PERSE: Personalized 3D Generative Avatars from A Single Portrait

Hyunsoo Cha  Inhee Lee  Hanbyul Joo

Seoul National University

CVPR 2025

teaser_pp.mp4

TL;DR: Given a reference portrait image input, PERSE generates an animatable 3D personalized avatar with disentangled and editable control over various facial attributes.

📣 News

• 2025.02.27: 🎉 PERSE got accepted into CVPR 2025!
• 2024.12.31: 👏 AK and MrNeRF introduced our work! Thank you for sharing it.
• 2024.12.30: 🎉 We release paper and project page.

🌟 TODOs

• Released the paper and project page on 2024.12.30.
• Release inference code and pretrained weights of avatar model.
• Release inference code and pretrained weights of portrait-CHAMP (image-to-video diffusion model).
• Release training code for portrait-CHAMP.
• Release training code for avatar model.

📢 License

Codes are available only for non-commercial research purposes.

✏️ Citation

If you find our work useful, please cite our paper:

@article{cha2024perse,
  title={PERSE: Personalized 3D Generative Avatars from A Single Portrait},
  author={Cha, Hyunsoo and Lee, Inhee and Joo, Hanbyul},
  journal={arXiv preprint arXiv:2412.21206},
  year={2024}
}