search

Yukun-Huang / DreamWaltz-G

Official implementation of the paper "DreamWaltz-G: Expressive 3D Gaussian Avatars from Skeleton-Guided 2D Diffusion".
https://yukun-huang.github.io/DreamWaltz-G/
49 stars 2 forks source link
3d-avatar 3d-generation animation diffusion-model

readme

DreamWaltz-G: Expressive 3D Gaussian Avatars from Skeleton-Guided 2D Diffusion

  _**[Yukun Huang](https://scholar.google.com/citations?user=lHb5gzoAAAAJ), [Jianan Wang](https://github.com/wendyjnwang), [Ailing Zeng](https://ailingzeng.site), [Zheng-Jun Zha](https://en.auto.ustc.edu.cn/2021/0616/c26828a513174/page.htm), [Lei Zhang](https://www.leizhang.org), [Xihui Liu](https://xh-liu.github.io)**_
## News - Our code, pre-trained models, and results will be released in a few weeks. ## Introduction


DreamWaltz-G utilizes skeleton-guided 2D diffusion for text-to-3D avatar generation and expressive whole-body animation, which supports diverse applications like shape control & editing, 2D video reenactment, and 3D scene composition.