Turn any video into a model-ready prompt
Drop in a clip. We read it shot by shot and write the prompt each video model understands best.
What a video prompt is made of
Every good prompt answers the same six questions. The demo colours them so you can see where each one lives in the shot.
- Subject
Who or what is on screen, and what they are doing.
“an older cook lifts noodles from a steaming pot”
- Camera
Shot size, lens feel and angle.
“medium close-up, eye level”
- Motion
How the camera moves — and how fast time runs.
“slow push-in, then slow motion”
- Light
Where light comes from, its colour and mood.
“warm bulb from the left, cool blue shadows”
- Style
The look: film stock, grade, texture, genre.
“35mm film look, subtle grain”
- Sound
What we hear, for models that generate audio.
“rain on the awning, bubbling pot”
Models we write for
Each model reads prompts differently. Costs are for the shortest clip; 10 and 15 seconds cost proportionally more.
| Model | Good at | Output | Cost |
|---|---|---|---|
| Grok Imagine 480pThe free lane — quick drafts with sound | The free lane — quick drafts with sound | 480p | Free, 1 a day |
| Grok ImagineCheapest, with sound — not for on-screen text | Cheapest, with sound — not for on-screen text | 720p | 6 credits (6 s) |
| Seedance 2 miniReadable brand names and text; start/end frames, references | Readable brand names and text; start/end frames, references | 720p | 8 credits (5 s) |
| Kling 3.0Camera language, multi-shot, consistent characters | Camera language, multi-shot, consistent characters | 720p | 14 credits (5 s) |
Made from a single prompt
Each clip below was generated by us from the prompt shown with it.
See each prompt with what the model followed and missed
Product orbitShow prompt
A minimalist frosted-glass perfume bottle with no label stands on a travertine pedestal. The camera slowly orbits 180 degrees around it at eye level. Soft studio lighting from a large softbox, gentle highlight sliding across the glass, pale sand-colored seamless background. Clean commercial product shot, sharp focus, no text, no logo.
Generated with Seedance via Doubao
Windy portraitShow prompt
Close-up portrait of a young woman with short silver hair wearing an oversized moss-green knitted scarf, standing on a windy hillside. Golden-hour backlight, wind gently moving the scarf and hair, slow dolly-in toward her face as she turns to look past the camera. Soft film tones, shallow depth of field, no text.
Generated with Seedance via Doubao
Surfing corgiShow prompt
A corgi wearing a small orange life vest rides a surfboard on a small turquoise wave. Low tracking shot from water level alongside the board, bright sunny day, water splashing toward the lens, the dog's ears flapping in the wind. Playful, vivid colors, slow motion, no text.
Generated with Seedance via Doubao
Dunes at sunriseShow prompt
Aerial drone shot gliding forward over rolling red sand dunes at sunrise, long soft shadows across the ridges, a single line of camels crossing far below. The camera slowly tilts down as it moves forward. Warm golden light, epic scale, calm atmosphere, no text.
Generated with Seedance via Doubao
Generate here, or take the prompt anywhere
Free
One 480p video a day with sound, for personal use. It joins a queue; when it’s quiet it starts within minutes.
Instant
No queue, 720p, vertical or landscape, 5–15 seconds with sound, commercial use. Pay once for credits — they never expire.
Questions
What does “video to prompt” mean?
You give it a video; it describes what is in each shot — subject, camera, motion, light, style and sound — and turns that into a prompt a video model can follow to make something similar.
Why a different prompt for each model?
Models read prompts differently. Seedance follows timed shot lists, Kling responds to camera language, Wan rewards dense description, Grok Imagine prefers one short shot. The same idea is written the way each one understands.
Is the analysis free?
Yes. Turning a video into prompts is free. Generating a video costs credits, except for one free 480p video a day.
What happens to the videos I upload?
They are used only to write your prompt and are deleted automatically afterwards. We never use them to train models.