ComfyUI_MangaNinjia

ComfyUI_MangaNinjia is a ComfyUI node of MangaNinja which‌ is a " Line Art Colorization with Precise Reference Following " method。

update

• 拆解流程，将峰值显存占用压到8G左右,8G显存是否能跑，要其他用户测试一下才行。
• Disassemble the process and reduce the peak memory usage to around 8GB. Other users need to test whether 8GB of memory can run.

1. Installation

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

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

---

2. Requirements

• no need, because it's base in sd1.5 and controlnet,Perhaps someone may be missing the library.没什么特殊的库,懒得删了

pip install -r requirements.txt

3. Models

• sd1.5 address v1-5-pruned-emaonly.safetensors #example
• controlnet lineart address control_v11p_sd15_lineart_fp16.safetensors #example
• Annotators address sk_model.pth #example
• MangaNinjia address #example
• clipvison address #clip-vit-large-patch14 #有空改成单体

--  ComfyUI/models/checkpoints
    |-- any sd1.5 safetensors #任意sd1.5模型，注意要带vae的
--  ComfyUI/models/controlnet
    |-- control_v11p_sd15_lineart_fp16.safetensors or control_v11p_sd15s2_lineart_anime_fp16.safetensors
--  ComfyUI/models/MangaNinjia
        |-- denoising_unet.pth
        |-- reference_unet.pth
        |-- point_net.pthnaz
        |-- controlnet.pth
        |-- sk_model.pth
-- any/path/clip-vit-large-patch14   #or auto download 
        |--config.json
        |--tokenizer.json
        |--tokenizer_config.json
        |--vocab.json
        |--special_tokens_map.json
        |--preprocessor_config.json
        |--model.safetensors

4.Tips

• is_lineart : if True you can link a normal image at lineart_image. 开启is_lineart，会自动预处理图片为线稿

5.Example

6.Citation

@article{liu2025manganinja,
  title={MangaNinja: Line Art Colorization with Precise Reference Following},
  author={Liu, Zhiheng and Cheng, Ka Leong and Chen, Xi and Xiao, Jie and Ouyang, Hao and Zhu, Kai and Liu, Yu and Shen, Yujun and Chen, Qifeng and Luo, Ping},
  journal={arXiv preprint arXiv:2501.08332},
  year={2025}
}