Number List Plot

Why you'd reach for it

Numeric schedules are invisible by default. You build a curve, wire it into a LoRA strength or a batch of renders, and find out what it did to the images twenty minutes later. If the answer is "the second half is flat at zero because I got reverse wrong", you've paid a render to learn something a line on a canvas would have told you in a second.

Number List Plot draws the numbers. Up to four lists, each as a line in its own colour, with its length, lowest and highest value in the legend and the value under your pointer read out underneath. It's a debugger for schedules - the one node in your graph whose whole job is to make you not queue.

It takes whatever a number list turns out to be: a whole LIST such as Number Sequence's values, a per-run output such as its floats (the node collects the runs into one series), or a single number, which draws as one lonely point.

How it works

It's a preview node in the same sense as Preview Image: an output node with no outputs at all. Nothing downstream can depend on it, so it never sits in the middle of anything - you wire it in parallel to whatever is actually using the list and let it draw.

Like Preview Image and Save Image, it's an output node, which in ComfyUI means it always runs when the graph executes rather than being skipped by the cache. That's the property you want here: the plot reflects this run, not the run whose numbers happened to be cached.

The drawing is frontend work. The backend hands each series to the node's UI, which renders the lines, computes length and spread per series, and caps what it draws at 10,000 entries per series - past that the extra entries are counted but not plotted, so a 50,000-entry list gives you a plot and a note rather than a frozen browser tab.

Each series gets its own colour, which is the whole point of the four inputs: two LoRA strengths against each other, or a curve and its mirror, on one set of axes.

Inputs and outputs

values is the only required input, and it's the one you'll wire from a list. values_b, values_c, values_d are optional and identical, each drawn over the first in its own colour, so you can compare two schedules at a glance - the tooltip's own example is the strengths of two LoRAs.

There are no outputs. There is nothing to wire onward. If you're looking for a way to turn the plot into an image file, that's not this node; this one plots on the node and stops there.

Install

Pack install, as with everything else here:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui.git

Or Manager → search WAS Node Suite v3 → install → restart. ComfyUI 0.14.0+, Python 3.10+, no pip installs (the repo's requirements.txt says as much in one line). First start writes config.yaml and the support folders under ComfyUI/user/was-node-suite/ and compiles bytecode once.

Nothing to download, no model needed. The plot is drawn in the browser, so if the node renders as a blank box you're looking at a frontend problem, not a weights problem - hard-refresh the page, since the pack's JS is loaded from the extension.

Where people get burned

Nothing drawn. Either the node didn't execute - it's bypassed, muted, or downstream of something that was - or the wire isn't carrying numbers. A LIST and a per-run list both work; a STRING that happens to contain 1, 0.5, 0 does not.

One flat point instead of a curve. You wired a scalar where you meant the list. Plugging Number Sequence's values gives you the curve; a single FLOAT gives you a dot, which is correct and useless.

Comparing two schedules of different lengths. Each series is stretched across the full width by its own entry count, so a 10-entry and a 20-entry list both fill the plot and "entry 8" means different points in the run for each. Give both lists the same count if you're checking whether one schedule finishes before the other.

Reading precise values off a tiny node. Drag the node bigger. The per-series legend gives length, low and high; the exact entry under your pointer is the readout below the plot. Small nodes are for shape, big nodes are for numbers.