Break Image Prompts Into Labelled Parts
You get a rough image prompt sorted into labelled parts, with clashes flagged and a tidy rebuilt version. Seeing each phrase's job makes it easier to spot padding and contradictions, such as asking for both watercolour and photorealistic.
Worked in our testIt followed every step. It split the phrases and labelled them with the right columns. It flagged photorealistic as clashing with watercolour, and flagged "very beautiful" and "masterpiece" as filler, with brief reasons tied to the poster. It correctly needed no translations. It reordered Subject, Style, Quality, Command, and gave one comma-separated line in a code block with nothing after it.
The prompt
Act as a prompt editor for AI image generators.
Task: Take my rough image prompt and turn it into a clean, well-ordered prompt by splitting it into labelled parts.
Steps:
1. Split {rough_image_prompt} into its individual phrases and keywords.
2. Label each one as Subject, Style, Quality, or Command (technical settings such as aspect ratio). If a phrase is not in English, give a short English translation.
3. Flag any phrase that clashes with another or adds nothing, and say why in a few words, bearing in mind {intended_use}.
4. Reorder the kept phrases: Subject first, then Style, then Quality, then Commands last.
Output: A table with the columns Phrase, Label, Keep or Remove, and Note. Below it, give the rebuilt prompt as one comma-separated line in a code block, with nothing after it.
Change the parts in curly braces. Use it when an image prompt has grown by trial and error and the results have become inconsistent, or when you want to reuse the same subject in a different style.
What happened when we ran it
We sent
Act as a prompt editor for AI image generators. Task: Take my rough image prompt and turn it into a clean, well-ordered prompt by splitting it into labelled parts. Steps: 1. Split a cosy independent bookshop at dusk, rain on the window, watercolour illustration, very beautiful, masterpiece, 4k, ultra detailed, soft warm lighting, photorealistic, --ar 3:2. The shop cat sleeps on a pile of hardbacks near the door. into its individual phrases and keywords. 2. Label each one as Subject, Style, Quality, or Command (technical settings such as aspect ratio). If a phrase is not in English, give a short English translation. 3. Flag any phrase that clashes with another or adds nothing, and say why in a few words, bearing in mind a printed poster for a bookshop's autumn reading event, in landscape format. 4. Reorder the kept phrases: Subject first, then Style, then Quality, then Commands last. Output: A table with the columns Phrase, Label, Keep or Remove, and Note. Below it, give the rebuilt prompt as one comma-separated line in a code block, with nothing after it.
The model replied
| Phrase | Label | Keep or Remove | Note |
|---|---|---|---|
| a cosy independent bookshop | Subject | Keep | Main subject; suits a bookshop event |
| at dusk | Subject | Keep | Sets the time of day and mood |
| rain on the window | Subject | Keep | Adds autumn atmosphere |
| The shop cat sleeps on a pile of hardbacks near the door | Subject | Keep | Good focal detail, reworded as a phrase |
| watercolour illustration | Style | Keep | Defines the look |
| soft warm lighting | Style | Keep | Fits dusk and a cosy autumn feel |
| photorealistic | Style | Remove | Clashes with watercolour illustration |
| very beautiful | Quality | Remove | Vague and adds nothing |
| masterpiece | Quality | Remove | Empty filler that doesn't guide the image |
| ultra detailed | Quality | Keep | Helps a printed poster stay sharp, though it may tighten the loose watercolour look slightly |
| 4k | Quality | Keep | Signals high resolution for print |
| --ar 3:2 | Command | Keep | Gives the landscape format the poster needs |
```
a cosy independent bookshop, at dusk, rain on the window, shop cat sleeping on a pile of hardbacks near the door, watercolour illustration, soft warm lighting, ultra detailed, 4k, --ar 3:2
```
Run on Claude Sonnet 5.5, 29 Oct 2026. Automatic check: 5 out of 5.
Mentioned for
Midjourney, ChatGPT, Claude
Where it came from
Found on GitHub, shared by Moonvy: Moonvy/OpenPromptStudio.
This is our own version of the idea, written and tested by us. The original may differ.