Taming Rectified Flow for Inversion and Editing

Jiangshan Wang^1,2, Junfu Pu², Zhongang Qi², Jiayi Guo¹, Yue Ma³,
Nisha Huang¹, Yuxin Chen², Xiu Li¹, Ying Shan²

¹ Tsinghua University, ² Tencent ARC Lab, ³ HKUST

We propose RF-Solver to solve the rectified flow ODE with less error, thus enhancing both sampling quality and inversion-reconstruction accuracy for rectified-flow-based generative models. Furthermore, we propose RF-Edit to leverage the RF-Solver for image and video editing tasks. Our methods achieve impressive performance on various tasks, including text-to-image generation, image/video inversion, and image/video editing.

🔥 News

[2024.11.18] More examples for style transfer are available!
[2024.11.18] Gradio Demo for image editing is available!
[2024.11.11] The homepage of the project is available!
[2024.11.08] Code for image editing is released!
[2024.11.08] Paper released!

👨‍💻 ToDo

☑️ Release the gradio demo
☑️ Release scripts to for more image editing cases
☐ Release the code for video editing

📖 Method

RF-Solver

We derive the exact formulation of the solution for Rectified Flow ODE. The non-linear part in this solution is processed by Taylor Expansion. Through higher order expansion, the approximation error in the solution is significantly reduced, thus achieving impressive performance on both text-to-image sampling and image/video inversion.

RF-Edit

Based on RF-Solver, we further propose the RF-Edit for image and video editing. RF-Edit framework leverages the features from inversion in the denoising process, which enables high-quality editing while preserving the structual information of source image/video. RF-Edit contains two sub-modules, espectively for image editing and video editing.

🛠️ Code Setup

The environment of our code is the same as FLUX, you can refer to the official repo of FLUX, or running the following command to construct the environment.

conda create --name RF-Solver-Edit python=3.10
conda activate RF-Solver-Edit
pip install -e ".[all]"

🚀 Examples for Image Editing

We have provided several scripts to reproduce the results in the paper, mainly including 3 types of editing: Stylization, Adding, Replacing. We suggest to run the experiment on a single A100 GPU.

Stylization

Ref Style
Editing Scripts	Trump	Marilyn Monroe	Einstein
Edtied image
Editing Scripts	Biden	Batman	Herry Potter
Edtied image

Adding & Replacing

Source image
Editing Scripts	+ hiking stick	horse -> camel	+ dog
Edtied image

🪄 Edit Your Own Image

Gradio Demo

We privide the gradio demo for image editing. Run the following command:

cd src
python gradio_demo.py

Here is an example for using the gradio demo to edit an image! Note that here "Number of inject steps" means the steps of feature sharing in RF-Edit, which is highly related to the quality of edited results. We suggest to tune this parameter, selecting the results with best visual quality.

Command Line

You can also run the following scripts to edit your own image.

cd src
python edit.py  --source_prompt [describe the content of your image or leaves it as null] \
                --target_prompt [describe your editing requirements] \
                --guidance 2 \
                --source_img_dir [the path of your source image] \
                --num_steps 30  \
                --inject [typically set to a number between 2 to 8] \
                --name 'flux-dev' --offload \
                --output_dir [output path]

Similarly, The --inject refers to the steps of feature sharing in RF-Edit, which is highly related to the performance of editing.

🖼️ Gallery

Inversion and Reconstruction

Image Stylization

Image Editing

Video Editing

🖋️ Citation

If you find our work helpful, please star 🌟 this repo and cite 📑 our paper. Thanks for your support!

@article{wang2024taming,
  title={Taming Rectified Flow for Inversion and Editing},
  author={Wang, Jiangshan and Pu, Junfu and Qi, Zhongang and Guo, Jiayi and Ma, Yue and Huang, Nisha and Chen, Yuxin and Li, Xiu and Shan, Ying},
  journal={arXiv preprint arXiv:2411.04746},
  year={2024}
}

Acknowledgements

We thank FLUX for their clean codebase.

Contact

The code in this repository is still being reorganized. Errors that may arise during the organizing process could lead to code malfunctions or discrepancies from the original research results. If you have any questions or concerns, please send email to wjs23@mails.tsinghua.edu.cn.