Image Mix RGB Channels

This one takes three separate channel images - a red, a green, a blue - and stacks them back into one full-color image. It's the reassembly half of channel surgery. The WAS README puts it plainly: "Mix together RGB channels into a single image." Pair it with Image Select Channel (which pulls a single channel out) and you've got a full split-edit-recombine loop for doing things to color channels individually that you can't do to the image as a whole.

Why would you want that? Channel-level control is a real trick in the toolbox. Deliberately misaligning the channels gives you chromatic-aberration and glitch looks. Blurring just one channel, or boosting contrast on green only, gets you color grades that are awkward to reach any other way. And if you're building something procedural - a fake-3D anaglyph, a displacement-driven effect - operating per channel is often the cleanest path.

How it works

An RGB image is really three grayscale layers riding together. Image Mix RGB Channels reverses the split: feed it one grayscale image per channel and it writes each into its slot, red into red, green into green, blue into blue, producing a single combined image. Whatever you did to those channel images in between - shift them, filter them, swap two of them - shows up in the recombined result.

The inputs and outputs that matter

Conceptually it's exactly what the name says:

• Three channel image inputs - one each for red, green and blue. These are typically grayscale images representing a single channel's intensity.
• One combined image output - the reassembled RGB image, ready to wire into a save node, a further filter, or a VAE-encode if you're theming an init.

I'm describing this from the node's stated job rather than a formal schema, so I won't invent exact widget labels - but the red/green/blue-in, image-out shape is the point.

How to install it

Part of the WAS Node Suite:

• ComfyUI Manager: search was-node-suite-comfyui, install, restart.
• Manually: cd ComfyUI/custom_nodes && git clone https://github.com/WASasquatch/was-node-suite-comfyui/, then pip install -r requirements.txt against your ComfyUI Python, and restart.

This is a plain pixel operation - no model download, no heavy dependency. The pack's weighty pieces (BLIP, SAM, MiDaS) only load for the nodes that call them, so a channel-mixing workflow stays light.

Common issues

The mistake people make is a mismatch: the three channel images need to be the same dimensions. Feed it a red channel at one size and a blue at another and you'll get an error or a garbage result, because there's no sane way to overlay two grids that don't line up. If you split from one source image and don't resize anything in between, you're fine by construction.

The other thing to keep straight is which channel is which. It's easy to accidentally route your edited "red" into the blue input and wonder why everything's tinted. If a recombined image comes out with a wild color cast you didn't ask for, check your wiring before you check your settings - a swapped channel looks exactly like a bug.

Everything else is the standard WAS caveat. The pack has been in retirement since December 2023 - still heavily used, just not maintained - so if a ComfyUI update makes every WAS node disappear with an "Import Failed", that's dependency drift across the whole suite, not this node. Re-run requirements.txt against the correct Python environment (or the bundled install.bat) and restart to bring them back.