Zero-1-to-A: Zero-Shot One image to Animatable Head Avatars Using Video Diffusion

Zhenglin Zhou · Fan Ma · Hehe Fan^* · Tat-Seng Chua

Zero-1-to-A is an image-to-4D avatar generation method. It synthesizes a spatial and temporal consistency dataset for 4D avatar reconstruction using the video diffusion model.

Citation

If you find Zero-1-to-A useful for your research and applications, please cite us using this BibTeX:

@inproceedings{zhou2025zero1toa,
    author    = {Zhou, Zhenglin and Ma, Fan and Fan, Hehe and Chua, Tat-Seng},
    title     = {Zero-1-to-A: Zero-Shot One image to Animatable Head Avatars Using Video Diffusion}, 
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    year      = {2025}
}