Prompt Stack Pixaroma
Build a prompt from chunks you can toggle on and off
- text
Prompt Stack Pixaroma is a prompt built from building blocks. Each row of the node holds one chunk of your prompt - a subject, a lighting recipe, a quality booster, a style word - with a short label so you remember what it does. Toggle any row's orange ON/OFF pill to include or skip it, drag rows to reorder, and the node joins all the enabled chunks into one text output. Testing prompt variants becomes clicking pills instead of editing and re-editing a wall of text.
It's the structured sibling of the plain Text node: same idea, but the prompt is decomposed into labeled, independently switchable parts. For anyone who iterates on prompts a lot, that decomposition is the whole game - "which words actually change this image?" becomes a toggle experiment instead of a guess.
How it works
Add rows, type a chunk in each, label it, and set each row's ON/OFF pill. All enabled rows get joined into the single text output (a STRING) with a separator - comma+space by default, configurable in Settings → 👑 Pixaroma → Prompt Stack to newline, space, pipe, or anything you type. Drag the handle on the left of a row to reorder, and the join order follows. Everything saves with your workflow, so the exact state comes back when you reopen it.
There are no input sockets at all - the schema is just the text output - so it's purely a prompt source. Wire the output straight into a CLIPTextEncode, or into the pack's Text Pixaroma if you want to inspect it first.
How to use it well
The workflow that makes this shine: keep the stable parts of your prompt in rows you rarely touch, and make the thing you're experimenting with its own row. Then, to test, you flip one pill rather than re-reading a full prompt to find the word to delete. A few tips:
- Label rows so you don't have to read the chunk to know what it is - "subject", "lighting", "quality", "negative" read faster than the text.
- Long chunks scroll inside their own row, so a quality booster that runs three lines doesn't blow up the node.
- Because each row is one semantic unit, you can stack alternative versions of the same idea and toggle between them - a "day" row and a "night" row, say, instead of editing text each time.
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.
Two things to keep in mind. If your joined prompt has weird separators, it's the settings - the separator choice is global, not per node. And the standing pack caveat: after updating, hard-refresh with Ctrl+Shift+R (Cmd+Shift+R on Mac) if a node looks blank; the browser cache is the usual suspect. Also worth knowing: rows you switch off are skipped, not deleted, so a half-configured stack is safe to leave in place while you build.
Inputs (0)
No inputs
Outputs (1)
| Name | Type | Description |
|---|---|---|
| text | STRING | All enabled chunks joined into one string using your separator. |