---
license: apache-2.0
language:
- en
tags:
- diffusion model
- stable diffusion
- SCEdit
- Scepter
- Scepter studio
- Controllable
- ControlNet
- Lora
---
<p align="center">

  <h2 align="center">🪄SCEdit: Efficient and Controllable Image Diffusion Generation via Skip Connection Editing</h2>
  <p align="center">
    <strong>Zeyinzi Jiang</strong>
    ·
    <strong>Chaojie Mao</strong>
    ·
    <strong>Yulin Pan</strong>
    ·
    <strong>Zhen Han</strong>
    ·
    <strong>Jingfeng Zhang</strong>
    <br>
    <br>
        <a href="https://arxiv.org/abs/2312.11392"><img src='https://img.shields.io/badge/arXiv-SCEdit-red' alt='Paper PDF'></a>
        <a href='https://scedit.github.io/'><img src='https://img.shields.io/badge/Project_Page-SCEdit-green' alt='Project Page'></a>
        <a href='https://github.com/modelscope/scepter'><img src='https://img.shields.io/badge/scepter-SCEdit-yellow'></a>
        <a href='https://github.com/modelscope/swift'><img src='https://img.shields.io/badge/swift-SCEdit-blue'></a>
    <br>
    <b>Alibaba Group</b>
  </p>
  
  <table align="center">
    <tr>
    <td>
      <img src="assets/figures/show.jpg">
    </td>
    </tr>
  </table>

SCEdit is an efficient generative fine-tuning framework proposed by Alibaba TongYi Vision Intelligence Lab. This framework enhances the fine-tuning capabilities for text-to-image generation downstream tasks and enables quick adaptation to specific generative scenarios, **saving 30%-50% of training memory costs compared to LoRA**. Furthermore, it can be directly extended to controllable image generation tasks, **requiring only 7.9% of the parameters that ControlNet needs for conditional generation and saving 30% of memory usage**. It supports various conditional generation tasks including edge maps, depth maps, segmentation maps, poses, color maps, and image completion.
#### Code Example
```shell
git clone https://github.com/modelscope/scepter.git
cd scepter
PYTHONPATH=. python scepter/tools/run_train.py --cfg scepter/methods/SCEdit/t2i_sdxl_1024_sce.yaml
```

To prepare  the training dataset.

```python
# pip install modelscope
from modelscope.msdatasets import MsDataset
ms_train_dataset = MsDataset.load('style_custom_dataset', namespace='damo', subset_name='3D', split='train_short')
print(next(iter(ms_train_dataset)))
```

## BibTeX

```bibtex
@article{jiang2023scedit,
    title = {SCEdit: Efficient and Controllable Image Diffusion Generation via Skip Connection Editing},
    author = {Jiang, Zeyinzi and Mao, Chaojie and Pan, Yulin and Han, Zhen and Zhang, Jingfeng},
    year = {2023},
    journal = {arXiv preprint arXiv:2312.11392}  
}
```