Black & White Converter

Some nodes try to do everything. This one does exactly one thing and doesn't even give you a slider to mess it up. Feed it an image, get back a grayscale version. That's the entire contract: one input, one output, zero options. It's called "Black & White Converter" on the canvas (class "Black and white"), part of Cyber-BlackCat's ComfyUI-MoneyMaker pack - "ComfyUI-Yuan" to its friends, where yuan is a currency pun and the README cheerfully promises the nodes will make you real money.

How it works

The conversion uses the standard Rec. 601 luma weights: 0.299 × red + 0.587 × green + 0.114 × blue, computed in torch directly on your image tensor. It's the same weighting your TV has been doing for decades, which is to say it's not a flat average of the channels - green contributes the most because human eyes are most sensitive there. A two-channel image gets a straight average of its two channels, and a one-channel image just passes through. Whatever goes in, the output is expanded back to three channels so it stays a drop-in IMAGE for anything downstream.

That three-channel detail matters more than it looks like it does. The README calls this "a Gray level map generator," which is the honest framing: this isn't a final B&W aesthetic, it's a luminance signal. Because the output stays an IMAGE, you can feed it into any image-expecting node without rewiring your graph - as a brightness map, a preprocessing step, or an input to something that wants a monochrome version of the same picture.

Where people get burned

The number one confusion is the output type. It's an IMAGE, not a MASK. If the node you're feeding wants an actual MASK input, you'll need to convert IMAGE → MASK yourself (ComfyUI has built-in nodes for that). A surprising number of people expect this to hand out masks and then wonder why the wire won't plug in.

The other trap is expectation-setting. This is a flat, mathematically correct luminance conversion - no tone curve, no grain, no contrast stretch, no film look. If you want moody B&W with actual vibe, run this as a starting point and then push it through your usual post-processing, because this node is deliberately not doing any of that. It's a utility, not a look.

Honest take

Is this the best grayscale node in the ecosystem? There are a dozen of them, and ComfyUI's built-ins can get you to the same place. What this one has going for it is pure zero-friction: no parameters to tune, no confusing extras, works on any image type in the pack's ecosystem. It's the node you grab when you want the operation done in one click and you don't care about branding. Given the pack's other, more interesting nodes (the Photoshop-style transfer is the keeper), this one is the dependable workhorse you install anyway because it came in the box.

Install

Same as the rest of the MoneyMaker pack. Fastest path is ComfyUI Manager - search "MoneyMaker" (pack title "ComfyUI-Yuan"). Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/Cyber-BlackCat/ComfyUI-MoneyMaker

Then restart ComfyUI. Dependencies are torch, torchvision, numpy, pillow, opencv-python, and scikit-image - all but scikit-image are typically already present in a working ComfyUI install. No models, no downloads beyond the repo itself.