VODOOText to video AI: how to write a prompt that comes out the way you saw it
By VODOO AI · · 4 min read
Text to video AI turns a paragraph into a moving shot. The difference between a clip you'd post and one you'd delete is almost always the paragraph: what it names, in what order, and what it leaves for the engine to guess.
Here's how the engines read a prompt, the order that works, the camera words they understand, and prompts you can copy for the shots people make most.
Write it in the order a director would
Engines weigh the start of a prompt most. Put the subject and the action first, then the place and the light, then the camera, then the finish. One shot per prompt: a clip of five or ten seconds is one moment, not a story.
- Subject and action: who or what, doing exactly what.
- Place and time: where, and when, because the time of day is the light.
- Camera: how it moves, or that it doesn't.
- Finish: lens, film look, colour, grain.
A grey-haired fisherman in a yellow oilskin coils a wet rope on a wooden pier, his hands rough and slow. Early morning, sea mist, pale sun breaking through. Slow dolly in from waist height. Shot on 35mm film, muted colour, natural skin texture.Make this in VODOO →
Camera words engines understand
Locked-off / static camera
- You get
- No camera movement at all
Slow push-in / dolly in
- You get
- The camera moves closer, smoothly
Orbit around
- You get
- The camera circles the subject
Handheld
- You get
- Small natural shake, feels filmed by a person
Overhead / top-down
- You get
- Looking straight down
Tracking shot
- You get
- The camera follows a moving subject
| Say | You get |
|---|---|
| Locked-off / static camera | No camera movement at all |
| Slow push-in / dolly in | The camera moves closer, smoothly |
| Orbit around | The camera circles the subject |
| Handheld | Small natural shake, feels filmed by a person |
| Overhead / top-down | Looking straight down |
| Tracking shot | The camera follows a moving subject |
Making it look filmed, not generated
The giveaways of AI video are too-smooth skin, light from nowhere, and motion that floats. Name the texture, the light source and the weight of things.
A heavy glass bottle of olive oil tips and pours a thick golden stream onto a wooden board of torn sourdough; drops splash and the oil spreads slowly. Kitchen window light from the right, dust in the air. Locked-off camera, close-up, 100mm macro look.Make this in VODOO →
A narrow Lisbon street at dusk after rain: wet cobbles reflecting yellow tram lights, a tram passes slowly, two people under one umbrella step aside. Tracking shot at walking pace. Natural colour, slight film grain, real reflections.Make this in VODOO →
For more on this, our guide to realistic AI video has twelve fixes for the plastic look.
Sound and speech
Some engines now make sound with the picture. Veo does dialogue and ambient sound well; others make silent video and you add a voice afterwards with ElevenLabs. If you want speech, write the exact words in quotes and say who says them and how.
A young baker in a flour-dusted apron leans on the counter of a small bakery and says to camera, warmly and a little tired: "We start at four so you don't have to." Morning light, bread on the shelves behind her, slight handheld. Ambient bakery sound.Make this in VODOO →
Which text to video engine to use
Kling is strong on people and motion, Veo on sound and realism, Sora on longer cinematic shots, Wan on affordable takes, and Hailuo on lively, expressive motion. In VODOO you write the prompt once in Create and the studio sends it to the engine that suits the shot, or you pick. Start with 300 free credits a month.
Questions people ask
- How do I write a good text to video prompt?
- Subject and action first, then place and light, then one camera move, then the look. Keep it to one shot, and name textures and light sources so it looks filmed.
- Can text to video AI make videos with sound?
- Some engines make sound and speech with the picture. Write spoken lines in quotes and say who says them and how. Silent engines can be paired with a separate AI voice.
- How long can a text to video AI clip be?
- Most engines make clips of about 5 to 10 seconds per prompt, some longer. Longer videos are made by joining shots, which keeps each one sharp and on brief.