😼> Divide Channels

Splits an image into its three color channels - red, green, blue - and hands each of them back as its own grayscale-looking image, alongside the original. The README's own framing is the right one: it's for when you want to manipulate just one channel without touching the other two, which a lot of ComfyUI's built-in image nodes don't give you an easy way to do.

Practical uses: pushing contrast or brightness on only the red channel for a warmth shift, isolating a channel to feed into something that expects a single-channel mask-like input, or classic RGB-shift/chromatic-aberration style effects where you offset one channel relative to the others before recombining. YANC ships a matching Combine Channels node (not covered here, but worth knowing it exists) to put the three channels back into one image once you're done manipulating them individually.

How it works

It reads the image tensor and slices out each of the three color channels independently, outputting each as its own IMAGE (still shaped like a standard image, just with that one channel's data - so it'll display in grayscale-looking form since there's no color information from the other two channels present).

The inputs and outputs that matter

• image - the source to split.

Four outputs: image (the unmodified original, passed through for convenience so you don't need a separate wire back to your source), plus red, green, and blue - each an isolated single-channel IMAGE you can process independently before recombining with Combine Channels.

How to install it

Via ComfyUI Manager: search "yanc" or "YANC - Yet Another Node Collection," install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/ALatentPlace/ComfyUI_yanc

then restart. No extra dependencies or downloads - straightforward tensor slicing.

Common issues & troubleshooting

The individual channel outputs look grayscale, not colored. That's expected - a single color channel has no color information on its own (a "pure red channel" isn't a red image, it's a grayscale map of how much red is at each pixel). If you want to see it tinted, you'd need to recombine it with zeroed-out other channels, which this node doesn't do automatically.

Forgot to recombine and the final output looks broken. Divide Channels only splits - it doesn't rejoin anything. If your workflow needs a normal color image at the end, route your processed red/green/blue outputs into YANC's separate Combine Channels node before anything downstream expects a standard RGB image again.

Channel order confusion. The three outputs are explicitly labeled red, green, blue - if a downstream effect looks tinted the wrong color, double check you didn't accidentally cross two of the three wires when connecting them back up after processing.