LC Sigma Resample

If you googled this class name, you've found a ghost - a perfectly healthy one. LCChangeStepCount and LCSigmaResample are two registrations of the same node in lonecatone23's ComfyUI_LC123_nodes pack. Both appear in the node menu under the display name LC Sigma Resample, and the pack's own README calls LCChangeStepCount an alias. Open the source and you'll see it: a single LCSigmaResample Python class mapped under both names in NODE_CLASS_MAPPINGS, with identical inputs, outputs, and behavior. Nothing about this node differs from its better-known twin.

So why does the old name exist at all? Backward compatibility. Renaming a node class breaks every saved workflow that references the old one - ComfyUI looks up the class name in the JSON and shows a dead, un-runnable node when it's gone. By keeping LCChangeStepCount registered, the pack lets graphs saved against the earlier name keep loading and running as if nothing happened. That's exactly the situation where you'll bump into this page: you opened an old workflow, or a shared graph, and this is the class name baked into its JSON. It loads, it runs, and you're free to swap in the newer LCSigmaResample name next time you rebuild.

What it does

It re-grids an existing sigma schedule onto a new step count without moving the endpoints. The formula, straight from the node description:

new_steps = round(old_steps * multiplier) + adder

Then it linearly interpolates the incoming curve so the sampler downstream really runs that many steps - densifying one band of a split schedule, or thinning another, while the overall path shape stays put.

The only widgets you'll ever touch:

• multiplier (default 1.0) - multiply the current step count first; 2.0 doubles the steps, 0.5 halves them.
• adder (default 0) - then add or subtract a fixed number, for when you want +4 rather than exactly double.

sigmas in (SIGMAS), sigmas out (SIGMAS). Put it after a split node, on the high or low slice you want denser - not in front of the splitter - and wire the result into an Advanced sampler's sigmas input. The full mechanism, positioning, and pitfalls are the same as its twin, and I'd point you to the LC Sigma Resample page for the long version rather than repeat it here.

Installing

Identical to the rest of the pack - ComfyUI Manager (search "ComfyUI_LC123_nodes") or:

cd ComfyUI/custom_nodes
git clone https://github.com/lonecatone23/ComfyUI_LC123_nodes

Restart ComfyUI and you're done. No extra pip packages, no model downloads for this node; the pack needs only ComfyUI's own environment. If the pack fails to load, check that __init__.py sits directly in custom_nodes/ComfyUI_LC123_nodes/ and isn't nested an extra level down.

One honest note: this is a thin utility node with two registration names, which is why this page is short. If you found it confusing that both names point to one tool, you're not wrong to be confused - but it's deliberate, and it's the reason an old workflow of yours still works.