Compression
Mesh Quantization
Reduce vertex attribute storage size using the KHR_mesh_quantization extension.
Default: Off
glTF Extension: KHR_mesh_quantization
By default, glTF stores vertex attributes such as normals, tangents, and UV coordinates as 32-bit floats. Mesh quantization allows the exporter to store these attributes in smaller integer types, reducing the size of the geometry data in the file.
What it does
- Normals, tangents, and texture coordinates are packed into smaller integer representations
- The
KHR_mesh_quantizationextension is registered as required in the output file - Geometry loads identically to the original — the viewer automatically converts integers back to float values
When to use it
- You have complex scenes with many meshes and geometry size is a download bottleneck
- Your target runtime explicitly supports
KHR_mesh_quantization - You are building for web delivery where reducing transfer size matters
When to avoid it
- Your runtime does not support
KHR_mesh_quantization— the file will be rejected at load time - You are handing off to a pipeline tool that expects standard float vertex attributes
Warning
This extension is marked required in the exported file. Any viewer or tool that does not support KHR_mesh_quantization will refuse to open the file. Always confirm runtime support before using this in production assets.