RiceRound Float

If you've landed on this page, you're probably trying to publish a ComfyUI workflow as a hosted "cloud node" through RiceRound rather than looking for a generation node - this pack doesn't touch samplers or models, it packages a working workflow so riceround.online can turn it into a public page and a callable endpoint. RiceRoundFloatNode is one of the building blocks for that page: it's how you expose a decimal value (a CFG scale, a denoise strength, a blend weight) as something the person using your published workflow can actually adjust, instead of a number frozen into the encrypted graph.

How it works

Same shape as its integer sibling, RiceRoundIntNode, just for floats. You set a name, a default value, and a range while building locally; when the workflow is published, that combination is presumably what renders as the editable field on RiceRound's generated page. Nothing about the node touches image data or models - it's pure parameter plumbing.

The inputs and output that matter

• name (STRING, default "数值" - "value/number" in Chinese) - the label. Worth overriding before you publish; the pack's defaults and its documentation both default to Chinese, a fair signal of who this tool was built for first.
• number (FLOAT, default 0) - the value you're setting while testing.
• min / max (FLOAT, defaults 0 and 100) - the bound on what an end user can set once this is live. Match it to whatever the field actually controls; a denoise strength wants 0–1, not 0–100, and there's no indication the node auto-clamps to anything sensible for you.
• Output: value (FLOAT) - feed it wherever a core FLOAT widget's output would go.

Installing it

ComfyUI Manager: search "RiceRound," click install, restart. By hand:

cd ComfyUI/custom_nodes
git clone https://github.com/RiceRound/ComfyUI_RiceRound

then restart. The README lists no requirements.txt, no model weights, nothing heavy - this pack is a packaging and upload layer, not a generation one, so install should be quick and there's nothing multi-gigabyte waiting at the end of it.

Common issues & troubleshooting

Is this actually safe to leave in when I publish? Yes - the README's one real structural warning is to build your publishable graph using nodes from RiceRound's Input category (which this is) and to strip out anything from the Output category before you run RiceRound Publish, since those are meant for local debugging only. You're using this correctly.

The range feels arbitrary and your published field lets people enter values that break your workflow. The 0–100 default is a generic placeholder, not a sensible bound for most float parameters in a diffusion workflow (CFG, denoise, LoRA strength all live in much tighter ranges). Set min/max deliberately for what you're actually controlling rather than leaving the defaults - this is the field standing between "reasonable user input" and whatever your workflow does with an out-of-range number, which the node itself doesn't appear to guard against.

You can't find English documentation past what's in this README. RiceRound's fuller docs live at a separate hosted site (help.riceround.online) and it's mostly Chinese; the stated fallback for anything not covered there is contacting the maintainer directly on WeChat. Budget for that friction if you're building something you plan to support for an English-speaking audience - you'll likely be figuring out edge cases from the node's schema and your own testing rather than from written docs.