ComfyUI-FBCNN

ComfyUI implementation of FBCNN

What is it?

This is a JPEG artifact/compression removal tool, which can work automatically or with the help of a user.

Installation

Download or git clone this repository inside ComfyUI/custom_nodes/ directory. The only required dependencies are Pytorch, numpy, and requests, which should be already installed if you have ComfyUI.

How to use it?

In ComfyUI, you can find this node under image/upscaling category. The model requires an estimate of the compression level, which is a number between 0 and 100 (the same you need to provide when compressing a JPEG image). You can either set auto_detect to enabled and use built-in estimator, or set it to disabled and provide the value yourself in compression_level.

⚠️ If you leave auto_detect as enabled, the value of compression_level will be ignored.

Credits

• FBCNN
• ComfyUI