How to write AI video prompts that actually work

Great AI video starts with a clear prompt. The models are powerful, but they cannot read your mind. Here is the formula our team uses every day.

The five-part formula

1. Subject. Who or what is on screen. Be specific about age, clothing and expression.
2. Action. What happens during the clip. One main action reads better than three.
3. Setting. Where and when. Time of day changes everything about the light.
4. Camera. Describe the move: slow push-in, orbit, handheld follow, drone pull-back.
5. Mood and style. Film stock, color palette, references like "warm golden hour" or "neon noir".

A before and after

Weak: a woman in a city.

Strong: a young woman in a yellow raincoat walks through a rain-soaked Tokyo alley at night, neon signs reflecting in puddles, slow tracking shot from behind, cinematic anamorphic look, moody teal and magenta palette.

Use the prompt enhancer

In the studio, switch on Prompt enhancement and the AI rewrites your idea with camera and lighting details. You can always edit the result before you generate.

Keep shots short and focused

Most models shine at 5 to 8 seconds. For longer stories, use the AI Director. It plans several connected shots and stitches them into one film.