RGB Channel Merger

RGB Channel Merger takes three IMAGE inputs - red_channel, green_channel, blue_channel - and combines them into a single RGB image. It's the "put it back together" node for anyone working with separated channels: masks rendered per channel, channel-split outputs from a processing step, or per-color effects you want to recombine into a viewable image.

This is a genuinely small utility, but it fills a real gap. ComfyUI has plenty of nodes that split an image into channels; far fewer that cleanly merge arbitrary channel images back into one, and this one is picky in a useful way.

How it works

Rather than blending the three inputs, it extracts one channel from each and concatenates them:

• Red input → its channel 0 becomes the R channel of the output.
• Green input → its channel 1 (or channel 0 if the input is single-channel) becomes G.
• Blue input → its channel 2 (or channel 0 if it doesn't have three) becomes B.

That detail matters more than it sounds: you don't need pre-split single-channel images. Feed a normal RGB image into red_channel and the node pulls its red channel out and uses it. So you can, for example, drive the output's R from one image's red, G from another's green, and B from a third's blue - a fast way to cross-pollinate color channels between renders without any channel-splitting nodes at all.

Output is a single merged_rgb IMAGE.

The constraints

The inputs must agree on batch size, height, and width, or the node throws a clear error - no silent size mismatch. It's also a merge, not an adjust: there's no opacity, blend mode, or mask here. If you want channels composited with softness, that's a different node; this one just stacks them.

Honest take

This is a three-channel concatenation with fallback logic - about 30 lines. It's in the pack because channel manipulation is a recurring ComfyUI need, and it does that one job without fuss. If you work with per-channel masks or channel-split pipelines, it's worth having around; if you never touch channels, you'll never miss it.

Installation

Part of Praveen's ComfyUI Tools. Install via ComfyUI Manager (search "Praveen" / "praveen-tools"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/Praveenhalder/praveen-tools

Restart ComfyUI. No model downloads, no extra pip dependencies - it's pure torch tensor slicing, on top of stock ComfyUI.