A great AI video starts with a great prompt. This guide shows you how to write prompts that get the result you want on the first try — whether you're generating from text or animating an image.
The anatomy of a good prompt
Every strong prompt answers four questions: what is in the scene, how it looks, how it moves, and what mood it carries.
Prompt recipes by use case
Product showcase
Copy"Studio product shot of [product] on a marble pedestal, soft key light, slow 360° rotation, clean minimal background, premium commercial feel"
Cinematic scene
Copy[Subject] in [setting], [time of day] lighting, [camera move] shot, shallow depth of field, film grain, [mood] atmosphere
Social clip
CopyVertical 9:16, [subject] doing [action], bold colors, fast cuts energy, trending aesthetic, eye-catching first frame
Pro Tips:
- 1
Front-load the subject. Put the most important element first — the model weights early words more heavily.
- 2
Name the camera. "Drone shot", "close-up", "tracking shot" give instant cinematic structure.
- 3
Describe motion explicitly. Static descriptions produce static-feeling video. Add verbs: gliding, panning, rising, swirling.
- 4
Use negative prompts for what to avoid. "No text overlays, no watermarks, no blurry faces" prevents common artifacts.
- 5
Iterate in small steps. Change one element at a time so you learn what each word does.
Try it now
Copy one of the recipes above, swap in your own subject, and paste it into the Video Generator. Compare the result against a vague prompt — the difference is night and day.
Open Video Generator →