EDEN OS V2 — Credits & Acknowledgements

Hallo — Portrait Image Animation

EDEN OS V2's face animation system is powered by Hallo, developed by the Fudan University Generative Vision Lab (fudan-generative-vision).

We extend our deepest gratitude and thanks to the Hallo team for their groundbreaking research in audio-driven portrait animation.

Hallo4: High-Fidelity Dynamic Portrait Animation via Direct Preference Optimization

Paper: SIGGRAPH Asia 2025
Authors: Jiahao Cui, Baoyou Chen, Mingwang Xu, Hanlin Shang, Yuxuan Chen, Yun Zhan, Zilong Dong, Yao Yao, Jingdong Wang, Siyu Zhu
Institutions: Fudan University, Baidu Inc, Nanjing University, Alibaba Group
Repository: https://github.com/fudan-generative-vision/hallo4
Model: https://huggingface.co/fudan-generative-ai/hallo4
arXiv: https://arxiv.org/abs/2505.23525

Hallo3: Highly Dynamic and Realistic Portrait Image Animation with Diffusion Transformer Networks

Authors: Jiahao Cui, Hui Li, Yun Zhan, et al.
Repository: https://github.com/fudan-generative-vision/hallo3
Model: https://huggingface.co/fudan-generative-ai/hallo3
arXiv: https://arxiv.org/abs/2412.00733

Citation

@misc{cui2025hallo4,
    title={Hallo4: High-Fidelity Dynamic Portrait Animation via Direct Preference Optimization},
    author={Jiahao Cui and Baoyou Chen and Mingwang Xu and Hanlin Shang and
            Yuxuan Chen and Yun Zhan and Zilong Dong and Yao Yao and
            Jingdong Wang and Siyu Zhu},
    year={2025},
    eprint={2505.23525},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

@misc{cui2024hallo3,
    title={Hallo3: Highly Dynamic and Realistic Portrait Image Animation
           with Diffusion Transformer Networks},
    author={Jiahao Cui and Hui Li and Yun Zhang and Hanlin Shang and
            Kaihui Cheng and Yuqi Ma and Shan Mu and Hang Zhou and
            Jingdong Wang and Siyu Zhu},
    year={2024},
    eprint={2412.00733},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Additional Technologies

Edge TTS — Microsoft Edge Text-to-Speech for Eve's voice
AvatarForcing — One-step streaming talking avatars (arXiv:2603.14331)
Wav2Vec2 — Facebook's audio encoder (facebook/wav2vec2-base-960h)
WAN2.1 — Base video generation model (Wan-AI/Wan2.1-T2V-1.3B)
MediaPipe — Google's face mesh detection

License

Hallo4 is a derivative of WAN2.1-1.3B, governed by the WAN LICENSE. Hallo3 is a derivative of CogVideo-5B, released under MIT license.