ali-vilab / MimicBrush

Official implementations for paper: Zero-shot Image Editing with Reference Imitation

Zero-shot Image Editing with Reference Imitation

Xi Chen · Yutong Feng · Mengting Chen · Yiyang Wang · Shilong Zhang · Yu Liu · Yujun Shen · Hengshuang Zhao


The University of Hong Kong   |   Alibaba Group |   Ant Group

News

• [2024.06.12] Release inference code, local gradio demo, online demo.
• [Todo] Release our benchmark.

Community Contributions

ComfyUI version by @AIFSH

Installation

Install with conda:

conda env create -f environment.yaml
conda activate mimicbrush

or pip:

#Python==3.8.5
pip install -r requirements.txt

Download Checkpoints

Download SD-1.5 and SD-1.5-inpainting checkpoint:

• You could download them from HuggingFace stable-diffusion-v1-5 and stable-diffusion-inpainting
• However, the repo above contains many models that would not be used, we provide a clean version at cleansd

Download MimicBrush checkpoint, along with a VAE, a CLIP encoder, and a depth model

• Download the weights on ModelScope xichen/MimicBrush
• The model is big because it contains two U-Nets.

You could use the following code to download them from modelscope

from modelscope.hub.snapshot_download import snapshot_download as ms_snapshot_download

sd_dir = ms_snapshot_download('xichen/cleansd', cache_dir='./modelscope')
print('=== Pretrained SD weights downloaded ===')
model_dir = ms_snapshot_download('xichen/MimicBrush', cache_dir='./modelscope')
print('=== MimicBrush weights downloaded ===')

or from Huggingface

from huggingface_hub import snapshot_download
snapshot_download(repo_id="xichenhku/cleansd", local_dir="./cleansd")
print('=== Pretrained SD weights downloaded ===')
snapshot_download(repo_id="xichenhku/MimicBrush", local_dir="./MimicBrush")
print('=== MimicBrush weights downloaded ===')

Gradio Demo

First, modify ./configs/inference.yaml to set the path of model weight. Afterwards, run the script:

python run_gradio3_demo.py

The gradio demo would look like the UI shown below.

*Please do not forget to click ''keep the original shape'' if you want condut texture transfer like the third case.

A biref tutorial:

• Upload/select a source image to edit.
• Draw the to-edit regionon the source image.
• Upload/select a reference image.
• Run.

Inference

1. Dowload our evaluation benchmark at Google Drive:

   • URL: [to be released]

2. Set the path to each dataset and checkpoints in ./config/inference.yaml :

3. Run inference with

   python run_inference_benchmark.py

Acknowledgements

This project is developped on the codebase of IP-Adapter and MagicAnimate . We appreciate this great work!

Citation

If you find this codebase useful for your research, please use the following entry.

@article{chen2024mimicbrush,
  title={Zero-shot Image Editing with Reference Imitation},
  author={Chen, Xi and Feng, Yutong and Chen, Mengting and Wang, Yiyang, and Zhang, Shilong and Yu, Liu and Shen, Yujun and Zhao, Hengshuang},
  journal={arXiv preprint arXiv:2406.07547},
  year={2024}
}

Star History