MuseV: Infinite-length and High Fidelity Virtual Human Video Generation with Visual Conditioned Parallel Denoising
Zhiqiang Xia ^*, Zhaokang Chen^*, Bin Wu^†, Chao Li, Kwok-Wai Hung, Chao Zhan, Yingjie He, Wenjiang Zhou (^*Equal Contribution, ^†Corresponding Author, benbinwu@tencent.com)
Lyra Lab, Tencent Music Entertainment

github huggingface HuggingfaceSpace [project](comming soon) Technical report (comming soon)

We have setup the world simulator vision since March 2023, believing diffusion models can simulate the world. MuseV was a milestone achieved around July 2023. Amazed by the progress of Sora, we decided to opensource MuseV, hopefully it will benefit the community. Next we will move on to the promising diffusion+transformer scheme.

We will soon release MuseTalk, a real-time high quality lip sync model, which can be applied with MuseV as a complete virtual human generation solution. Please stay tuned!

Overview

MuseV is a diffusion-based virtual human video generation framework, which

1. supports infinite length generation using a novel Visual Conditioned Parallel Denoising scheme.
2. checkpoint available for virtual human video generation trained on human dataset.
3. supports Image2Video, Text2Image2Video, Video2Video.
4. compatible with the Stable Diffusion ecosystem, including base_model, lora, controlnet, etc.
5. supports multi reference image technology, including IPAdapter, ReferenceOnly, ReferenceNet, IPAdapterFaceID.
6. training codes (comming very soon).

News

• [03/27/2024] release MuseV project and trained model musev, muse_referencenet, muse_referencenet_pose.

Model

overview of model structure

parallel denoising

Cases

All frames are generated from text2video model, without any post process.

Bellow Case could be found in configs/tasks/example.yaml

Text/Image2Video

Human

scene

VideoMiddle2Video

pose2video

In duffy case, pose of the vision condition frame is not aligned with of the first frame of control video. posealign module can solve it.

MuseTalk

Quickstart

please refer to MuseV

Acknowledgements

1. MuseV has referred much to TuneAVideo, diffusers, Moore-AnimateAnyone, animatediff, IP-Adapter, AnimateAnyone, VideoFusion, insightface.
2. MuseV has been built on ucf101 and webvid datasets.

Thanks for open-sourcing!

Limitation

There are still many limitations, including

1. Lack of generalization ability. Some visual condition image perform well, some perform bad. Some t2i pretraied model perform well, some perform bad.
2. Limited types of video generation and limited motion range, partly because of limited types of training data. The released MuseV has been trained on approximately 60K human text-video pairs with resolution 512*320. MuseV has greater motion range while lower video quality at lower resolution. MuseV tends to generate less motion range with high video quality. Trained on larger, higher resolution, higher quality text-video dataset may make MuseV better.
3. Watermarks may appear because of webvid. A cleaner dataset withour watermarks may solve this issue.
4. Limited types of long video generation. Visual Conditioned Parallel Denoise can solve accumulated error of video generation, but the current method is only suitable for relatively fixed camera scenes.
5. Undertrained referencenet and IP-Adapter, beacause of limited time and limited resources.
6. Understructured code. MuseV supports rich and dynamic features, but with complex and unrefacted codes. It takes time to familiarize.

Citation

@article{musev,
  title={MuseV: Infinite-length and High Fidelity Virtual Human Video Generation with Visual Conditioned Parallel Denoising},
  author={Xia, Zhiqiang and Chen, Zhaokang and Wu, Bin and Li, Chao and Hung, Kwok-Wai and Zhan, Chao and He, Yingjie and Zhou, Wenjiang},
  journal={arxiv},
  year={2024}
}

Disclaimer/License

1. code: The code of MuseV is released under the MIT License. There is no limitation for both academic and commercial usage.
2. model: The trained model are available for non-commercial research purposes only.
3. other opensource model: Other open-source models used must comply with their license, such as insightface, IP-Adapter, ft-mse-vae, etc.
4. The testdata are collected from internet, which are available for non-commercial research purposes only.
5. AIGC: This project strives to impact the domain of AI-driven video generation positively. Users are granted the freedom to create videos using this tool, but they are expected to comply with local laws and utilize it responsibly. The developers do not assume any responsibility for potential misuse by users.