ComfyUI-Color_Transfer
Implementation of color transfer using KMeans algorithm
Nodes (9)
The 30-second way to feed colors into every palette node
Steal a photo's whole color mood with one reference image
Pull the five dominant colors out of any photo
Optimal transport, no posterization
Smooth, continuous palette recoloring via radial basis functions
The color-palette node with every knob, and the one to learn first
The two-input palette node that's a whole look transfer
The 'just recolor it' node — cluster mapping with the boring knobs removed
Palette recoloring with soft edges and a real blend control
ComfyUI-Color_Transfer
Postprocessing nodes that implement color palette transfer in images. It replaces the dominant colors in an image with a target color palette.
Prevention/troubleshooting of import error
To avoid import issues install python package using CLI pip install scikit-learn.
Installation
Clone repo to your custom_nodes folder
git clone https://github.com/45uee/ComfyUI-Color_Transfer.git
Usage
- Create a "Color Palette" node containing RGB values of your desired colors. Color must be defined in this format: [(Value, Value, Value), ...], for example [(30, 32, 30), (60, 61, 55), (105, 117, 101), (236, 223, 204)]
- Create a "Palette Transfer" node, and connect your image and palette as input, that's all.
- You can specify color clustering and comparing distance method, by default MiniBatchKMeans faster but can be less accurate
Example
