TripoSR
TripoSR is a feed-forward reconstruction model jointly developed by Stability AI and Tripo AI. It generates a 3D object from one image using an architecture based on Large Reconstruction Model research.
Model design and access
The card describes changes to data preparation, architecture and training over the original LRM approach. It points to an implementation in the TripoSR repository. This earlier model remains downloadable, although the card directs readers to Stable Fast 3D for a newer asset pipeline.TripoSR official model card
Practical use
Practical guidance. TripoSR is a useful baseline for image-to-3D research and tools that require permissive weight terms. Its joint origin should remain visible when comparing providers. Review mesh quality and hidden surfaces independently of generation speed; a quick reconstruction can still need considerable asset cleanup.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Developers | Stability AI and Tripo AI.TripoSR official model card |
| License | MIT.TripoSR official model card |
| Training source | A selected subset of Objaverse renders.TripoSR official model card |
Related models
- Stable Fast 3D
Stability AI · Stable 3D
Stable Fast 3D reconstructs a textured mesh from a single object image.
Launch date not confirmedAPI · Downloadable - Stable Point Aware 3D
Stability AI · Stable 3D
A 3D reconstruction model that uses a generated point cloud to control textured-mesh creation from an image.
Launch date not confirmedAPI · Downloadable