Glossary · Making it with AI
Text-to-video
Text-to-video is a kind of AI that makes a short moving clip from a written description alone, such as a scene, an action and a camera move.
Why it matters
It is the fastest way from a written idea to a moving picture. But with words as the only guide, the result can differ from what was meant, and from one attempt to the next.
The written description, often called a prompt, says what should be on screen, what happens, where, in what light, and how the camera behaves. The AI makes a short clip that tries to match it. Longer videos are built by making many short clips and editing them together.
Because words leave a lot open, two attempts with the same description can look quite different. Faces, clothing and small details may change between clips, and the AI may add things nobody asked for.
Careful makers use it where an exact look doesn't matter, such as a sky, a crowd or a stretch of sea, and start from an approved picture instead when a person, object or place has to look the same every time.
For a short about the history of lighthouses, a wide shot of waves breaking on a rocky coast at dusk is made from a written description, while every shot of the lighthouse keeper starts from an approved picture so his face doesn't change.
Last reviewed 15 October 2026.