BEFORE YOU SHIP

Small enough.
Ready to load.

A GLB file extension does not guarantee that every application can load its contents. Check the extensions used by the output and test your destination.

Match the settings to your loader

These requirements can combine. The result panel reports compatibility information for your output. Our preview includes decoders, so a successful preview does not prove that your own application has them configured.

The presets use geometry quantization by default. Disabling it can remove the geometry quantization requirement, but Meshopt or converted textures still need their corresponding loaders. Check the extensions reported for the actual result.

Using Three.js

Configure a Meshopt decoder through GLTFLoader.setMeshoptDecoder() for compressed geometry. For KTX2, configure a KTX2Loader, detect renderer support and attach it through setKTX2Loader(). Serve the matching transcoder assets with your application. Consult the GLTFLoader documentation and KTX2Loader documentation for your version.

Names, extras and scene behavior

The three preservation toggles are enabled by default. Leave them enabled when your application finds objects or materials by name or reads extras. They do not promise identical node indices, hierarchy, animation samples or custom attributes. Optimization can merge or reorganize data and change precision.

Test selection, material changes, animation playback, morph targets and object transforms in the actual application. Preserve extras is not a way to preserve arbitrary custom extensions.

Input restrictions

This service rejects Draco-compressed input, external resources and unsupported required extensions. Unknown optional extensions may be discarded. Export an uncompressed, embedded-resource GLB from the source application when necessary.

For the engine's extension list and optimization behavior, see upstream gltfpack documentation. The web service applies additional validation and resource limits.