Support instancing in OpenGL ES 2.0 if available

Checking for extension and enabling it if available
This commit is contained in:
Ray 2021-03-17 13:56:00 +01:00
parent 3c76b5cc8e
commit aba69146f2

View File

@ -892,11 +892,12 @@ typedef struct rlglData {
} State; // Renderer state
struct {
bool vao; // VAO support (OpenGL ES2 could not support VAO extension)
bool texNPOT; // NPOT textures full support
bool texDepth; // Depth textures supported
bool texFloat32; // float textures support (32 bit per channel)
bool texCompDXT; // DDS texture compression support
bool vao; // VAO support (OpenGL ES2 could not support VAO extension) (GL_ARB_vertex_array_object)
bool instancing; // Instancing supported
bool texNPOT; // NPOT textures full support (GL_ARB_texture_non_power_of_two, GL_OES_texture_npot)
bool texDepth; // Depth textures supported (GL_ARB_depth_texture, GL_WEBGL_depth_texture)
bool texFloat32; // float textures support (32 bit per channel) (GL_OES_texture_float)
bool texCompDXT; // DDS texture compression support (GL_EXT_texture_compression_s3tc)
bool texCompETC1; // ETC1 texture compression support
bool texCompETC2; // ETC2/EAC texture compression support
bool texCompPVRT; // PVR texture compression support
@ -930,9 +931,14 @@ static rlglData RLGL = { 0 };
#if defined(GRAPHICS_API_OPENGL_ES2)
// NOTE: VAO functionality is exposed through extensions (OES)
static PFNGLGENVERTEXARRAYSOESPROC glGenVertexArrays; // Entry point pointer to function glGenVertexArrays()
static PFNGLBINDVERTEXARRAYOESPROC glBindVertexArray; // Entry point pointer to function glBindVertexArray()
static PFNGLDELETEVERTEXARRAYSOESPROC glDeleteVertexArrays; // Entry point pointer to function glDeleteVertexArrays()
static PFNGLGENVERTEXARRAYSOESPROC glGenVertexArrays = NULL;
static PFNGLBINDVERTEXARRAYOESPROC glBindVertexArray = NULL;
static PFNGLDELETEVERTEXARRAYSOESPROC glDeleteVertexArrays = NULL;
// NOTE: Instancing functionality could also be available through extension
static PFNGLDRAWARRAYSINSTANCEDEXTPROC glDrawArraysInstanced = NULL;
static PFNGLDRAWELEMENTSINSTANCEDEXTPROC glDrawElementsInstanced = NULL;
static PFNGLVERTEXATTRIBDIVISOREXTPROC glVertexAttribDivisor = NULL;
#endif
//----------------------------------------------------------------------------------
@ -1617,10 +1623,9 @@ void rlglInit(int width, int height)
GLint numExt = 0;
#if defined(GRAPHICS_API_OPENGL_33) && !defined(GRAPHICS_API_OPENGL_21)
// NOTE: On OpenGL 3.3 VAO and NPOT are supported by default
// OpenGL 3.3 extensions supported by default (core)
RLGL.ExtSupported.vao = true;
// Multiple texture extensions supported by default
RLGL.ExtSupported.instancing = true;
RLGL.ExtSupported.texNPOT = true;
RLGL.ExtSupported.texFloat32 = true;
RLGL.ExtSupported.texDepth = true;
@ -1685,7 +1690,17 @@ void rlglInit(int width, int height)
if ((glGenVertexArrays != NULL) && (glBindVertexArray != NULL) && (glDeleteVertexArrays != NULL)) RLGL.ExtSupported.vao = true;
}
// Check instanced rendering support
if (strcmp(extList[i], (const char *)"GL_ANGLE_instanced_arrays") == 0)
{
glDrawArraysInstanced = (PFNGLDRAWARRAYSINSTANCEDEXTPROC)eglGetProcAddress("glDrawArraysInstancedANGLE");
glDrawElementsInstanced = (PFNGLDRAWELEMENTSINSTANCEDEXTPROC)eglGetProcAddress("glDrawElementsInstancedANGLE");
glVertexAttribDivisor = (PFNGLVERTEXATTRIBDIVISOREXTPROC)eglGetProcAddress("glVertexAttribDivisorANGLE");
if ((glDrawArraysInstanced != NULL) && (glDrawElementsInstanced != NULL) && (glVertexAttribDivisor != NULL)) RLGL.ExtSupported.instancing = true;
}
// Check NPOT textures support
// NOTE: Only check on OpenGL ES, OpenGL 3.3 has NPOT textures full support as core feature
if (strcmp(extList[i], (const char *)"GL_OES_texture_npot") == 0) RLGL.ExtSupported.texNPOT = true;
@ -2925,95 +2940,95 @@ void rlDrawMesh(Mesh mesh, Material material, Matrix transform)
// Draw a 3d mesh with material and transform
void rlDrawMeshInstanced(Mesh mesh, Material material, Matrix *transforms, int count)
{
#if defined(GRAPHICS_API_OPENGL_33)
// Bind shader program
glUseProgram(material.shader.id);
// Upload to shader material.colDiffuse
if (material.shader.locs[SHADER_LOC_COLOR_DIFFUSE] != -1)
glUniform4f(material.shader.locs[SHADER_LOC_COLOR_DIFFUSE], (float)material.maps[MATERIAL_MAP_DIFFUSE].color.r/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.g/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.b/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.a/255.0f);
// Upload to shader material.colSpecular (if available)
if (material.shader.locs[SHADER_LOC_COLOR_SPECULAR] != -1)
glUniform4f(material.shader.locs[SHADER_LOC_COLOR_SPECULAR], (float)material.maps[MATERIAL_MAP_SPECULAR].color.r/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.g/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.b/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.a/255.0f);
// Bind active texture maps (if available)
for (int i = 0; i < MAX_MATERIAL_MAPS; i++)
#if defined(GRAPHICS_API_OPENGL_33) || defined(GRAPHICS_API_OPENGL_ES2)
if (RLGL.ExtSupported.instancing)
{
if (material.maps[i].texture.id > 0)
// Bind shader program
glUseProgram(material.shader.id);
// Upload to shader material.colDiffuse
if (material.shader.locs[SHADER_LOC_COLOR_DIFFUSE] != -1)
glUniform4f(material.shader.locs[SHADER_LOC_COLOR_DIFFUSE], (float)material.maps[MATERIAL_MAP_DIFFUSE].color.r/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.g/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.b/255.0f,
(float)material.maps[MATERIAL_MAP_DIFFUSE].color.a/255.0f);
// Upload to shader material.colSpecular (if available)
if (material.shader.locs[SHADER_LOC_COLOR_SPECULAR] != -1)
glUniform4f(material.shader.locs[SHADER_LOC_COLOR_SPECULAR], (float)material.maps[MATERIAL_MAP_SPECULAR].color.r/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.g/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.b/255.0f,
(float)material.maps[MATERIAL_MAP_SPECULAR].color.a/255.0f);
// Bind active texture maps (if available)
for (int i = 0; i < MAX_MATERIAL_MAPS; i++)
{
glActiveTexture(GL_TEXTURE0 + i);
if ((i == MATERIAL_MAP_IRRADIANCE) || (i == MATERIAL_MAP_PREFILTER) || (i == MATERIAL_MAP_CUBEMAP))
glBindTexture(GL_TEXTURE_CUBE_MAP, material.maps[i].texture.id);
else glBindTexture(GL_TEXTURE_2D, material.maps[i].texture.id);
if (material.maps[i].texture.id > 0)
{
glActiveTexture(GL_TEXTURE0 + i);
if ((i == MATERIAL_MAP_IRRADIANCE) || (i == MATERIAL_MAP_PREFILTER) || (i == MATERIAL_MAP_CUBEMAP))
glBindTexture(GL_TEXTURE_CUBE_MAP, material.maps[i].texture.id);
else glBindTexture(GL_TEXTURE_2D, material.maps[i].texture.id);
glUniform1i(material.shader.locs[SHADER_LOC_MAP_DIFFUSE + i], i);
glUniform1i(material.shader.locs[SHADER_LOC_MAP_DIFFUSE + i], i);
}
}
// Bind vertex array objects (or VBOs)
glBindVertexArray(mesh.vaoId);
// At this point the modelview matrix just contains the view matrix (camera)
// For instanced shaders "mvp" is not premultiplied by any instance transform, only RLGL.State.transform
glUniformMatrix4fv(material.shader.locs[SHADER_LOC_MATRIX_MVP], 1, false,
MatrixToFloat(MatrixMultiply(MatrixMultiply(RLGL.State.transform, RLGL.State.modelview), RLGL.State.projection)));
float16* instanceTransforms = RL_MALLOC(count*sizeof(float16));
for (int i = 0; i < count; i++) instanceTransforms[i] = MatrixToFloatV(transforms[i]);
// This could alternatively use a static VBO and either glMapBuffer or glBufferSubData.
// It isn't clear which would be reliably faster in all cases and on all platforms, and
// anecdotally glMapBuffer seems very slow (syncs) while glBufferSubData seems no faster
// since we're transferring all the transform matrices anyway.
unsigned int instancesB = 0;
glGenBuffers(1, &instancesB);
glBindBuffer(GL_ARRAY_BUFFER, instancesB);
glBufferData(GL_ARRAY_BUFFER, count*sizeof(float16), instanceTransforms, GL_STATIC_DRAW);
// Instances are put in SHADER_LOC_MATRIX_MODEL attribute location with space for 4x Vector4, eg:
// layout (location = 12) in mat4 instance;
unsigned int instanceA = material.shader.locs[SHADER_LOC_MATRIX_MODEL];
for (unsigned int i = 0; i < 4; i++)
{
glEnableVertexAttribArray(instanceA+i);
glVertexAttribPointer(instanceA + i, 4, GL_FLOAT, GL_FALSE, sizeof(Matrix), (void *)(i*sizeof(Vector4)));
glVertexAttribDivisor(instanceA + i, 1);
}
glBindBuffer(GL_ARRAY_BUFFER, 0);
// Draw instanced
if (mesh.indices != NULL) glDrawElementsInstanced(GL_TRIANGLES, mesh.triangleCount*3, GL_UNSIGNED_SHORT, 0, count);
else glDrawArraysInstanced(GL_TRIANGLES, 0, mesh.vertexCount, count);
glDeleteBuffers(1, &instancesB);
RL_FREE(instanceTransforms);
// Unbind all binded texture maps
for (int i = 0; i < MAX_MATERIAL_MAPS; i++)
{
glActiveTexture(GL_TEXTURE0 + i); // Set shader active texture
if ((i == MATERIAL_MAP_IRRADIANCE) || (i == MATERIAL_MAP_PREFILTER) || (i == MATERIAL_MAP_CUBEMAP)) glBindTexture(GL_TEXTURE_CUBE_MAP, 0);
else glBindTexture(GL_TEXTURE_2D, 0); // Unbind current active texture
}
// Unind vertex array objects (or VBOs)
glBindVertexArray(0);
// Unbind shader program
glUseProgram(0);
}
// Bind vertex array objects (or VBOs)
glBindVertexArray(mesh.vaoId);
// At this point the modelview matrix just contains the view matrix (camera)
// For instanced shaders "mvp" is not premultiplied by any instance transform, only RLGL.State.transform
glUniformMatrix4fv(material.shader.locs[SHADER_LOC_MATRIX_MVP], 1, false,
MatrixToFloat(MatrixMultiply(MatrixMultiply(RLGL.State.transform, RLGL.State.modelview), RLGL.State.projection)));
float16* instanceTransforms = RL_MALLOC(count*sizeof(float16));
for (int i = 0; i < count; i++) instanceTransforms[i] = MatrixToFloatV(transforms[i]);
// This could alternatively use a static VBO and either glMapBuffer or glBufferSubData.
// It isn't clear which would be reliably faster in all cases and on all platforms, and
// anecdotally glMapBuffer seems very slow (syncs) while glBufferSubData seems no faster
// since we're transferring all the transform matrices anyway.
unsigned int instancesB = 0;
glGenBuffers(1, &instancesB);
glBindBuffer(GL_ARRAY_BUFFER, instancesB);
glBufferData(GL_ARRAY_BUFFER, count*sizeof(float16), instanceTransforms, GL_STATIC_DRAW);
// Instances are put in SHADER_LOC_MATRIX_MODEL attribute location with space for 4x Vector4, eg:
// layout (location = 12) in mat4 instance;
unsigned int instanceA = material.shader.locs[SHADER_LOC_MATRIX_MODEL];
for (unsigned int i = 0; i < 4; i++)
{
glEnableVertexAttribArray(instanceA+i);
glVertexAttribPointer(instanceA + i, 4, GL_FLOAT, GL_FALSE, sizeof(Matrix), (void *)(i*sizeof(Vector4)));
glVertexAttribDivisor(instanceA + i, 1);
}
glBindBuffer(GL_ARRAY_BUFFER, 0);
// Draw call!
if (mesh.indices != NULL) glDrawElementsInstanced(GL_TRIANGLES, mesh.triangleCount*3, GL_UNSIGNED_SHORT, 0, count);
else glDrawArraysInstanced(GL_TRIANGLES, 0, mesh.vertexCount, count);
glDeleteBuffers(1, &instancesB);
RL_FREE(instanceTransforms);
// Unbind all binded texture maps
for (int i = 0; i < MAX_MATERIAL_MAPS; i++)
{
glActiveTexture(GL_TEXTURE0 + i); // Set shader active texture
if ((i == MATERIAL_MAP_IRRADIANCE) || (i == MATERIAL_MAP_PREFILTER) || (i == MATERIAL_MAP_CUBEMAP)) glBindTexture(GL_TEXTURE_CUBE_MAP, 0);
else glBindTexture(GL_TEXTURE_2D, 0); // Unbind current active texture
}
// Unind vertex array objects (or VBOs)
glBindVertexArray(0);
// Unbind shader program
glUseProgram(0);
#else
TRACELOG(LOG_WARNING, "VAO: Instanced rendering requires GRAPHICS_API_OPENGL_33");
#endif
}