Safe-Sora

KO | EN

gitlite — search

#react #javascript #css #java #android #nodejs #gulp #webpack #docker #swift #machine-learning #ios

★ 20 Open GitHub ↗

[NeurIPS 2025] The official implementation of paper "Safe-Sora: Safe Text-to-Video Generation via Graphical Watermarking"

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

<h2 align="center">Safe-Sora: Safe Text-to-Video Generation via Graphical Watermarking</h2>

<a href="https://github.com/Sugewud">Zihan Su1</a>,
 <a href="https://scholar.google.com/citations?user=bMwW4e8AAAAJ&hl=zh-CN">Xuerui Qiu2</a>,
 <a href="https://scholar.google.com/citations?hl=zh-CN&user=mRC_emoAAAAJ">Hongbin Xu3</a>,
 <a href="https://dblp.org/pid/369/7675.html">Tangyu Jiang1</a>,
 <a href="https://scholar.google.co.in/citations?user=J3olRccAAAAJ&hl=en">Junhao Zhuang1</a>, 
 <a href="https://scholar.google.com/citations?hl=en&user=fYdxi2sAAAAJ">Chun Yuan1†</a>,
 <a href="https://scholar.google.com/citations?user=2wySPkcAAAAJ&hl=zh-CN">Ming Li4†</a>,
 <a href="https://scholar.google.com/citations?user=rBWnK8wAAAAJ&hl=en">Shengfeng He5</a>,
 <a href="https://dblp.org/pid/16/6654.html">Fei Richard Yu4</a>

1 Tsinghua University 
 2 Institute of Automation, Chinese Academy of Sciences 
 3 South China University of Technology 
 4 Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) 
 5 Singapore Management University 
 †Corresponding Author

</div>

## Release
- [09/19] 🚀 🚀  Code Released! 
- [09/18] 🎉 🎉 Safa-Sora is accepted by NeurIPS 2025! 
- [05/23] Initial Preview Release 🔥 Coming Soon!

## 🔆 Introduction
Safe-Sora is the first framework that integrates graphical watermarks directly into the video generation process.
 
<img src="fig/Application.png" style="width:100%;">
 
The following results show the original video, the watermarked video, the difference between them (×5), the original watermark, the recovered watermark, and the difference between them (×5).

## 📋 File Preparation

Download the files and place them in the root directory.

- [checkpoints](https://drive.google.com/file/d/1ciKIafSQFbulL3xvQjARYPxxnvnpXGaI/view?usp=sharing) contains the pretrained weights for Safe-Sora, VideoCrafter2, the VAE, and the 3D-CNN (simulating H.264 compression).
- [dataset](https://drive.google.com/file/d/1-aIaPyhkufrwWMVva_bAZ2I4kh-_OMII/view?usp=sharing) contains the Logo-2K dataset and the Panda-70M dataset.
- [mamba](https://drive.google.com/file/d/1fnm67gq_0xfTWpTG08ByKXlK4wfG39Qz/view?usp=sharing) is provided for setting up the environment.
- [causal-conv1d](https://drive.google.com/file/d/1RIR4i9MyCfgnmeXrjhQOBR0bkIV-VDFT/view?usp=sharing) is provided for setting up the environment.

## 💻 Requirements

To install requirements:

```setup
conda create -n safe-sora python=3.9
conda activate safe-sora
conda install pytorch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 pytorch-cuda=11.8 -c pytorch -c nvidia

pip install packaging ninja==1.11.1.1
cd causal-conv1d
python setup.py install
cd ../mamba
python setup.py install

cd ..
pip install -r requirements.txt
```

## 🐶 Training
To train Safe-Sora, run this command:

```train
bash train.sh 
```

We use DDP to train Safe-Sora. You can modify the parameters in train.sh to select which GPUs to use.

## 🚀 Evaluation

To evaluate our models, run:

```eval
bash test.sh 
```

## 🙌🏻 Acknowledgement
Our code is based on these awesome repos:
* [Mamba](https://github.com/state-spaces/mamba)
* [RainMamba](https://github.com/TonyHongtaoWu/RainMamba)
* [FreqMamba](https://github.com/aSleepyTree/FreqMamba)

## 📖 BibTeX
If you find our repo helpful, please consider leaving a star or cite our paper :)
```bibtex
@article{su2025safe,
  title={Safe-Sora: Safe Text-to-Video Generation via Graphical Watermarking},
  author={Su, Zihan and Qiu, Xuerui and Xu, Hongbin and Jiang, Tangyu and Zhuang, Junhao and Yuan, Chun and Li, Ming and He, Shengfeng and Yu, Fei Richard},
  journal={arXiv preprint arXiv:2505.12667},
  year={2025}
}
```

Safe-Sora

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README