This repository is the official implementation of SEINE:

SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction (ICLR2024)

SEINE is a video diffusion model and is part of the video generation system Vchitect. You can also check our Text-to-Video (T2V) framework LaVie.

conda create -n seine python==3.9.16
conda activate seine
pip install -r requirement.txt

Our model is based on Stable diffusion v1.4, you may download Stable Diffusion v1-4 to the director of pretrained . Download our model checkpoint (from google drive or hugging face) and save to the directory of pretrained

Now under ./pretrained, you should be able to see the following:

├── pretrained
│   ├── seine.pt
│   ├── stable-diffusion-v1-4
│   │   ├── ...
└── └── ├── ...
        ├── ...

Run the following command to get the I2V results:

python sample_scripts/with_mask_sample.py --config configs/sample_i2v.yaml

The generated video will be saved in ./results/i2v.

You may modify ./configs/sample_i2v.yaml to change the generation conditions. For example:

ckpt is used to specify a model checkpoint.

text_prompt is used to describe the content of the video.

input_path is used to specify the path to the image.

python sample_scripts/with_mask_sample.py --config configs/sample_transition.yaml

The generated video will be saved in ./results/transition.

Input Image	Output Video

Input Images		Output Video

@article{chen2023seine,
title={SEINE: Short-to-Long Video Diffusion Model for Generative Transition and Prediction},
author={Chen, Xinyuan and Wang, Yaohui and Zhang, Lingjun and Zhuang, Shaobin and Ma, Xin and Yu, Jiashuo and Wang, Yali and Lin, Dahua and Qiao, Yu and Liu, Ziwei},
journal={arXiv preprint arXiv:2310.20700},
year={2023}
}

@article{wang2023lavie,
  title={LAVIE: High-Quality Video Generation with Cascaded Latent Diffusion Models},
  author={Wang, Yaohui and Chen, Xinyuan and Ma, Xin and Zhou, Shangchen and Huang, Ziqi and Wang, Yi and Yang, Ceyuan and He, Yinan and Yu, Jiashuo and Yang, Peiqing and others},
  journal={arXiv preprint arXiv:2309.15103},
  year={2023}
}

We disclaim responsibility for user-generated content. The model was not trained to realistically represent people or events, so using it to generate such content is beyond the model's capabilities. It is prohibited for pornographic, violent and bloody content generation, and to generate content that is demeaning or harmful to people or their environment, culture, religion, etc. Users are solely liable for their actions. The project contributors are not legally affiliated with, nor accountable for users' behaviors. Use the generative model responsibly, adhering to ethical and legal standards.

Xinyuan Chen: [email protected] Yaohui Wang: [email protected]

The code is built upon LaVie, diffusers and Stable Diffusion, we thank all the contributors for open-sourcing.

The code is licensed under Apache-2.0, model weights are fully open for academic research and also allow free commercial usage. To apply for a commercial license, please contact [email protected].