Int

The most boring node in ComfyUI-FairLab is also one of the most useful if you've ever changed a seed in four places and gotten four different images. Int is a constant integer source: type one number, get that number back on a wire you can fan out to as many sockets as you want. Change it once, every consumer updates.

ComfyUI already ships this - the core PrimitiveInt does exactly this job - so FairLab's version is a convenience duplicate with the same shape: one value widget (default 0, range −65535 to 65535), one value INT output. You reach for it when the alternative is right-click → Convert Widget to Input on five different nodes and wiring a number everywhere by hand. If you're building anything with a shared seed, a shared step count, or a shared width, a single Int node is the one authoritative source and it kills the silent-mismatch class of bug before it happens.

Inputs and outputs

• value - the integer. That's it. One widget in, one INT out.

The node is marked as an output node, which just means it always fires when the graph runs - harmless for a constant.

How it works

There's no magic here: node_function returns whatever you typed. The value lives on the wire, so anything downstream - a KSampler seed, a batch count, an image dimension - reads the same number every run, and if you want it random or stepped per run, that's the widget's control_after_generate doing the work, not this node.

Where it fits

Put Int nodes next to anything with numeric widgets you keep re-entering. The pattern that pays off immediately: seed + width + height + steps as four Int nodes at the top of a workflow, everything below wired to them, so a workflow you share still has its "settings" in one place instead of scattered across the canvas.

Install

FairLab installs as one pack:

cd ComfyUI/custom_nodes
git clone https://github.com/yanhuifair/ComfyUI-FairLab.git
cd ComfyUI-FairLab
pip install -r requirements.txt

or use ComfyUI Manager → Install Custom Nodes → search ComfyUI-FairLab. Then restart. The pack's dependencies (opencv, googletrans, perfect-pixel, requests) are only needed by specific nodes; this one runs with zero extras. Search the canvas for Int or FairLab to find it - it lives under the Fair/string category, which is a bit of a misnomer for a numeric node.

Gotchas

• Don't confuse FairLab Int with the polymorphic core Primitive node, which adopts whatever type you plug it into. This one is unambiguously an integer, which is honestly a plus - one less thing to surprise you.
• The widget clamps to ±65535. If you need bigger numbers (say, a 131k seed), FairLab's Number node or the core PrimitiveInt may serve you better.