♾️🌊✨ Integer to Float

The opposite direction from the pack's float-to-int bridge, and just as necessary. ESS Integer to Float takes an INT and hands back a FLOAT. ComfyUI is strict about these types - a math node expecting floats will reject an integer wire, and sometimes the number you have (a seed, a step count, a resolution) is a plain int that needs to become 28.0 before it'll play nice with a formula. That's this node's entire job, and it does it without drama.

It's part of the newer branded converter family in the Endless pack (♾️🌊✨ "ESS" nodes). The family covers the common bridges - int to float, float to int, number to int, string to float - so whichever direction your graph is stuck in, there's a matching node.

Inputs and outputs

• IntegerValue - the one input, an INT, default 0.
• FLOAT - the one output, an integer as a float. 28 becomes 28.0.

There's no rounding or loss - integers convert cleanly to floats. The only gotcha, in principle, is huge integers exceeding float precision, and you're not going to hit that with image-generation numbers.

When you'll actually use it

Classic case: a CFG or guidance math chain. You want to compute something like "steps × 0.7" and the steps value is an integer - feed the int through this node first and the math node stops complaining. Same for resolution math (width/height pairs as ints) and for feeding a seed into a chain that adds a bias before the sampler sees it. It's the kind of node you install once, use in three workflows, and never think about again.

Installing it

Ships in Endless 🌊✨ Nodes by tusharbhutt:

cd ComfyUI/custom_nodes
git clone https://github.com/tusharbhutt/Endless-Nodes
# restart ComfyUI

Or ComfyUI Manager → search "Endless". No dependencies or model files.

Gotchas

Very little to trip on here - the conversion is lossless for any value you'll realistically feed it. The one honest caveat is the pack-level one: single hobbyist maintainer, AI-assisted, rewritten in June 2025. A trivial cast like this isn't where that risk materializes, but if you don't otherwise use Endless, your existing math/utility pack likely has an int-to-float bridge too.