Grayscale Image (WLSH)

This one does exactly what the name says and nothing more: feed it a color image, get back the same image in grayscale. There's no clever trick here, no model, no config beyond "here's an image." It's the kind of node that's boring by design - a single-purpose utility that earns its keep by being one click instead of you going and finding an image-processing node from some other pack.

The README's own reason for including it is ControlNet prep. Some ControlNet preprocessors and conditioning setups work better - or are specifically expecting - a desaturated input, especially for things like line art extraction or depth-adjacent workflows where color information is noise you don't want influencing the conditioning. Stripping color out before it hits a preprocessor is a five-second fix for artifacts that would otherwise be confusing to debug.

Inputs and outputs, and there really is only one of each: original takes an IMAGE, and the node returns a single output named grayscale, also typed IMAGE. That's the entire interface. Wire your source image in, wire the grayscale result wherever you need it - a ControlNet preprocessor, a second branch for a black-and-white variant, whatever.

One thing worth knowing since it's easy to assume otherwise: the output is still an IMAGE type, not a MASK. It's a genuinely desaturated RGB image (three channels, all carrying the same luminance value), not a single-channel mask. If what you actually need is a mask - say you're chaining this into something that wants a grayscale-derived selection - you'll need to run the result through ComfyUI's own "Convert Image to Mask" node afterward. The WLSH pack's own Generate Border Mask node, in fact, calls out exactly that pattern in its README description: pass its output through "Convert Image to Mask" using the green channel. Same idea applies here if you're chasing a mask rather than a picture.

Installing it: grab the whole pack, not just this node - WLSH ships as one bundle. Easiest path is ComfyUI Manager: search "wlsh_nodes" or "WLSH Nodes" and install from there. Manual install is just as quick:

cd ComfyUI/custom_nodes
git clone https://github.com/wallish77/wlsh_nodes

Restart ComfyUI afterward and it'll show up under WLSH Nodes/image in the node menu. No models to download, no extra Python packages called out in the README - it's a plain image-processing node, so once the pack is cloned in you're done.

Troubleshooting: there's genuinely not much that goes wrong with this one. If the node doesn't appear after install, that's a pack-loading problem (check the ComfyUI console log for import errors on wlsh_nodes and confirm the folder actually landed in custom_nodes/), not something specific to this node. If your "grayscale" output still looks tinted, double check you're not accidentally viewing it through a color-managed preview or a downstream node that's re-tinting it - the conversion itself is a standard luminance desaturation and doesn't leave a color cast.