Image-to-3D
Hunyuan3D-2
English
Chinese

Update model card: pipeline tag, remove invalid library, add explicit links

#1
by nielsr HF Staff - opened

This PR improves the model card for Hunyuan3D-Omni by:

  • Updating the pipeline_tag in the metadata from image-to-3d to text-to-3d to accurately reflect the model's capability for controllable 3D asset generation through various signals, including text.
  • Removing the library_name: hunyuan3d-2 from the metadata. This entry does not correspond to a standard Hugging Face library and therefore does not enable automated usage snippets as intended for this metadata field. The model uses a custom hy3dshape framework.
  • Updating the main heading to the full paper title.
  • Adding explicit links to the paper (Hugging Face papers page), project page, and GitHub repository at the top of the model card for better visibility and discoverability.

These changes ensure the model card is more informative, correctly categorized, and aligns with Hugging Face Hub best practices.

Tencent org

Hi there,

Thanks for your pr.

  1. Our model only supports image-to-3d and so text-to-3d is not accurate. Our controllable 3D asset generation refers to cross-modal generation such as image+skeleton, but not text.
  2. We want to keep hunyuan3d-2 so users can easily find our model family.
    3/4. Sure.

Great, thanks for your feedback!

As for library_name: hunyuan3d-2, you could enable a custom code snippet when people click on the "Use this model" button at the top right.

See here an example on how to define such a custom code snippet for your library: https://github.com/huggingface/huggingface.js/pull/1756

Cannot merge
This branch has merge conflicts in the following files:
  • README.md

Sign up or log in to comment