GLTF Pipeline Optimization

What this builds

A complete asset pipeline that takes raw glTF/GLB models (from Blender, Cinema 4D, Maya, or any DCC tool) and produces web-optimized assets with compressed geometry (Draco or meshopt), compressed textures (KTX2/Basis Universal), deduplicated data, welded vertices, automatic instancing, and LOD generation. Includes the runtime decoder setup for Three.js and R3F, and concrete size/performance budgets to hit.

• Build-time CLI pipeline using gltf-transform and gltfpack.
• Draco compression for geometry (70-90% smaller vertex data).
• KTX2/Basis Universal texture compression (75% smaller textures with GPU-native decompression).
• Mesh deduplication, vertex welding, and automatic instancing.
• LOD generation for distance-based quality scaling.
• Runtime loader configuration: DRACOLoader, KTX2Loader, MeshoptDecoder.
• Size budget targets with concrete thresholds per asset class.
• Verification workflow: before/after comparison with visual diff.