<div align="center">

UniAnimate: Taming Unified Video Diffusion Models for Consistent Human Image Animation

Xiang Wang1, Shiwei Zhang2, Changxin Gao1, Jiayu Wang2, Xiaoqiang Zhou3, Yingya Zhang2 , Luxin Yan1 , Nong Sang1
1HUST 2Alibaba Group 3USTC

🎨 Project Page

Demo cases generated by the proposed UniAnimate

</div>

🔥 News

[2025/04/15] 🔥🔥🔥🔥🔥 We released a new model based on UniAnimate and Wan2.1, named UniAnimate-DiT, including the training and inference code. Please refer to UniAnimate-DiT for more details.
[2024/07/19] 🔥 We added a noise prior to the code (refer to line 381: noise = diffusion.q_sample(random_ref_frame.clone(), getattr(cfg, "noise_prior_value", 939), noise=noise) in tools/inferences/inference_unianimate_long_entrance.py), which can help achieve better appearance preservation (such as background), especially in long video generation. In addition, we are considering releasing an upgraded version of UniAnimate if we obtain an open source license from the company.
[2024/06/26] For cards with large GPU memory, such as A100 GPU, we support multiple segments parallel denoising to accelerate long video inference. You can change context_batch_size: 1 in configs/UniAnimate_infer_long.yaml to other values greater than 1, such as context_batch_size: 4. The inference speed will be improved to a certain extent.
[2024/06/15] 🔥🔥🔥 By offloading CLIP and VAE and explicitly adding torch.float16 (i.e., set CPU_CLIP_VAE: True in configs/UniAnimate_infer.yaml), the GPU memory can be greatly reduced. Now generating a 32x768x512 video clip only requires ~12G GPU memory. Refer to this issue for more details. Thanks to @blackight for the contribution！
[2024/06/13] 🔥🔥🔥 We released the code and models for human image animation, enjoy it!
[2024/06/13] We have submitted the code to the company for approval, and the code is expected to be released today or tomorrow.
[2024/06/03] We initialized this github repository and planed to release the paper.

TODO

[x] Release the models and inference code, and pose alignment code.
[x] Support generating both short and long videos.
[ ] Release the models for longer video generation in one batch.
[ ] Release models based on VideoLCM for faster video synthesis.
[ ] Training the models on higher resolution videos.

Introduction

Overall framework of UniAnimate

</div>

Recent diffusion-based human image animation techniques have demonstrated impressive success in synthesizing videos that faithfully follow a given reference identity and a sequence of desired movement poses. Despite this, there are still two limitations: i) an extra reference model is required to align the identity image with the main video branch, which significantly increases the optimization burden and model parameters; ii) the generated video is usually short in time (e.g., 24 frames), hampering practical applications. To address these shortcomings, we present a UniAnimate framework to enable efficient and long-term human video generation. First, to reduce the optimization difficulty and ensure temporal coherence, we map the reference image along with the posture guidance and noise video into a common feature space by incorporating a unified video diffusion model. Second, we propose a unified noise input that supports random noised input as well as first frame conditioned input, which enhances the ability to generate long-term video. Finally, to further efficiently handle long sequences, we explore an alternative temporal modeling architecture based on state space model to replace the original computation-consuming temporal Transformer. Extensive experimental results indicate that UniAnimate achieves superior synthesis results over existing state-of-the-art counterparts in both quantitative and qualitative evaluations. Notably, UniAnimate can even generate highly consistent one-minute videos by iteratively employing the first frame conditioning strategy.

Getting Started with UniAnimate

(1) Installation

Installation the python dependencies:

git clone https://github.com/ali-vilab/UniAnimate.git
cd UniAnimate
conda create -n UniAnimate python=3.9
conda activate UniAnimate
conda install pytorch==2.0.1 torchvision==0.15.2 torchaudio==2.0.2 pytorch-cuda=11.8 -c pytorch -c nvidia
pip install -r requirements.txt

We also provide all the dependencies in environment.yaml.

Note: for Windows operating system, you can refer to this issue to install the dependencies. Thanks to @zephirusgit for the contribution. If you encouter the problem of The shape of the 2D attn_mask is torch.Size([77, 77]), but should be (1, 1)., please refer to this issue to solve it, thanks to @Isi-dev for the contribution.

(2) Download the pretrained checkpoints

Download models:

!pip install modelscope
from modelscope.hub.snapshot_download import snapshot_download
model_dir = snapshot_download('iic/unianimate', cache_dir='checkpoints/')

Then you might need the following command to move the checkpoints to the "checkpoints/" directory:

mv ./checkpoints/iic/unianimate/* ./checkpoints/

Finally, the model weights will be organized in ./checkpoints/ as follows:

./checkpoints/
|---- dw-ll_ucoco_384.onnx
|---- open_clip_pytorch_model.bin
|---- unianimate_16f_32f_non_ema_223000.pth 
|---- v2-1_512-ema-pruned.ckpt
└---- yolox_l.onnx

(3) Pose alignment (Important)

Rescale the target pose sequence to match the pose of the reference image:

# reference image 1
python run_align_pose.py  --ref_name data/images/WOMEN-Blouses_Shirts-id_00004955-01_4_full.jpg --source_video_paths data/videos/source_video.mp4 --saved_pose_dir data/saved_pose/WOMEN-Blouses_Shirts-id_00004955-01_4_full 

# reference image 2
python run_align_pose.py  --ref_name data/images/musk.jpg --source_video_paths data/videos/source_video.mp4 --saved_pose_dir data/saved_pose/musk 

# reference image 3
python run_align_pose.py  --ref_name data/images/WOMEN-Blouses_Shirts-id_00005125-03_4_full.jpg --source_video_paths data/videos/source_video.mp4 --saved_pose_dir data/saved_pose/WOMEN-Blouses_Shirts-id_00005125-03_4_full

# reference image 4
python run_align_pose.py  --ref_name data/images/IMG_20240514_104337.jpg --source_video_paths data/videos/source_video.mp4 --saved_pose_dir data/saved_pose/IMG_20240514_104337

We have already provided the processed target pose for demo videos in data/saved_pose, if you run our demo video example, this step can be skipped. In addition, you need to install onnxruntime-gpu (pip install onnxruntime-gpu==1.13.1) to run pose alignment on GPU.

✔ Some tips:

In pose alignment, the first frame in the target pose sequence is used to calculate the scale coefficient of the alignment. Therefore, if the first frame in the target pose sequence contains the entire face and pose (hand and foot), it can help obtain more accurate estimation and better video generation results.

(4) Run the UniAnimate model to generate videos

(4.1) Generating video clips (32 frames with 768x512 resolution)

Execute the following command to generate video clips:

python inference.py --cfg configs/UniAnimate_infer.yaml

After this, 32-frame video clips with 768x512 resolution will be generated:

<table> <center> <tr> <td ><center> <image height="260" src="assets/1.gif"></image> </center></td> <td ><center> <image height="260" src="assets/2.gif"></image> </center></td> </tr> <tr> <td ><center> Click <a href="https://cloud.video.taobao.com/vod/play/cEdJVkF4TXRTOTd2bTQ4andjMENYV1hTb2g3Zlpmb1E/Vk9HZHZkdDBUQzZQZWw1SnpKVVVCTlh4OVFON0V5UUVMUDduY1RJak82VE1sdXdHTjNOaHc9PQ">HERE</a> to view the generated video. </center></td> <td ><center> Click <a href="https://cloud.video.taobao.com/vod/play/cEdJVkF4TXRTOTd2bTQ4andjMENYYzNUUWRKR043c1FaZkVHSkpSMnpoeTZQZWw1SnpKVVVCTlh4OVFON0V5UUVMUDduY1RJak82VE1sdXdHTjNOaHc9PQ">HERE</a> to view the generated video. </center></td> </tr> </center> </table> </center> <!-- <table> <center> <tr> <td ><center> <video height="260" controls autoplay loop src="https://cloud.video.taobao.com/vod/play/cEdJVkF4TXRTOTd2bTQ4andjMENYYzNUUWRKR043c1FaZkVHSkpSMnpoeTZQZWw1SnpKVVVCTlh4OVFON0V5UUVMUDduY1RJak82VE1sdXdHTjNOaHc9PQ" muted="false"></video> </td> <td ><center> <video height="260" controls autoplay loop src="https://cloud.video.taobao.com/vod/play/cEdJVkF4TXRTOTd2bTQ4andjMENYV1hTb2g3Zlpmb1E/Vk9HZH

UniAnimate

Install / Use

README