Int To Float (Soze)

Mirror image of Float To Int, and you'll reach for it just as often: an INT output refuses to plug into a FLOAT input, because ComfyUI's sockets don't auto-cast between numeric types even though the underlying values are perfectly compatible. Int To Float exists purely to bridge that gap.

The typical trigger is a seed, a batch index, or a loop counter (all naturally INT) that you need to feed into something expecting a float - a strength slider, a scheduler parameter, a math node built around floats. Instead of rewriting the upstream logic, you drop this in the wire.

How it works

It takes whatever integer arrives on its input and outputs the same numeric value as a FLOAT. Because every integer is exactly representable as a float at the ranges ComfyUI deals with, this is lossless in both directions - unlike Float To Int, there's no rounding ambiguity to worry about here. What you put in comes out with a .0 on it, nothing more.

The inputs and outputs that matter

• int - the value to convert. Default 0, range −2,147,483,648 to 2,147,483,647 (standard 32-bit int bounds), step 1.
• Output: FLOAT - feed it into whatever FLOAT-typed input rejected your integer.

That's the whole node.

How to install it

Via ComfyUI Manager, search "Quality of Life Nodes for ComfyUI" (or "Soze") and install from there. Or manually:

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

then restart ComfyUI. No model downloads, no heavy dependencies - this node (like the rest of the converter group) doesn't touch any of the pack's optional integrations, so the CD_API_KEY/FAL_KEY/ELEVENLABS_API_KEY env vars the README mentions for ComfyDeploy, FAL, and ElevenLabs nodes are irrelevant here.

Common issues & troubleshooting

This is one of the least-discussed corners of an already obscure pack. ComfyUI_Soze barely registers in general ComfyUI discussion online, so there's no troubleshooting thread waiting for you if something looks off - the node is short enough that reading the source on GitHub is faster than searching for an explanation.

You probably don't need it if the downstream node is flexible. Some custom nodes accept INT,FLOAT as a combined type and cast internally; if a plain int-to-float link does work without this node in a specific case, that target node is just more permissive than the ComfyUI default, not evidence this converter is broken.

Missing after install. If the node doesn't show up post-install, it's almost always an incomplete clone or a pip install -r requirements.txt that didn't run - re-clone rather than dig, since there's nothing else in this node's dependency chain to go wrong.