小珠光大字展示

Sometimes the most useful node in a workflow is the one that just shows you a value. 小珠光大字展示 (Xiaozhuguang Big Display) takes any input - a string, a number, an integer, even a list - and renders it as large text directly on the node. It's the 小珠光 version of the showAnything-class display node, and it's the thing you drop at the end of a messy graph when you need to see what a value actually resolved to.

You'd think this is trivial, and it is. That's the point. When you're debugging a pipeline full of wildcard plumbing, converted widgets, and data-block gates, the fastest way to learn "did the seed change?" or "what did the get-widget node output?" is to wire it into a big display and look at the canvas instead of the console.

How it works

It's an output node with a single optional wildcard input (input). Whatever arrives gets stringified: strings and numbers pass through directly, short lists render each element, anything bigger gets dumped as JSON. There are no outputs - it's a terminal, like a preview or save node. It uses an IS_CHANGED that always returns true (a @classmethod that returns True, after a version where a plain boolean property tripped ComfyUI's cache signature check and threw warnings - fixed in v12.11.1), so the display always refreshes with the latest value.

On the front end the text scales freely with the node size - drag the node bigger, the text grows with it - and it's clipped to the content area so it never draws over the input port. It's unapologetically a UI convenience, and it costs you nothing in graph performance.

The input

• input - optional, any type. Wire up a STRING, INT, FLOAT, or even a wildcard output. Leave it unconnected and you get an empty display, which is about as useful as a note node.

That's the whole API: one wire, no settings.

Install

From the ComfyUI-xiaozhuguang pack - Manager (search "ComfyUI-xiaozhuguang"), or:

cd ComfyUI/custom_nodes/
git clone https://github.com/xiaozhuguang/ComfyUI-xiaozhuguang.git

Restart. Nothing to download.

Gotcha

Because it always re-runs (that's the point of a live display), it sits at the end of the graph's cache path - but since it has no downstream, the always-dirty cost is contained to itself. If you leave a huge batch wired into it, the JSON dump can be slow and ugly; for debugging a single value, feed it one scalar, not the whole tensor.