abyz22_blend_onecolor
Tint a region with R, G, B and a slider
- imageA
- mask
- image
Sometimes you don't want a second image to blend with - you want to pour a single color over part of the frame and fade it. That's abyz22_blend_onecolor, the sibling of the pack's abyz22_blendimages. It blends your image with a solid R/G/B fill through a mask, which sounds trivial until you need to tint a cloth region, warm up a shadow area, or push a piece of skin toward a specific tone and want it to stay a number you can tune instead of a guess.
It lives in abyz22/image_control, the Impact-Pack-derived fork that does body editing and outpainting. This node is the utility-grade component: no model, no weights, just a color math pass.
Inputs that matter
imageA- the image to tint (IMAGE).mask- another "mask" that is actually an IMAGE, same convention asabyz22_blendimages. White = color shows, black = original shows, grey = partial. Feed it a feathered mask or the blend edge will look cut with scissors.blend_percentage(0–1, default 0.5) - how strongly the color wins where the mask is white.R,G,B(each 0–255) - the fill color. Default is 255/255/255, pure white.
Output is a single image. There's no color-picker UI here - the three integer sliders are the entire palette - so for anything more exotic than a flat tint you'll want a color wheel node upstream and this node's simpler sibling instead.
How it works
The code builds a tensor the shape of your image filled with your R/G/B values, then calls WAS Node Suite's "Image Blend by Mask" once per batch frame, exactly like abyz22_blendimages. Which means the same dependency lands on you: WAS Node Suite must be installed, or this node throws at execution when it looks up the node mapping. The pack's own requirements.txt (openpyxl, pytorch-lightning, kornia) won't pull WAS in for you.
Install
Same as every node in this pack:
cd ComfyUI/custom_nodes
git clone https://github.com/abyz22/image_control
or install "image_control" via ComfyUI Manager, then restart. First startup is slow because of the pytorch-lightning dependency.
Where people get burned
The mask-is-an-IMAGE thing again - if you're hooking up a MASK-typed output from a segmentation node, convert it to an image first or the shapes won't line up. Also, blend_percentage above ~0.6 with a pure color tends to read as "painted", because a flat fill has no texture to fall back on; if you're tinting something with detail, keep it around 0.3–0.5 and let the original texture carry through. The classic use in this pack's world is the see-through-clothes trick - blend a skin-tone fill through a garment mask at low strength so the fabric reads as translucent rather than recolored.
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| imageA | IMAGE | — | |
| mask | IMAGE | — | |
| blend_percentage | FLOAT | 0.500–1 | — |
| R | INT | 2550–255 | — |
| G | INT | 2550–255 | — |
| B | INT | 2550–255 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| image | IMAGE | — |