Prompts · Lesson 1 of 6 · 6 min read

How to write AI video prompts that work

AI video models follow what you describe, so a clear prompt is the fastest way to a usable clip. This structure works across Kling, Veo, Seedance and Wan.

The six parts of a good prompt

  • Subject: what is on screen, described plainly. “A glass bottle of cold brew coffee.”
  • Action: what moves. “Condensation slides down the glass.”
  • Setting: where it is. “On a wet black stone surface.”
  • Camera: how we watch. “Slow push-in.”
  • Light: the mood. “Soft side light, dark background.”
  • Style: the finish. “Premium commercial look, realistic.”

Put it together

Weak
Coffee ad, make it cool.
Strong
A glass bottle of cold brew coffee on wet black stone, condensation sliding down the glass, slow push-in, soft side light on a dark background, premium commercial look, realistic detail.

Tips that save credits

  • One idea per clip. Several actions in five seconds usually look rushed.
  • Describe motion, not only the scene: AI video is about what changes.
  • Say what must stay the same when you animate a photo: “the product stays exactly as in the photo”.
  • Keep text out of the video prompt. Add words later in Finish, where they are sharp and spelled right.
  • Short clips first. Try 5 seconds, then make the best idea longer.

Starting from a photo

When you animate a photo, the photo already sets the subject, setting and light. Your prompt should mostly describe the motion and the camera: “Slow orbit around the product, gentle reflections moving across the label.”

Updated 29 September 2026