XY Inputs: Denoise //EasyUse

Denoise strength is one of those settings everyone eyeballs - 0.5 sounds reasonable, 0.7 sounds "more", but you don't actually know what either does to your image until you run it. easy XYInputs: Denoise generates a range of denoise values and hands them to the pack's easy XY Plot Advanced node so you get every value laid out as a grid in one pass, instead of nudging a slider and re-queuing five separate times.

Why you'd reach for it

Anywhere denoise strength is the knob you're tuning: img2img restyles, a hires-fix / upscale refiner pass, or any second-pass sampler where you're deciding how much of the input latent to keep versus regenerate. The range matters a lot and isn't intuitive - roughly 0.0-0.2 is a subtle touch-up, 0.3-0.5 keeps the composition but allows real style drift, 0.5-0.7 lets the model reinterpret elements more freely, and 0.7-1.0 is close to a full regeneration where only vague structure survives. A grid across that range shows you exactly where your image tips from "refined" to "reimagined," which is the whole judgment call this setting exists for.

How it works

Like the rest of the XY Inputs family, this node doesn't sample anything itself. Give it a start value, an end value, and a count, and it linearly interpolates the range and packages it tagged as the "Denoise" axis. Wire the output into the X or Y slot of easy XY Plot Advanced (alongside a pipe from upstream), and that node runs your sampling pass once per value, building the comparison grid.

The inputs and outputs that matter

• batch_count (default 3) - how many denoise values to test between your two endpoints. Each one is a full sampling pass, so a wider sweep costs more time.
• first_denoise / last_denoise (default 0.5 → 1.0, step 0.1) - the range. The default already leans toward the "significant change to full regeneration" end; if you're testing a subtle refiner pass instead, pull both down toward 0.1-0.4.

Output is a single X or Y, wired into whichever axis of easy XY Plot Advanced you want it on.

How to install it

ComfyUI Manager: search ComfyUI Easy Use, install, restart. Or by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Use

Windows: double-click install.bat. Linux/Mac: pip install -r requirements.txt. Restart ComfyUI. No models, no extra dependencies - just numbers feeding a plot.

Common issues & troubleshooting

easy XY Plot Advanced needs a pipe in addition to this node's output - connect a loader's pipe into it, or the plot has no workflow to actually run.

The value that trips people up most is where "0" and "1" actually sit: at denoise 0, the sampler does effectively nothing to the input latent, so the low end of a wide sweep can look like a wasted generation slot - that's expected, not a bug. At denoise 1, you're closer to a from-scratch generation than a refinement, so if your goal was a subtle touch-up and you're sweeping 0.5 to 1.0 (the node's default), you'll mostly see the model overwriting your source image rather than polishing it. Narrow the range toward the low end for refiner-style passes, and only sweep the wide 0.5-1.0 range when you actually want to see how far you can push a restyle before the original composition disappears.

As with any grid sweep, judge it on more than one seed if you can - a single run at each denoise value tells you what happened on that particular seed, and denoise strength interacts with seed noticeably enough that one lucky or unlucky roll can make a value look better or worse than it typically is.