♻ Loop (Integer)

The integer version of Bjornulf's range-loop nodes - good for anything you'd normally scrub through as a whole number: sampler steps, batch counts, a seed offset, anything on a KSampler or elsewhere that takes an INT widget. Set a start, an end, a step size, and let ComfyUI's execution engine run through every value in between.

How it works

from_this, to_that, and jump describe the range, and the node outputs it as a list of integers. Because ComfyUI treats a list output as "run everything downstream once per item," connecting this into, say, a KSampler's steps input means your workflow will render once at each step count in the range - no manual re-queuing between values.

Inputs and outputs

from_this (INT, default 0, 0–1000, step 1), to_that (INT, default 10, same range), jump (INT, default 1, same range). Output: an INT list. To actually wire this into a KSampler parameter, right-click the widget (steps, for example) and choose "Convert widget to input" first - plain number widgets in ComfyUI don't expose a socket until you do that.

Installing it

Ships with the rest of Bjornulf_custom_nodes. ComfyUI Manager → search the pack title → install → restart, or cd ComfyUI/custom_nodes && git clone https://github.com/justUmen/Bjornulf_custom_nodes and restart manually. No dependency specific to this node.

Common issues & troubleshooting

If you're using this to sweep steps on a KSampler for a quick "what step count is enough" comparison, know going in that the author's own README flags this specific use case as not optimized - the node itself is fine, but looping steps re-runs the entire sampling process from scratch at each value rather than reusing any work between them. It's genuinely useful for eyeballing results across a range, but don't expect it to be fast; treat it as "good enough for quick testing," which is the exact phrase the README uses, not a production-speed comparison tool.

The usual list-loop budgeting applies: a wide range with a tight jump queues a lot of full runs. Sweeping steps from 10 to 50 by 1 is 41 complete generations. Start with a coarse step to find the interesting range, then narrow it if you need finer resolution.

And the same seed-consistency note as the Float version applies here: a loop by itself keeps the same seed across every iteration, which is what you want for a fair steps-vs-quality comparison, but it means back-to-back outputs can look suspiciously similar if you were expecting variety rather than a controlled sweep. If you actually wanted different results each iteration, combine this with Bjornulf's Text with Random Seed node rather than assuming the loop alone randomizes anything.