diff --git a/examples/shaders/shaders_deferred_render.c b/examples/shaders/shaders_deferred_render.c index 4f652fe3..52c713aa 100644 --- a/examples/shaders/shaders_deferred_render.c +++ b/examples/shaders/shaders_deferred_render.c @@ -95,11 +95,19 @@ int main(void) rlEnableFramebuffer(gBuffer.framebuffer); - // Since we are storing position and normal data in these textures, - // we need to use a floating point format. - gBuffer.positionTexture = rlLoadTexture(NULL, screenWidth, screenHeight, RL_PIXELFORMAT_UNCOMPRESSED_R32G32B32, 1); + // NOTE: Vertex positions are stored in a texture for simplicity. A better approach would use a depth texture + // (instead of a detph renderbuffer) to reconstruct world positions in the final render shader via clip-space position, + // depth, and the inverse view/projection matrices. + + // 16-bit precision ensures OpenGL ES 3 compatibility, though it may lack precision for real scenarios. + // But as mentioned above, the positions could be reconstructed instead of stored. If not targeting OpenGL ES + // and you wish to maintain this approach, consider using `RL_PIXELFORMAT_UNCOMPRESSED_R32G32B32`. + gBuffer.positionTexture = rlLoadTexture(NULL, screenWidth, screenHeight, RL_PIXELFORMAT_UNCOMPRESSED_R16G16B16, 1); + + // Similarly, 16-bit precision is used for normals ensures OpenGL ES 3 compatibility. + // This is generally sufficient, but a 16-bit fixed-point format offer a better uniform precision in all orientations. + gBuffer.normalTexture = rlLoadTexture(NULL, screenWidth, screenHeight, RL_PIXELFORMAT_UNCOMPRESSED_R16G16B16, 1); - gBuffer.normalTexture = rlLoadTexture(NULL, screenWidth, screenHeight, RL_PIXELFORMAT_UNCOMPRESSED_R32G32B32, 1); // Albedo (diffuse color) and specular strength can be combined into one texture. // The color in RGB, and the specular strength in the alpha channel. gBuffer.albedoSpecTexture = rlLoadTexture(NULL, screenWidth, screenHeight, RL_PIXELFORMAT_UNCOMPRESSED_R8G8B8A8, 1);