Guides

How to Write AI Video Prompts

A simple structure for AI video prompts: subject, action, camera, style and length, with examples that work for text-to-video and image-to-video tools.

Published October 7, 2026

A good video prompt describes one short shot. Most tools handle a few seconds of footage, so keep the idea small and clear.

The structure

Write the prompt in this order:

Subject: who or what is in the shot.
Action: one clear movement.
Camera: for example a slow push-in, a pan, a handheld follow or a static wide shot.
Setting and light: place, time of day, mood.
Style: film look, colour palette, lens feel.

A fisherman pulls a net onto a wooden boat at sunrise. Slow push-in, handheld feel, warm golden light, light mist over calm water, 35mm film look.

Image to video

When you start from a still image, the picture already defines the subject and style. Describe only the motion and the camera.

The woman turns her head and smiles. Gentle camera drift to the right, hair moves in a light breeze.

Tips

One action per shot. Several actions in one clip often fail.
Name the camera move, not just the scene.
Keep it short; long prompts can contradict themselves.
Generate several versions and keep the best, then edit them together.
For a consistent character across shots, start each clip from the same reference image.

Compare tools in best image-to-video AI tools and read How to write better AI image prompts for the basics.