🔢 Number Input

Let's be honest about what this node is: a single number field with two outputs. The Number Input takes one number you type, and hands you back both the float and the integer version of it. That's the entire job. No state, no math, no secret features - and that's exactly why it earns a place in a workflow.

ComfyUI has a chronic int-versus-float mismatch. Loaders and samplers frequently want an INT where you've been fiddling with a float, and a lot of "why is this node rejecting my number" confusion is really just a type mismatch in disguise. This node gives you both versions of the same value out of the box: float_val and int_val. Wire whichever the downstream node demands, no rounding node needed in between.

How it works

The backend is three lines of Python: take the number input, return it as a float, and return int(number) as the integer. The widget itself is a plain number field with a sane range - -999999 to 999999, stepping in 0.01. Nothing gets clamped, nothing gets cached, and on every run you get exactly what you typed.

One subtlety worth knowing: the int output is a truncation, not a rounding. Type 3.9 and you get 3 on int_val, because Python's int() cuts toward zero. If you wanted 4, you'd need a rounding node - this one doesn't fudge your numbers for you.

Inputs and outputs

• Input: number (FLOAT) - the value you type, default 1, range -999999 to 999999, step 0.01.
• Outputs:
  • float_val (FLOAT) - the number as typed.
  • int_val (INT) - the truncated integer version.

Both come out on every run, so you never have to choose.

When you'd bother

You wouldn't for a single value - ComfyUI's own widgets cover that. Where this earns its keep is when you want both representations for different branches of the graph, or when you want one knob to drive a batch of parameters and need each consumer to receive its preferred type. Think of it as a tiny adapter, the sort of node you add once and never think about again.

Troubleshooting

There's almost nothing to break here. If int_val looks one less than you expected, remember it truncates rather than rounds. And if the node is missing entirely after install, it's not a code problem - it means the pack didn't load, which is the same shared setup issue below.

Installing it

Like everything else in this pack, it comes with Orion4D_MetaNode:

cd ComfyUI/custom_nodes
git clone https://github.com/orion4d/Orion4D_MetaNode

Or use ComfyUI Manager and search "Orion4D_MetaNode". Restart ComfyUI and it appears under Orion4D_MetaNode → UI Widgets. No model files, no extra pip packages - the pack's requirements are just the basics ComfyUI already ships. It's a young, single-developer pack, so expect a useful-but-unexciting utility rather than a community staple.