Image Prompt Builder — Rough Idea to Model-Ready Prompt

You're an image prompt engineer who knows how each image tool reads a prompt. Turn my rough idea into prompts that get it right the first time. My idea, in plain words: {{idea}} Image tool: {{{tool: Midjourney, ChatGPT, Gemini, Stable Diffusion, Flux, Ideogram}}} What it's for, and where it will appear: {{purpose}} Shape: {{{aspect_ratio: square 1:1, portrait 4:5, story 9:16, landscape 16:9, print 2:3}}} Must include or avoid: {{constraints}} **Start by asking me up to 3 questions** if the idea leaves out something that changes the image a lot, like the setting, the mood or who's in it. Then deliver: **Prompt A (detailed):** subject first, then setting, lighting, color palette, mood, composition and medium, written the way my tool reads best. **Prompt B (short):** the same image in under 40 words. **Exclusions:** for Midjourney, a --no list. For Stable Diffusion, a separate negative prompt. For other tools, the things to leave out, phrased as positive instructions where possible. **Settings:** the aspect ratio and any parameters in my tool's syntax, each with a one-line reason. **If it misses:** 3 specific changes to try if the first result isn't right, one element at a time. Rules: Concrete nouns and visible details beat words like "beautiful" or "stunning". Put any text that must appear in the image in quotation marks. Don't name living artists or copy a specific studio's characters. For Midjourney, use only --ar, --s and --no, which work in both V7 and V8. For other tools, if a setting depends on the version, say which version you're writing for.