NGs Image - Fill with Color

Same idea as its sibling, NGs Create Solid Color, but you don't type in a width and height - you hand it an existing image, and it repaints every pixel that color at that image's exact dimensions. Useful whenever you need a solid-color layer that matches something you already have: a same-size background to swap in behind a transparent subject, a flat comparison frame at a mask's exact resolution, or just resetting a canvas mid-graph without re-entering numbers you already had once.

How it works

It takes the image input purely for its dimensions - the actual pixel content is discarded entirely - and paints in the RGBA color you specify across the whole thing.

The inputs and outputs that matter

• image - the source you're borrowing width/height from. Whatever's actually in it doesn't matter.
• red, green, blue, alpha - 0–255 each, the fill color.

Output is "Filled Image" (IMAGE).

One thing worth knowing: unlike Create Solid Color, there's no mode (RGB/RGBA) toggle here - alpha is just a required field with no switch to drop it. So expect the output to always carry an alpha channel, even at alpha=255 (fully opaque). If something downstream insists on a strict RGB image, you may need to strip that channel before it.

If you don't have a source image yet and just want to specify a size directly, reach for NGs Create Solid Color instead - same output, manual dimensions.

Where this genuinely earns its keep over just typing numbers into Create Solid Color: any time the size you need isn't a round number you'd naturally type - an image that got resized by an upstream node to some odd resolution, or a mask whose dimensions you don't want to go check and copy by hand. Feeding the actual image in and letting the node read its dimensions removes an entire class of "off by a few pixels, doesn't line up" bugs that come from manually re-entering a width and height you already had sitting one node upstream.

How to install it

ComfyUI Manager: search "NeoGriever," install, restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/NeoGriever/ComfyUI-NeoGriever

then restart. No dependencies, no models - this whole pack installs and updates in seconds.

Common issues & troubleshooting

Output has an alpha channel you weren't expecting. As noted above, there's no RGB-only mode on this one - if a downstream node errors or behaves oddly on an unexpected channel count, that's the likely cause. Set alpha to 255 for a visually opaque result, but know the channel is still there.

Content of the source image "leaking through." It shouldn't - the node is meant to discard the input pixels entirely and paint flat color. If you're seeing any trace of the original image, double-check you're not accidentally looking at a cached preview of the input node rather than this one's output.

Beyond that, there's not much to trip over - it's a small, single-purpose node doing exactly what its name says.