🅐 Integer

Every ComfyUI workflow ends up needing a plain whole number that isn't buried inside some other node. Steps, batch size, latent dimensions, how many LoRA repeats - all of those are integers, and all of them are more pleasant to drive when there's a dedicated node sitting in the graph that says exactly what it is. 🅐 Integer is that node. It does one thing: takes an integer you type and hands it back as a properly typed INT. If that sounds too simple to need a paragraph, remember that half of ComfyUI is fighting about types, and a node that just makes the type unambiguous is quietly useful.

How it works

Nothing clever. There's a single int widget with a default of 0, a step of 1, and a range wide enough for any legitimate use (the full 64-bit span, roughly −1.8×10¹⁹ to +1.8×10¹⁹). The node passes whatever you set straight through as its INT output. It doesn't clamp, doesn't round, doesn't randomize - which is exactly what you want when you need a value that stays put.

The most common wiring is the obvious one: int → steps on a KSampler, int → batch_size on an empty latent, or int → seed (though the pack also ships a dedicated 🅐 Seed node for that). Because it's a real typed INT, it slots into any port that declares INT, and the graph won't let you accidentally feed it somewhere it doesn't belong - that validation is the entire point of a typed primitive versus the loose core Primitive node.

Inputs and outputs

• int - INT, default 0, step 1.
• Output: INT.

That's the complete contract. If you find yourself needing the output to vary between runs, feed it into the pack's 🅐 Expression Eval and compute from it, or drive a switch with it - this node is the raw material those nodes consume.

Install

It ships in hay86/ComfyUI_AceNodes, so installation is the pack's standard dance:

cd ComfyUI/custom_nodes
git clone https://github.com/hay86/ComfyUI_AceNodes
cd ComfyUI_AceNodes
pip install -r requirements.txt

Restart ComfyUI and look for 🅐 Integer in the Ace Nodes category. Manager users: search ComfyUI_AceNodes.

What to watch for

Two small things. First, the pack installs a heavy requirements.txt (rembg, transformers, boto3, and friends) even though this node needs none of it - you're paying the pack's install tax either way. Second, don't use this when you actually want a seed with randomization semantics; there's a 🅐 Seed node in the same pack whose job is the seed role specifically. For a plain, repeatable, always-the-same number, this is the one.