Start with four useful decisions
Decide who is in the scene, where they are, what changes and where the camera sits. One clear sentence for each decision gives you a brief that is easier to debug than a long list of visual adjectives.
- Character: which reference belongs to each person.
- Setting: one place and a simple lighting condition.
- Action: one main movement per shot.
- Camera: framing and one movement, if needed.
Describe what should stay the same
For connected shots, describe the direction of movement and the characters’ positions. These details give the transition a visual anchor. They matter more than asking for a transition to be cinematic.
Change one instruction at a time
If a result misses the gesture, simplify the action before adding more description. Keep a short note about what changed between attempts. The sample prompt on this site is an editable starting point, not a tested promise for a particular model.
A note from the AI Trend Clips guide library