(uint32_t));
+  memcpy(limits->workgroupCount, state.limits.workgroupCount, 3 * sizeof(uint32_t));
-  memcpy(limits->computeWorkgroupSize, state.limits.computeWorkgroupSize, 3 * sizeof(uint32_t));
+  memcpy(limits->workgroupSize, state.limits.workgroupSize, 3 * sizeof(uint32_t));
-  limits->computeWorkgroupVolume = state.limits.computeWorkgroupVolume;
+  limits->totalWorkgroupSize = state.limits.totalWorkgroupSize;
   limits->computeSharedMemory = state.limits.computeSharedMemory;
   limits->shaderConstantSize = MIN(state.limits.pushConstantSize, 256);
   limits->indirectDrawCount = state.limits.indirectDrawCount;
@@ -1556,7 +1556,12 @@ Shader* lovrShaderCreate(const ShaderInfo* info) {
   }
 
   if (info->type == SHADER_COMPUTE) {
-    memcpy(shader->localWorkgroupSize, spv[0].localWorkgroupSize, 3 * sizeof(uint32_t));
+    memcpy(shader->workgroupSize, spv[0].workgroupSize, 3 * sizeof(uint32_t));
+    lovrCheck(shader->workgroupSize[0] <= state.limits.workgroupSize[0], "Shader workgroup size exceeds the 'workgroupSize' limit");
+    lovrCheck(shader->workgroupSize[1] <= state.limits.workgroupSize[1], "Shader workgroup size exceeds the 'workgroupSize' limit");
+    lovrCheck(shader->workgroupSize[2] <= state.limits.workgroupSize[2], "Shader workgroup size exceeds the 'workgroupSize' limit");
+    uint32_t totalWorkgroupSize = shader->workgroupSize[0] * shader->workgroupSize[1] * shader->workgroupSize[2];
+    lovrCheck(totalWorkgroupSize <= state.limits.totalWorkgroupSize, "Shader workgroup size exceeds the 'totalWorkgroupSize' limit");
   }
 
   uint32_t constantStage = spv[0].pushConstantSize > spv[1].pushConstantSize ? 0 : 1;
@@ -1818,8 +1823,8 @@ bool lovrShaderHasAttribute(Shader* shader, const char* name, uint32_t location)
   return false;
 }
 
-void lovrShaderGetLocalWorkgroupSize(Shader* shader, uint32_t size[3]) {
+void lovrShaderGetWorkgroupSize(Shader* shader, uint32_t size[3]) {
-  memcpy(size, shader->localWorkgroupSize, 3 * sizeof(uint32_t));
+  memcpy(size, shader->workgroupSize, 3 * sizeof(uint32_t));
 }
 
 // Material
@@ -5087,9 +5092,9 @@ void lovrPassCompute(Pass* pass, uint32_t x, uint32_t y, uint32_t z, Buffer* ind
 
   Shader* shader = pass->pipeline->shader;
   lovrCheck(shader && shader->info.type == SHADER_COMPUTE, "Tried to run a compute shader, but no compute shader is bound");
-  lovrCheck(x <= state.limits.computeDispatchCount[0], "Compute %s count exceeds computeDispatchCount limit", "x");
+  lovrCheck(x <= state.limits.workgroupCount[0], "Compute %s count exceeds workgroupCount limit", "x");
-  lovrCheck(y <= state.limits.computeDispatchCount[1], "Compute %s count exceeds computeDispatchCount limit", "y");
+  lovrCheck(y <= state.limits.workgroupCount[1], "Compute %s count exceeds workgroupCount limit", "y");
-  lovrCheck(z <= state.limits.computeDispatchCount[2], "Compute %s count exceeds computeDispatchCount limit", "z");
+  lovrCheck(z <= state.limits.workgroupCount[2], "Compute %s count exceeds workgroupCount limit", "z");
 
   gpu_pipeline* pipeline = state.pipelines.data[shader->computePipeline];
 

src/modules/graphics/graphics.h

@@ -72,9 +72,9 @@ typedef struct {
   uint32_t clipDistances;
   uint32_t cullDistances;
   uint32_t clipAndCullDistances;
-  uint32_t computeDispatchCount[3];
+  uint32_t workgroupCount[3];
-  uint32_t computeWorkgroupSize[3];
+  uint32_t workgroupSize[3];
-  uint32_t computeWorkgroupVolume;
+  uint32_t totalWorkgroupSize;
   uint32_t computeSharedMemory;
   uint32_t shaderConstantSize;
   uint32_t indirectDrawCount;
@@ -315,7 +315,7 @@ void lovrShaderDestroy(void* ref);
 const ShaderInfo* lovrShaderGetInfo(Shader* shader);
 bool lovrShaderHasStage(Shader* shader, ShaderStage stage);
 bool lovrShaderHasAttribute(Shader* shader, const char* name, uint32_t location);
-void lovrShaderGetLocalWorkgroupSize(Shader* shader, uint32_t size[3]);
+void lovrShaderGetWorkgroupSize(Shader* shader, uint32_t size[3]);
 
 // Material
 
@@ -330,7 +330,6 @@ typedef struct {
   float clearcoat;
   float clearcoatRoughness;
   float occlusionStrength;
-  float glowStrength;
   float normalScale;
   float alphaCutoff;
   float pointSize;