Objaverse trained model?

[ajhuofa]

Hi, the HuggingFace release only contains the DL3DV scene checkpoints. Is there a plan to release the
Objaverse-trained checkpoint used for the object-centric results on the project page?

Thanks!

[cwchenwang]

Thanks for your interest! The Objaverse-trained checkpoint used for the object-centric results is based on a commercial/internal model, so unfortunately we are not able to release that checkpoint at this time. We sincerely appreciate your understanding.

If you would like to reimplement, we suggest you use LaCT codebase and change the output to GS: https://github.com/a1600012888/LaCT/blob/main/lact_nvs/README.md.