Rigyd vs NVIDIA SimReady asset library
They solve different halves of the same problem. NVIDIA's SimReady library is a curated catalog of ready-made assets. Rigyd converts raw 3D, images, and text into SimReady OpenUSD. Here's how they compare and when to use each.
Updated May 19, 2026 · by Ugur Yekta
The short answer
NVIDIA's SimReady library offers roughly 1,000 free, hand-validated assets — ideal for prototyping and generic scenes. Rigyd converts raw 3D models, images, and text descriptions into SimReady OpenUSD in minutes — ideal when simulation must match your real environment's specific objects at scale. Most teams use both.
Side-by-side comparison
| NVIDIA SimReady Library Curated free catalog | Rigyd Convert your own catalog | |
|---|---|---|
| Asset source | ~1,000 NVIDIA-curated assets | Your own raw 3D (.glb, .fbx, .obj), images, or text |
| Catalog size | ~1,000 fixed assets | Unlimited — bounded only by your catalog |
| Cost | Free | $29/mo Starter; Enterprise custom |
| Physics accuracy | Hand-validated, high precision | AI-estimated: mass ±15-20%, friction ±0.1 |
| Covers your specific SKUs | No — generic objects only | Yes — your exact products and equipment |
| Output formats | OpenUSD | OpenUSD + MJCF, URDF, FBX on demand |
| Time to a new asset | N/A — use what exists | ~5 minutes per asset |
| Bulk / API processing | Manual download | Enterprise API for entire catalogs |
| Best fit | Prototyping, generic scenes | Production sim matching your real environment |
NVIDIA SimReady library: curated quality, fixed scope
NVIDIA's SimReady asset library is a set of roughly 1,000 hand-validated 3D assets built to the SimReady specification — correct USDPhysics schemas, semantic labels, and material bindings. Because each asset is curated, physics quality is high and consistent. The trade-off is scope: the library covers common, generic objects (pallets, crates, basic furniture, tools), not the specific SKUs in your warehouse or the custom equipment on your factory floor. It is free and the fastest way to start prototyping a scene.
Rigyd: your catalog, automated
Rigyd takes the opposite approach: instead of a fixed catalog, it converts raw 3D models, images, or text descriptions you provide into SimReady OpenUSD in about five minutes. Mass, friction, collision meshes, and semantic labels are estimated automatically, calibrated within domain-randomization variance ranges (mass ±15-20%, friction ±0.1). This is what you need when the simulation has to match your real environment — the 50,000 unique SKUs in a warehouse, or the proprietary fixtures in a manufacturing line — rather than generic stand-ins.
They are complementary, not mutually exclusive
The most common production setup uses both: NVIDIA's curated library for common, generic objects where a stand-in is fine, and Rigyd for the long tail of catalog-specific assets that no curated library could contain. The gap NVIDIA itself names — roughly 1,000 SimReady assets exist while a single warehouse needs 50,000+ — is precisely the gap automated conversion fills.
When to choose each
NVIDIA SimReady Library
Prototyping, generic scenes, zero budget, and getting a simulation running quickly with common objects.
Rigyd
Production simulation that must match your real environment — your specific SKUs, custom equipment, and asset diversity beyond ~1,000 objects.
Both together
Most production teams: curated assets for common objects, automated conversion for the catalog-specific long tail.
Where Rigyd fits
Rigyd outputs the same SimReady-spec OpenUSD that NVIDIA's library uses, so assets from both sources drop into the same Isaac Sim or Omniverse scene without friction. If you have started with NVIDIA's library and hit its coverage ceiling, Rigyd extends it with your own catalog rather than replacing it.
Frequently asked questions
Is Rigyd a replacement for NVIDIA's SimReady asset library?
No — they are complementary. NVIDIA's library provides curated, generic assets for free. Rigyd converts raw 3D, images, and text into SimReady OpenUSD. Teams typically use the library for common objects and Rigyd for catalog-specific assets the library does not contain. Both output SimReady-spec OpenUSD, so they coexist in the same scene.
How accurate is Rigyd's physics compared to NVIDIA's hand-validated assets?
NVIDIA's curated assets have hand-validated, high-precision physics. Rigyd's are AI-estimated within domain-randomization variance — mass ±15-20%, friction ±0.1 coefficient. For policies trained with domain randomization, that accuracy is sufficient; for a small set of mission-critical hero assets, hand-measured values remain the gold standard.
Can I use NVIDIA SimReady assets and Rigyd assets in the same simulation?
Yes. Both produce OpenUSD that conforms to the SimReady specification with valid USDPhysics schemas and semantic labels. They drop into the same Isaac Sim or Omniverse scene without conversion, so you can mix curated library objects with your own Rigyd-converted catalog freely.
Why not just wait for NVIDIA to expand its SimReady library?
A curated library cannot scale to every team's specific catalog — your warehouse SKUs, custom tooling, and proprietary equipment will never be in a general-purpose library. Automated conversion exists to fill exactly that long tail, which is why NVIDIA itself frames the ~1,000-asset library against the 50,000+ objects a single warehouse simulation needs.
Convert your own catalog to SimReady
Upload any 3D model and get a physics-enabled OpenUSD asset in minutes — exports to MJCF, URDF, and FBX too.
Related reading
Synthetic data generation for computer vision in robotics
Real-world labeled robotics data is expensive and slow to capture. Synthetic data is fast and unlimited — but only useful if the underlying simulation has correct physics, semantic labels, and domain randomization. Here's the complete pipeline.
Digital twin creation pipeline for manufacturing
A factory digital twin needs every object to behave physically, not just render. This is the end-to-end pipeline: CAD intake, BIM merge, physics layer, semantic labeling, simulation runtime — at the asset volumes (10K+ unique SKUs) real factories actually contain.
How to set up mass, friction, and joint properties for robot training
The three pillars of robot physics setup — mass, friction, joints — determine whether your trained policy transfers to real hardware. Here's the calibration target for each, the schemas, and the common mistakes that quietly break training.