Usd Optimize — Developer Library

Usd Optimize is a USD scene optimization library: a broad set of operations for processing and optimizing Universal Scene Description (USD) stages. A published package includes C++ headers, Python bindings, and prebuilt libraries so you can embed Usd Optimize in your own applications and pipelines — without installing Omniverse Kit.

This project is currently not accepting contributions.

Detailed docs live alongside the source:

Support: GitHub Issues · Security: SECURITY.md · Governance: Code of Conduct

After cloning this repository (see Requirements for the C++ toolchain):

./repo.sh build
./repo.sh test

./repo.sh ci format checks formatting the way CI does. OpenUSD, Python, premake, and other build-time dependencies are fetched automatically via Packman during build — no manual install required.

To consume a published binary drop instead of building from source, follow the install guide for Windows or Linux (x86_64, aarch64).

Usd Optimize ships 45+ operations — mesh cleanup and decimation, geometry/material deduplication, hierarchy flattening, UV generation, remeshing, and more. Each operation declares typed arguments and can run individually or chained into a JSON stack.

See the full catalog with per-operation arguments and JSON examples in docs/operations.rst, and guidance on which to apply for a given goal in docs/choosing-operations.rst.

Note: A few operations (Python Script, Delete Hidden Prims, Remove Untyped Prims, Move Materials) are Python-plugin operations — they run via the usd-optimize wheel/bindings but not from the standalone usdOptimize CLI, which does not host a Python interpreter. See docs/cli.rst.

Usd Optimize integrates with the usd-validation-nvidia PyPI package to expose its performance and geometry checks as validation rules under the Usd:Performance and Omni:Geometry categories (each rule wraps the analysis mode of an operation). With the usd-optimize wheel installed, the rules register themselves through a package entry point — no register_all() needed:

from usd_validation_nvidia import ValidationEngine
from pxr import Usd

engine = ValidationEngine()  # rules auto-register on import of usd_validation_nvidia
results = engine.validate(Usd.Stage.Open("source/tests/data/simpleFourCubes.usda"))
for issue in results.issues():
    print(issue.severity, issue.rule.__name__, issue.message)

From a source checkout (no installed wheel), call register_all() yourself, importing usd_validation_nvidia first:

from usd_validation_nvidia import ValidationEngine
from usd_optimize.validators import register_all

register_all()

For running the validators from a source checkout, prefer tools/validators/run.sh (see the run-validators skill) — it aligns PYTHONPATH/loader paths with the build tree and avoids the CLI footguns. Full details, including raw nvidia_usd_validate usage and the libusd/pxr alignment matrix, are in docs/performance-validators.rst and the run-validators skill.

Prefer not to build from source? Consume a published binary drop — C++ headers, prebuilt libraries, and Python bindings — via the per-OS install guide:

Source builds use the usd_ver / python_ver tokens in repo.toml; supported values are listed in deps/usd_flavors.json. Published drops are built once per USD flavor, so a drop's own package name is authoritative for both — this file ships unchanged into every flavor and cannot name a single version.

The usd-optimize wheel produced by ./repo.sh py_package declares a specific Python minor in its tags (see requires-python in tools/pyproject/pyproject.toml and the cp3xx segment in the wheel filename). It is a cp312 wheel, so set PYTHON_BIN to a python3.12 interpreter and use "$PYTHON_BIN" -m pip for install and for any python -m … invocations.

./repo.sh --set-token usd_ver:26.08 build && ./repo.sh py_package --test

./repo.sh --set-token usd_flavor:usd --set-token usd_ver:26.08 build -r

• Git and Git LFS (LFS tracks USD assets, textures, and other binary fixtures via .gitattributes).
• (Windows, C++) Visual Studio 2019/2022 (or Build Tools) with the Desktop development with C++ workload, plus the Windows SDK.
• (Linux) build-essential (sudo apt-get install build-essential).

All other build-time dependencies (USD, Python, third-party libraries, premake) are pulled via Packman during ./repo.sh build.

Platform-specific build notes — Windows host-toolchain discovery, PYTHONUTF8 on non-English locales, and the extra patchelf/PyPI requirements of ./repo.sh py_package on Linux — are in docs/developer.rst.

Usd Optimize Core is licensed under the Apache License, Version 2.0. Prebuilt drops carry the same text at PACKAGE-LICENSES/LICENSE.

Copyright (c) 2022-2026, NVIDIA CORPORATION.

For third-party and bundled components in a published package, see PACKAGE-LICENSES/.