ImageToMask (🚫Deprecated)

This node does one thing: pick a color channel - red, green, or blue - and it hands you that channel back as a MASK. It's a channel-extraction shortcut, useful back when it was the fast way to turn a red-painted region or a color-coded reference image into a usable mask without a general-purpose color-to-mask node.

It's sitting in the pack's Deprecated category now, so treat it the way you'd treat any deprecated node: fine if it's already in a workflow you downloaded and it still works, not the node to reach for when you're building something new.

How it works

The node reads the selected channel of the input image and returns its pixel values directly as a mask - brighter in that channel means more "on" in the mask, essentially treating one color plane of the image as grayscale mask data.

The inputs and outputs that matter

• image - the source image.
• channel - red, green, or blue. Whichever channel you pick becomes the mask.
• MASK (output) - the extracted channel, ready to wire into an inpaint node, a composite, or anywhere else that consumes a mask.

Why it's deprecated, and what to use instead

Core ComfyUI ships its own "Image to Mask"-style functionality, and mask-focused packs generally offer more channel options (including alpha, and luminance rather than a single RGB channel) plus finer control over the threshold. If you're building fresh, reach for whatever mask-extraction node your current mask/segmentation pack of choice provides, and lean on this one only if you're maintaining an older workflow that already has it wired in.

Installing it

You get this with the base pack, no separate install for a deprecated node specifically. ComfyUI Manager: search ComfyUI Easy Use, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Use

then install.bat (Windows) or pip install -r requirements.txt, restart. No models - it's pure channel math.

Common issues & troubleshooting

Mask looks inverted from what you expected. Channel-as-mask extraction has no concept of "this is the subject, this is the background" - it just reports the raw channel brightness. If your source image encodes the mask as dark-on-light rather than light-on-dark for the channel you picked, the mask will come out backwards from your intent. Add an invert step (most mask-utility packs have one) rather than assuming the node did something wrong.

Picked the wrong channel and got noise instead of a clean mask. If your source image wasn't specifically painted or generated with one clean channel carrying the mask data, extracting any single RGB channel from a normal photo just gives you that channel's natural brightness variation - not a clean binary-ish mask. This node works well on deliberately color-coded reference images and poorly on arbitrary photos; it's not a segmentation tool.

Looking for this node and it's not showing up. Double-check you're searching by its class name easy imageToMask or its display name - deprecated nodes sometimes get tucked into a collapsed submenu in the node browser rather than showing at the top level, since the pack is nudging you toward newer alternatives.