Show Text Pixaroma
Inspect anything in your graph — even tensors — and still pass it on
- source
- text
Show Text Pixaroma is the "what is actually flowing through this wire" node. Wire anything into it - a string, a number, a latent, an image tensor, even a type you don't recognize - and it prints a human-readable summary in a text box you can select and copy from. Then, because it's polite, it passes the same text out a STRING output so you can inspect a prompt and still feed it forward.
Debugging a workflow is 80% "what does this node actually output," and the built-in ComfyUI answers to that question are thin. This node is the version that understands tensors: it doesn't dump a million numbers, it summarizes them.
How it works
One input, source (wildcard *, so it accepts anything), and one output. What you see in the box depends on what you fed it:
- Tensors - shape, dtype, min, max. That's the summary a debugging session actually wants: is it the right shape, is it the right range, is it float or int?
- Latents - the sample shape.
- Anything else -
str(value), so prompts, numbers, filenames, whatever, just print.
The box is a real read-only text widget: you can select text and copy it, which sounds trivial until you've fought with ComfyUI text you can't select. Long text scrolls instead of forcing the node to grow, and you can resize the node freely. The same string passes through unchanged to the text output, so you can drop the node mid-wire - inspect the prompt, and still send it on to CLIPTextEncode or wherever it was going.
When you'd reach for it
- A prompt node's output looks wrong in the result, and you want to see the exact string before it hits the encoder.
- A node is erroring with "expected X shape" and you need to know what shape you actually have.
- You're learning how a custom node works and want to see what its hidden values look like - a STRING output of a node you're curious about beats guessing.
It's a debugging tool first and a passthrough second, so it's not something that lives in a finished workflow - but during the build, it earns its place.
Install and gotchas
Standard Pixaroma install - ComfyUI Manager (search "Pixaroma") or:
cd ComfyUI/custom_nodes
git clone https://github.com/pixaroma/ComfyUI-Pixaroma
Restart. No dependencies or model downloads.
One honest caveat: the summary is a summary - for a tensor it tells you shape, dtype, min, and max, not the values. If you need raw numbers, this isn't that node. And the pack-wide note: after an update, hard-refresh with Ctrl+Shift+R (Cmd+Shift+R on Mac) if a node renders blank, because the browser cache is the usual suspect.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| source | * | Anything to inspect. Tensors are summarized as shape / dtype / min / max; latents show their sample shape; everything else uses str(value). The string representation is also passed through unchanged to the 'text' output. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| text | STRING | The same text shown in the box, passed through so you can keep chaining it. |