MuseV

MuseV: Infinite-length and High Fidelity Virtual Human Video Generation with Parallel Denoising
Zhiqiang Xia ^*, Zhaokang Chen^*, Bin Wu^†, Chao Li, Kwok-Wai Hung, Chao Zhan, Wenjiang Zhou (^*co-first author, ^†Corresponding Author)

[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.

Our next move will switch to the promising diffusion+transformer scheme. Please stay tuned.

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

Intro

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

1. supports infinite length generation using a novel 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).

Model

overview of model structure

parallel denoise

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

)

image	video	prompt
	yongen_c.mp4	(masterpiece, best quality, highres:1),(1girl, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	jinkesi2.mp4	(masterpiece, best quality, highres:1),(1girl, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	seaside4_c.mp4	(masterpiece, best quality, highres:1), peaceful beautiful sea scene
	real_girl_seaside2_c.mp4	(masterpiece, best quality, highres:1), peaceful beautiful sea scene
	seaside_girl.mp4	(masterpiece, best quality, highres:1), peaceful beautiful sea scene
	boy_play_guitar.mp4	(masterpiece, best quality, highres:1), playing guitar
	girl_play_guitar2_c.mp4	(masterpiece, best quality, highres:1), playing guitar
	boy_play_guitar2_c.mp4	(masterpiece, best quality, highres:1), playing guitar
	girl_play_guitar4_c.mp4	(masterpiece, best quality, highres:1), playing guitar
	dufu.mp4	(masterpiece, best quality, highres:1),(1man, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	Mona_Lisa._by_Leonardo_da_Vinci._from_C2RMF_retouched_c.mp4	(masterpiece, best quality, highres:1),(1girl, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	Portrait-of-Dr.-Gachet_c.mp4	(masterpiece, best quality, highres:1),(1man, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	Self-Portrait-with-Cropped-Hair.mp4	(masterpiece, best quality, highres:1),(1girl, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)
	The-Laughing-Cavalier_c.mp4	(masterpiece, best quality, highres:1),(1girl, solo:1),(beautiful face, soft skin, costume:1),(eye blinks:{eye_blinks_factor}),(head wave:1.3)

scene

image	video	prompt
	waterfall4_c.mp4	(masterpiece, best quality, highres:1), peaceful beautiful waterfall, an endless waterfall
	river.mp4	(masterpiece, best quality, highres:1), peaceful beautiful river
	seaside2_c.mp4	(masterpiece, best quality, highres:1), peaceful beautiful sea scene

VideoMiddle2Video

Video2Video

News

• [03/22/2024] release MuseV project and trained model musev, muse_referencenet.

TODO:

• technical report (comming soon).
• training codes.
• release pretrained unet model, which is trained with controlnet、referencenet、IPAdapter, which is better on pose2video.
• support diffusion transformer generation framework.

Quickstart

prepare python environment and install extra package like diffusers, controlnet_aux, mmcm.

suggest to use docker primarily to prepare python environment.

prepare environment

docker

1. pull docker image

docker pull docker pull anchorxia/musev:latest

2. run docker

docker run --gpus all -it --entrypoint /bin/bash anchorxia/musev:latest

The default conda env is musev.

conda

create conda environment from environment.yaml

conda env create --name musev --file ./environment.yml

requirements

pip install -r requirements.txt

install mmlab package

pip install--no-cache-dir -U openmim 
mim install mmengine 
mim install "mmcv>=2.0.1" 
mim install "mmdet>=3.1.0" 
mim install "mmpose>=1.1.0" 

custom package / modified package

git clone --recursive https://github.com/TMElyralab/MuseV.git

prepare PYTHONPATH

current_dir=$(pwd)
export PYTHONPATH=${PYTHONPATH}:${current_dir}/MuseV
export PYTHONPATH=${PYTHONPATH}:${current_dir}/MuseV/MMCM
export PYTHONPATH=${PYTHONPATH}:${current_dir}/MuseV/diffusers/src
export PYTHONPATH=${PYTHONPATH}:${current_dir}/MuseV/controlnet_aux/src
cd MuseV

1. MMCM: multi media, cross modal process package。
2. diffusers: modified diffusers package based on diffusers
3. controlnet_aux: modified based on controlnet_aux

download models

git clone https://huggingface.co/TMElyralab/MuseV ./checkpoints

• motion: text2video model.
  • musev/unet: only has and train unet motion module.
  • musev_referencenet: train unet module, referencenet, IPAdapter
    • unet: motion module, which has to_k, to_v in Attention layer refer to IPAdapter
    • referencenet: similar to AnimateAnyone
    • ip_adapter_image_proj.bin: images clip emb project layer, refer to IPAdapter
• t2i/sd1.5: text2image model, paramter are frozen when training motion module.
  • majicmixRealv6Fp16: example, could be replaced with other t2i base. download from majicmixRealv6Fp16
• IP-Adapter/models: download from IPAdapter
  • image_encoder: vision clip model.
  • ip-adapter_sd15.bin: original IPAdapter model checkpoint.
  • ip-adapter-faceid_sd15.bin: original IPAdapter model checkpoint.

Inference

prepare model_path

skip this step when run example task with example inference command. set model path and abbreviation in config, to use abbreviation in inference script.

• T2I SD：ref to musev/configs/model/T2I_all_model.py
• Motion Unet: refer to musev/configs/model/motion_model.py
• task: refer to musev/configs/tasks/example.yaml

musev_referencenet

text2video

python scripts/inference/text2video.py   --sd_model_name majicmixRealv6Fp16   --unet_model_name musev_referencenet --referencenet_model_name musev_referencenet --ip_adapter_model_name musev_referencenet   -test_data_path ./configs/tasks/example.yaml  --output_dir ./output  --n_batch 1  --target_datas yongen  --vision_clip_extractor_class_name ImageClipVisionFeatureExtractor --vision_clip_model_path ./checkpoints/IP-Adapter/models/image_encoder  --time_size 12 --fps 12  

common parameters:

• test_data_path: task_path in yaml extention
• target_datas: sep is ,, sample subtasks if name in test_data_path is in target_datas.
• sd_model_cfg_path: T2I sd models path, model config path or model path.
• sd_model_name: sd model name, which use to choose full model path in sd_model_cfg_path. multi model names with sep =,, or all
• unet_model_cfg_path: motion unet model config path or model path。
• unet_model_name: unet model name, use to get model path in unet_model_cfg_path, and init unet class instance in musev/models/unet_loader.py. multi model names with sep=,, or all. If unet_model_cfg_path is model path, unet_name must be supported in musev/models/unet_loader.py
• time_size: num_frames per diffusion denoise generation。default=12.
• n_batch: generation numbers. Total_frames=n_batch*time_size+n_viscond, default=1。
• context_frames: context_frames num. If time_size > context_frame，time_size window is split into many sub-windows for parallel denoising"。 default=12。

model parameters： support referencenet, IPAdapter, IPAdapterFaceID, Facein.

• referencenet_model_name: referencenet model name.
• ImageClipVisionFeatureExtractor: ImageEmbExtractor name, extractor vision clip emb used in IPAdapter.
• vision_clip_model_path: ImageClipVisionFeatureExtractor model path.
• ip_adapter_model_name: from IPAdapter, it's ImagePromptEmbProj, used with ImageEmbExtractor。
• ip_adapter_face_model_name: IPAdapterFaceID, from IPAdapter to keep faceid，should set face_image_path。

Some parameters that affect the amplitude and effect of generation：

• video_guidance_scale: similar to text2image, control influence between cond and uncond，default=3.5
• guidance_scale: The parameter ratio in the first frame image between cond and uncond, default=3.5
• use_condition_image: Whether use the given first frame for video generation.
• redraw_condition_image: whether redraw the given first frame image.
• video_negative_prompt: abbreviation of full negative_prompt in config path. default=V2.

video2video

python scripts/inference/video2video.py --sd_model_name majicmixRealv6Fp16  --unet_model_name musev_referencenet --referencenet_model_name   musev_referencenet --ip_adapter_model_name musev_referencenet    -test_data_path ./configs/tasks/example.yaml    --vision_clip_extractor_class_name ImageClipVisionFeatureExtractor --vision_clip_model_path ./checkpoints/IP-Adapter/models/image_encoder      --output_dir ./output  --n_batch 1 --controlnet_name dwpose_body_hand  --which2video "video_middle"  --target_datas  bilibili_queencard   --fps 12 --time_size 12

import parameters

Most of paramters are same as musev_text2video. Special parameters of video2video are

1. need set video_path in test_data. Now support rgb video and controlnet_middle_video。

• need_video2video: whether rgb video influence initial noise.
• controlnet_name：whether use controlnet condition, such as dwpose,depth.
• video_is_middle: video_path is rgb video or controlnet_middle_video. could set for every test_data in test_data_path.
• video_has_condition: whether condtion_images is aligned with the first frame of video_path. If Not, firstly generate condition_images and align with concatation. set in test_data。

musev

text2video

python scripts/inference/text2video.py   --sd_model_name majicmixRealv6Fp16   --unet_model_name musev   -test_data_path ./configs/tasks/example.yaml  --output_dir ./output  --n_batch 1  --target_datas yongen  --time_size 12 --fps 12

video2video

python scripts/inference/video2video.py --sd_model_name majicmixRealv6Fp16  --unet_model_name musev    -test_data_path ./configs/tasks/example.yaml --output_dir ./output  --n_batch 1 --controlnet_name dwpose_body_hand  --which2video "video_middle"  --target_datas  bilibili_queencard   --fps 12 --time_size 12

Gradio demo

MuseV provides gradio script to generate GUI in local machine to generate video conveniently.

cd scripts/gradio
python app.py

Acknowledgements

MuseV builds on TuneAVideo, diffusers. Thanks for open-sourcing!

Citation

paper comming soon

@article{musev,
  title={MuseV: Infinite-length and High Fidelity Virtual Human Video Generation with Parallel Denoising},
  author={Xia, Zhiqiang and Chen, Zhaokang and Wu, Bin and Li, Chao and Hung, Kwok-Wai and Zhan, Chao 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. 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.