Render float v5.1.0

Render float does one deeply simple thing: takes a FLOAT and draws it as text on an image, so it shows up in the image preview pane instead of vanishing into a widget. It uses matplotlib under the hood - a figure is created, your number is placed as a large text label on a white canvas, and the result is returned as a normal IMAGE. Feed it 0.7 and you get a small white image that literally says 0.7.

That sounds almost too trivial to justify a node, but if you've ever tried to debug a pipeline by watching a float cross a graph, you know the pain: numbers don't render in the preview pane, they only live on nodes you have to mouse over. This turns any float into something you can pin to the canvas, save alongside a batch, or burn into an image for a record. In the pack's context it's a debug aid for the numeric-heavy nodes - histogram limits, mean/std, threshold values - where "what value did that actually compute?" is a question you ask constantly while tuning.

Input and output

• value - a FLOAT, 0.0 to 1.0 (default 0.0). That range is the only real limitation: this isn't a general number-to-text tool, it's calibrated for the 0–1 normalized values this pack deals in.

One output, an IMAGE (a 10×4-inch figure rendered as a tensor). Wire it to a PreviewImage or a SaveImage and the value lands in your output folder alongside everything else.

How to install

Same pack story as the rest of the tri3d nodes - Manager (search "tri3d") or:

cd ComfyUI/custom_nodes
git clone https://github.com/TRI3D-LC/tri3d-comfyui-nodes
# restart ComfyUI

Matplotlib is in the pack's requirements, so if the pack installed cleanly this node has what it needs. Keep the folder named tri3d-comfyui-nodes - the module path depends on it.

Where it earns its keep

The honest use case is debugging and documentation. Couple it with the pack's tri3d-get_mean_and_standard_deviation or tri3d-get_threshold_for_bg_swap and you can preview the computed values right in the graph instead of reading the console. In a batch run, saving the float alongside each image gives you a per-sample audit trail of what value each output was tuned to - genuinely handy when you're sweeping a threshold across fifty images.

Gotchas

• The input is capped at 1.0. If your float lives outside 0–1 (an arbitrary threshold, a raw histogram bin), normalize or scale it before it gets here, or the widget won't even let you type the value in.
• It renders text, nothing else - don't come here looking for sparklines or graphs. It's a value label with extra steps.
• Matplotlib first-import is slow (a second or two) the first time it fires in a session. Harmless, just don't panic at the delay.

Is it worth installing a whole pack for? On its own, no. As one node in a try-on rig you're already running, it's a nice little quality-of-life addition that turns invisible numbers into something you can look at. That's the whole job, and it does it.