ComfyUI_MS_Diffusion

MS-Diffusion multi-subject zero-shot image personalization with layout guidance

Upadtes

借助agent（自动）使用comfyUI核心重新封装此方法 / Re-implement this method with the help of agent to re-package it into the core of comfyUI

Installation

In the ./ComfyUI/custom_nodes directory, run the following:

git clone https://github.com/smthemex/ComfyUI_MS_Diffusion

Models

sdxl ms-ipadapter

├── ComfyUI/models/checkpoints
|     ├── sd_xl_base_1.0.safetensors #any sdxl
├── ComfyUI/models/photomaker #ipadapter
|        ├──ms_adapter.bin

Multi-subject (multiple reference images)

#	Sequence	Example (2 subjects)
1	IMAGE batch order (native `ImageBatch`: image1, image2, ...)	`ms_dog.jpg` → image1, `ms_cat.jpg` → image2
2	`[subject]` order in `prompt`	`a [dog] and a [cat] on the beach`
3	`boxes`, `;`-separated, one `x1,y1,x2,y2` each	`0., 0.25, 0.4, 0.75;0.6, 0.25, 1., 0.75`

Example

Citation

@inproceedings{
  wang2025msdiffusion,
  title={{MS}-Diffusion: Multi-subject Zero-shot Image Personalization with Layout Guidance},
  author={Xierui Wang and Siming Fu and Qihan Huang and Wanggui He and Hao Jiang},
  booktitle={The Thirteenth International Conference on Learning Representations},
  year={2025},
  url={https://openreview.net/forum?id=PJqP0wyQek}
}