Integer Multiply

JWIntegerMul multiplies two integers and gives you back an integer. a * b, whole numbers in, whole number out. It's the integer twin of JWFloatMul, and the integer type is exactly why you'd pick it over the float version.

A lot of ComfyUI inputs demand a true integer, not a float that happens to look round - frame counts, batch sizes, pixel dimensions, step counts, tile counts. Feed one of those a 12.0 and some nodes complain. JWIntegerMul keeps the math in integer-land the whole way through, so the number that comes out is the right type for those slots, no conversion node needed.

The classic use is dimensions and frames. Multiply a base tile size by a count. Work out total frames as clips × frames-per-clip. Scale a batch count. Compute a resolution as a base times a multiplier and hand it straight to an empty-latent or resize node. Because it's wired rather than typed by hand, changing the base ripples through automatically.

What it actually does

a times b, both INT, one INT out. Both default to 0. The range here is genuinely enormous - roughly ±1.8e19, which is the span of a 64-bit integer - so you will never bump the limit with anything sane. Multiplication is exact; there's no floating-point fuzz because everything stays a whole number.

The inputs and outputs that matter

Two required inputs, one output:

• a - first factor (INT).
• b - second factor (INT).

Output: a single INT equal to a * b. Wire it into anything that wants an integer - an empty latent's width/height, a batch size, a frame count.

How to install it

Pack is jamesWalker55/comfyui-various. Via ComfyUI Manager, search Various ComfyUI Nodes by Type, install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/jamesWalker55/comfyui-various

then restart. No dependencies. If you only want the math nodes, the author's suggested approach is to drop the single comfyui_primitive_ops.py file into custom_nodes/ - it contains all the Integer and Float nodes and nothing else, keeping your node menu tidy.

Common issues & troubleshooting

Float vs. int matters more than you'd think. If a downstream node is throwing a type error about expecting an integer, make sure you're using this node and not JWFloatMul upstream. And if you multiplied two integers but need a float afterward (say for a denoise value), you'll want the pack's integer-to-float conversion node in between - JWIntegerMul will only ever hand you a whole number.

Zero default. Both inputs start at 0, so an untouched node outputs 0. If your result is stuck at zero, one input isn't set.