deduplicate-hierarchies

USD hierarchy deduplication works by detecting repeated prim subtrees under the default prim and replacing them with internal instanceable references to the first prototype. It matches subtrees by structure, prim types, authored property names, and property values.

Yes, you can safely collapse duplicate USD subtrees without flattening or losing authored prototype paths. The tool validates the input USD path and exports from the root layer to preserve prototype names during hierarchy replacement.

Subtree matching allows tolerance for float-array and scalar float or double drifts. It verifies subtree shape, prim types, authored property names, and property values to ensure only truly identical hierarchies are merged.

The best way to catch remaining identical meshes is to pair hierarchy dedup with per-mesh dedup in a canonical two-op workflow. This pipeline approach ensures identical meshes not folded at the subtree level are still caught.

USD hierarchy deduplication works on the entire stage or specific prim paths. You provide the full asset path and can optionally restrict the scan to one or more prim paths under the default prim.

No additional dependencies are needed to run the USD scene optimizer pipeline. The tool invokes registered Scene Optimizer operations via a standalone runner pipeline without requiring external dependencies.