Every product parses prompts differently, but structure travels well. Fill in this order instead of stacking adjectives blindly.

Suggested order

  1. Subject: who/what, count, pose or action.
  2. Environment: indoor/outdoor, era, tidy vs messy.
  3. Style: photo vs illustration vs 3D; reference artists or film stock if useful.
  4. Light & time: soft/hard light, golden hour, neon night.
  5. Lens: wide vs tele, depth of field, high/low angle.
  6. Negative prompts (if supported): no watermark text, no extra fingers, no heavy distortion.

Tips

  • Broad first, tight second—explore composition, then iterate details.
  • Run the same scaffold across models to see which fits your taste.
  • Copyright & likeness: check training policies and portrait rights before commercial use.

Image gen is trial and error—structured prompts cut wasted rolls.