I want an image of: [rough idea]. Use: [purpose, e.g. blog header 16:9].
Write one prompt for each of these models, using each model's own syntax and strengths:
- Midjourney V8.2 (parameters at the end: --ar, --s, --raw, --hd as appropriate)
- GPT Image 2.5 (clear paragraph, quoted text, size and quality settings)
- Nano Banana 2.1 (narrative description, camera and light, aspect ratio)
- FLUX 3 Image (medium/style first, then subjects and placement, then light, color, background; no negative prompt)
- Ideogram 4.5 (plain language, or JSON if there is text to place)
For each, add one line on why it's written that way.