Text to video
Describe a scene in words and the AI builds the video. No photos, no clips — just a sufficiently specific prompt.
How to use it
- Write the video prompt. It is required and has a character counter.
- Attach images if you want the AI to follow a specific look. Optional.
- Enter narration if you want audio, or switch the voice-over off.
- Press Create and save to library. Track progress on the Videos page.
Writing the prompt
Describe the shot you want; the more specific, the closer the result. Five things worth including:
| Element | Example |
|---|---|
| Setting | a Hanoi pavement café in the morning |
| Subject | a middle-aged woman running a street stall |
| Action | pouring coffee, then looking up and smiling |
| Lighting | early sunlight slanting through leaves |
| Camera | slow dolly in from a distance |
Do not put dialogue in the prompt
The prompt builds visuals. Dialogue written here is never spoken — audio comes from the Narration field below. This is the most common misunderstanding with this tool.
Narration
The narration field has a character budget shown beside it, derived from the video length. Writing past it means the narration cannot fit the visuals.
If you do not want audio, switch the voice-over off rather than leaving the field empty — the system asks you to do one or the other before submitting.
Notes
- This tool bills per second. Going past the reference duration still renders; you pay for the extra seconds.
- For longer pieces, render several short segments and join them rather than forcing one prompt to carry a long video — quality holds up better.
- To build from your own photos, use video from photos; to assemble existing clips, use Storytelling.