Int Relay

"Int Relay" is a pass-through node for integers - in one end, out the other, nothing changes. It's part of a family of typed relays in this pack, and like its siblings it's a graph-organizing tool rather than a processing node.

But the integer relay has a second job worth knowing about. Because the input is a widget with a default value, the node doubles as a constant source: type 512 into the widget and you've got a fixed INT you can wire to anything that wants an integer. That's genuinely useful, because ComfyUI's Primitive node - the usual "feed a number into a wire" tool - is finicky about converting between widget types. A typed relay sidesteps that entirely: the input type is locked to INT, so the value on the wire is always an INT, no conversion surprises.

How it works

Trivially: return (int_,). The widget holds an integer (default 0, step 1), and whatever value is there comes out the output unchanged.

Inputs and output

• int_ (INT, default 0, step 1) - the value, either fed in from elsewhere or typed straight into the widget.
• INT - the same value out.

When you'd reach for it

Three cases. One: you're cleaning up a crowded graph and want a named, visible anchor for an integer wire instead of a bare Reroute. Two: you need a fixed integer somewhere and don't want to fight Primitive's type conversion. Three: you're mid-reorganize and want to keep a value pinned while you replace the node that used to supply it. None of these are daily-driver scenarios, which is fine - relays are tools you reach for when the graph becomes the problem.

The honest take

If you already live happily with Reroute and Primitive, you can skip this one. It's a convenience, not a capability. The whole pack installs the same way regardless, so it's there if you want it.

Installing it

It's part of longgui0318/comfyui-common-util:

cd ComfyUI/custom_nodes
git clone https://github.com/longgui0318/comfyui-common-util
cd comfyui-common-util && pip install -r requirements.txt

Restart after, or install via ComfyUI Manager by searching "comfyui-common-util". Note that although requirements.txt pins only opencv-python, the pack imports scipy, scikit-image, and PyWavelets at load - if it fails to appear, pip install scipy scikit-image PyWavelets then restart.