wenjieshen
wenjieshen
Replace GlSimpleBlendTask with GlDirectBlendTask to perform blending directly in the shape shader without a separate compose pass. This copies only the required destination region and blends in a single draw....
Allows the engine to render directly to the screen's back buffer, bypassing intermediate framebuffers when possible. This change introduces a new `mDirectTarget` flag to indicate whether the rendering target is...
- Move straight-alpha premultiplication from fragment shader to CPU during texture upload, using a buffered conversion path with OpenMP to keep upload O(n) but off the GPU hot path. -...
We've implemented a comprehensive binary shader caching system to significantly improve OpenGL rendering performance on macOS by eliminating redundant shader compilation overhead on subsequent application runs. This feature introduces the...
## Problem All shapes currently go through the stencil-then-cover pipeline. Even rectangles and other simple convex shapes trigger two raster passes and two draw calls. In large scenes, this produces...