Reducing GLB transfer size can shorten downloads, but transfer size, decode time, GPU memory, draw calls, triangle count, and frame rate are different measurements. Optimize the bottleneck you observe rather than assuming that a smaller file automatically renders faster.
Outcomes to measure
- Transferred bytes and time to first useful render
- Decode time and peak CPU/GPU memory
- Frame time, draw calls, triangles, and overdraw in the target scene
Optimization checklist
- Reduce texture resolution and compress to KTX2/WebP.
- Use Draco or mesh quantization for geometry.
- Remove unused materials, nodes, and animations.
- Validate performance on the target device.
Ready to optimize? Start with OptimizeGLB and explore advanced workflows in the texture optimization guide.