60 lines
1.8 KiB
Markdown
60 lines
1.8 KiB
Markdown
# ComfyUI_InstantIR_Wrapper
|
||
You can InstantIR to Fix blurry photos in ComfyUI
|
||
[InstantIR](https://github.com/instantX-research/InstantIR):Blind Image Restoration with Instant Generative Reference
|
||
喜欢这个项目的，请给InstantIR项目个星星！（If you like this project, please give the InstantIR project a star!）
|
||
|
||
----
|
||
|
||
Update
|
||
-----
|
||
**2024-11-11**
|
||
* change some code for lowram,The inference speed of 4070 12G is 20 times faster than before(21s at 20 steps );
|
||
* 修改了一些代码，目前4070 12G的推理速度比之前演示的快20倍（感觉这个才是正常的，显存峰值10G左右,20步跑完20秒左右)
|
||
|
||
|
||
1.Installation
|
||
-----
|
||
In the ./ComfyUI /custom_node directory, run the following:
|
||
```
|
||
git clone https://github.com/smthemex/ComfyUI_InstantIR_Wrapper.git
|
||
```
|
||
2.requirements
|
||
----
|
||
```
|
||
pip install -r requirements.txt
|
||
```
|
||
|
||
3.checkpoints
|
||
----
|
||
3.1 any SDXL checkpoint
|
||
3.2 InstantIR checkpoints [InstantIR](https://huggingface.co/InstantX/InstantIR)
|
||
```
|
||
├── ComfyUI/models/InstantIR/models
|
||
| ├── adapter.pt
|
||
| ├── aggregator.pt
|
||
| ├──previewer_lora_weights.bin
|
||
```
|
||
3.3 dino: online or any local path [dinov2-large](https://huggingface.co/facebook/dinov2-large)
|
||
3.4 lcm lora [lcm-lora-sdxl](https://huggingface.co/latent-consistency/lcm-lora-sdxl)
|
||
|
||
----
|
||
|
||
4.Example
|
||
----
|
||
|
||
|
||
|
||
----
|
||
|
||
5.Citation
|
||
------
|
||
|
||
**instantX-research/InstantIR**
|
||
``` python
|
||
@article{huang2024instantir,
|
||
title={InstantIR: Blind Image Restoration with Instant Generative Reference},
|
||
author={Huang, Jen-Yuan and Wang, Haofan and Wang, Qixun and Bai, Xu and Ai, Hao and Xing, Peng and Huang, Jen-Tse},
|
||
journal={arXiv preprint arXiv:2410.06551},
|
||
year={2024}
|
||
}```
|