ImagePass

There's a small category of nodes in ComfyUI that do literally nothing to your data and exist purely to make the graph itself easier to work with. ImagePass is one of them: an image goes in, the exact same image comes out, unmodified. The value isn't in transformation - it's in what a labeled, typed pass-through point gives you in a graph that's getting complicated.

Why you'd want this

A generic Reroute node does something similar, but it's untyped and generic-looking. ImagePass is specifically an image-in, image-out node, which means it reads clearly in a crowded graph - anyone looking at your workflow (including future-you) can immediately tell "this is the image branch, nothing happens here yet" rather than having to trace a reroute back to figure out what type it's actually carrying. That makes it genuinely useful for organizing big workflows: as a clearly-labeled anchor point where you might insert a real processing node later, as a stable connection point when you're rearranging a graph and don't want to rewire everything at once, or simply as a visual marker that breaks up a long, hard-to-follow chain of connections.

Don't expect it to do anything functional beyond that. If you're looking for a node that resizes, adjusts, or transforms an image in some small way, this isn't it - by design.

How it works

The image input is optional, and whatever comes in goes straight to the IMAGE output unchanged. There's no processing step in between.

The inputs and outputs that matter

• image (optional) - the image to pass through.

One output: IMAGE - identical to whatever was fed in.

That's the entire node. No other settings.

How to install it

• ComfyUI Manager - search KJNodes for ComfyUI, install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/kijai/ComfyUI-KJNodes, then pip install -r ComfyUI-KJNodes/requirements.txt (portable: python_embeded\python.exe -m pip install -r ComfyUI\custom_nodes\ComfyUI-KJNodes\requirements.txt), then restart.

Nothing to download - it has no dependencies of its own beyond the pack.

Common issues & troubleshooting

Expecting it to do something and being confused when it doesn't. This is the whole node - if your image looks unchanged after passing through it, that's correct behavior, not a bug. If you meant to resize, color-correct, or otherwise modify the image, you want a different node entirely.

Running it with nothing connected to image. Since the input is optional, the graph won't refuse to run just because it's unconnected - but what comes out of the IMAGE output in that case depends on how the rest of your graph handles an empty/None image. If a downstream node errors on a missing image, check back here first rather than assuming the problem is further along the chain.

Wondering why you'd add an extra node that does nothing. It's purely organizational - if your workflow is small and simple, you genuinely don't need this. It earns its place in larger workflows where a clearly-labeled, correctly-typed anchor point makes the graph easier to read and safer to rewire later.