How to write a prompt for text to video AI
Video models need more direction than image models: they have to know what moves, how the camera moves and for how long.
Describe motion, not just the picture
A cup of tea gives the model nothing to animate. The camera slowly pushes in as steam rises from a cup of masala chai on a window sill, rain outside gives it a subject, an action, a camera move and an atmosphere.
A simple prompt structure
- 1Shot: close-up, wide shot, aerial, handheld.
- 2Subject and action: who or what, doing what.
- 3Setting and light: where, what time of day, what kind of light.
- 4Camera move: push in, pan left, orbit, static.
- 5Mood or style: cinematic, documentary, anime.
Pick the model for the job
Models differ in realism, motion and length. Some make clips of a few seconds; others run longer. Some generate sound. The model picker shows what each one supports and costs, so try a short, low-resolution version first, then generate the final at full quality.
Shape and length for each platform
Generate in 9:16 for Reels and Shorts and 16:9 for YouTube. Short clips of 4–8 seconds are easier to keep consistent; join several in Motion Studio to build a longer video.
Label AI video where platforms ask
YouTube and Instagram ask creators to disclose realistic AI-generated content. Read our guide to AI content labels before you post.
Build a longer video from short clips
Most great AI videos are several short clips cut together, not one long generation. Plan the shots like a storyboard — an opening wide shot, a close-up, a reaction, an ending — generate each one, then arrange them on the timeline in Motion Studio. Keep the same style words and light in every prompt so the shots match. Add captions, music and titles there, and export one MP4 for Reels, Shorts or YouTube.
Credits and the Free plan
Every account gets 50 credits a month and can use every video model. Longer and higher-resolution clips cost more, and the cost is always shown before you generate. See pricing.
