ComfyUI-Inspyrenet-Rembg

ComfyUI node for background removal, implementing InSPyReNet

I've tested a lot of different AI rembg methods (BRIA - U2Net - IsNet - SAM - OPEN RMBG, ...) but in all of my tests InSPyReNet was always ON A WHOLE DIFFERENT LEVEL!

The cherry on top is that InSPyReNet has MIT License which allows for Commercial use (for example BRIA does not allow Commercial use to my knowledge)

Check the Licenses for yourself I will not be held accountable :)

Updates

Added an advanced node that supports threshold

Features

Superior rembg quality compared to other methods (just give it a try!)

Can take batch of images as input

Optimized for image batch to be the fastest rembg node (perfect for video frames)

Outputs both the image and the corresponding mask

Shows the progress in terminal

Installation

Simple way:

search for ComfyUI-Inspyrenet-Rembg in ComfyUI-Manager and hit install.

Done!

Manual:

1. Go to your custom_nodes folder in ComfyUI, open the terminal and run the following command:

git clone https://github.com/john-mnz/ComfyUI-Inspyrenet-Rembg.git

2. To install requirements, run the following commands:

cd ComfyUI-Inspyrenet-Rembg
pip install -r requirements.txt

3. Download models to ComfyUI/models/transparent-background

ckpt_base.pth:                      https://drive.google.com/file/d/13oBl5MTVcWER3YU4fSxW3ATlVfueFQPY/view?usp=sharing

ckpt_base.pth:                      https://github.com/plemeri/transparent-background/releases/download/1.2.12/ckpt_base.pth
ckpt_fast.pth:                      https://github.com/plemeri/transparent-background/releases/download/1.2.12/ckpt_fast.pth
ckpt_base_nightly.pth:              https://github.com/plemeri/transparent-background/releases/download/1.2.12/ckpt_base_nightly.pth

TODO: model loader for trained models

InSPyReNet_SwinB_HU.pth:            https://drive.google.com/file/d/1HB02tiInEgo-pNzwqyvyV6eSN1Y2xPRJ/view?usp=sharing
InSPyReNet_SwinB_DIS5K.pth:         https://drive.google.com/file/d/1aCxHMbhvj8ah77jXVgqvqImQA_Y0G-Yg/view?usp=sharing
InSPyReNet_SwinB_Plus_Ultra.pth:    https://drive.google.com/file/d/13oBl5MTVcWER3YU4fSxW3ATlVfueFQPY/view?usp=sharing

note

If torchscript_jif is set to on, it will trace model with pytorch built-in torchscript JIT compiler. May cause delay in initialization, but reduces inference time and gpu memory usage.

basic workflow

download this file and drag and drop it into your comfyui as basic workflow:

inspyrenet-rembg-basic-workflow.json

show case