Nifty Math

Every ComfyUI graph eventually grows a little tower of math nodes: add two floats, multiply by a scale, clamp the result, feed it into a latent size. It works, but it's five nodes to express one formula. Nifty Math collapses that into a single node with a text field - type a * b + clamp(c, 0, 1) and you're done. If you've ever wished ComfyUI math were more like a calculator and less like plumbing, this is the node.

How it works

The node gives you up to 16 named inputs, a through p. Connect values to them, write an expression in the expression field using those letters as variables, and it evaluates. Unconnected slots default to 0, so a + b works the moment you connect two things.

Under the hood it's a Python eval - but a carefully sandboxed one. The author ships a whitelist of safe globals with __builtins__ disabled, so you get math functions, trig, and the listed helpers, not arbitrary code execution. That's the right instinct for a node that evaluates strings; you're still running whatever the workflow author typed, so only load expressions you trust.

What you can express:

• Arithmetic: + - * / // % ** (including integer division and powers).
• Comparisons: a == b, a != b, a < b, a > b, a <= b, a >= b.
• Boolean logic: a and b, a or b, not a, plus ternary: a if a > 0 else b.
• Functions: abs, round, min, max, pow, sqrt, floor, ceil, log, log2, log10, exp, and full trig (sin, cos, tan, asin, acos, atan, atan2, radians, degrees, hypot).
• Helpers: clamp(v, lo, hi), lerp(a, b, t), remap(v, a, b, c, d) - the remap alone kills a lot of chained math nodes.
• Constants: pi, tau, e, inf.

The three outputs - and the trap

The result is cast to all three output types at once, so you can grab whichever you need:

• float - the actual result.
• int - the result cast to integer. Note: it truncates, it doesn't round. 2.9 becomes 2, not 3. If you need proper rounding, wrap it: round(a / b).
• boolean - False when the result is 0/0.0, True otherwise. Handy for feeding a switch.

Two edge-case behaviors from the source are worth remembering: division by zero returns 0 instead of erroring (a deliberate choice), and any other expression error raises a descriptive exception so you can see what went wrong.

Installing it

Part of the Nifty Nodes pack:

cd ComfyUI/custom_nodes
git clone https://github.com/Stibo/comfyui-nifty-nodes

or search "Nifty Nodes" in ComfyUI Manager, then restart. No models, no heavy dependencies - just the pack's own requirements.

Gotchas

The int-output truncation is the one that actually bites people - wire a latent dimension through and you'll get 511 instead of 512 if you forgot to round. And because slots default to 0, a typo'd variable name silently evaluates as zero rather than erroring; if a formula looks wrong, check for a misspelled letter first. It's the most direct math node in the pack, and for the price of remembering one expression syntax it deletes a lot of clutter from your graph.