Write in the 8-part order
Models respond best to one fixed order. Keep it every time:
- Format & duration
- Subject
- Action
- Environment
- Camera
- Look
- Timing
- Audio & constraints
Every prompt here is written in a production format — subject, action, camera, light, timing, audio — so you get a usable take instead of a lucky accident. Search it, copy it, drop in your own details.
∅ No prompts match those filters yet. Try a different keyword or reset the filters.
Two rules do most of the heavy lifting. Follow both and your output stops looking like a stock AI clip.
Models respond best to one fixed order. Keep it every time:
One continuous take with a timestamp per beat. One camera behaviour per beat, one location per film.
Never rewrite a prompt from scratch. Replace only these:
[SUBJECT] — species, age, wardrobe[LOCATION] — one single environment[STYLE] — photoreal / 3D / documentary[DIALOGUE] — or explicitly “none”