Skip to content
OpenMicroDrama

AI video prompt generator

Describe your cast and shots once, then pick Seedance, Kling or Veo. You get a prompt in that model's own format, with settings and the mistakes to avoid.

Describe the scene, let AI fill the form

Free, 3 a day without an account. It fills the form; you still check it.

Clip
Characters

Use the exact same look in every prompt of the episode. That's what keeps the face from changing.

Scene
Shots
Shot 1

One clear action. Show feelings through the body, not with big emotion words.

Shot 2

One clear action. Show feelings through the body, not with big emotion words.

Shot 3

One clear action. Show feelings through the body, not with big emotion words.

Seedance prompt
Elena: 24-year-old woman, chestnut hair in a low ponytail, cream silk blouse, tan trench coat.
Adrian: 32-year-old man, sharp jaw, slicked-back black hair, charcoal three-piece suit, silver cufflinks.
Exactly two people in frame.
Setting: a dark penthouse office with floor-to-ceiling windows, warm desk-lamp light, dark room.
Live-action, cinematic, vertical 9:16, glossy, polished interiors.
0–3s: Medium shot, slow push in. Elena pushes the office door open and stops, holding a printed contract.
3–7s: Close-up, static camera. Adrian looks up from his desk without standing. Adrian, coldly, says in English {You signed it without reading it?}
7–10s: Extreme close-up, slow push in. Elena's jaw tightens, her eyes wet, she doesn't blink.
No subtitles, no on-screen text.

Settings: Seedance 2.5 · 9:16 · 10s · draft at 480p, finish the keeper at 1080p

What the Seedance tab does

  • Puts things in ByteDance's order: subject, action, scene, style, camera, sound.
  • Uses whole-second timestamps on 2.5 ("0–3s", "3–7s") and "Shot 1, Shot 2" labels on 2.0.
  • Writes dialogue in the official format: says in English {You signed it without reading it?}
  • Pastes each character's look word for word and adds a headcount line, so you don't get twins.
  • Ends every prompt with "no subtitles, no on-screen text".
  • Flags what the guides warn about: big emotion words, readable text in the frame, and rain or echo cues on 2.5.

Want this model on its own page? Seedance prompt generator.

Questions

Is this AI video prompt generator free?
Yes. The form runs in your browser and there's no sign-up. Only the optional AI fill sends anything: your one-sentence idea goes to our server and an AI model, and we don't store it.
Why are there separate tabs for each model?
Each model reads prompts differently. Seedance 2.5 wants whole-second timestamps and dialogue in braces, Kling wants shot lengths and a negative prompt, and Veo wants timed cuts inside one 8-second clip. Pick the tab for the model you'll paste into.
Which model should I pick?
Seedance 2.5 for long dialogue takes, Kling for two people talking in one frame, Veo for lip-synced close-ups. Our model guides compare them in detail.
Does it only work for dramas?
The prompts are written for vertical micro dramas, but the format works for any short clip: ads, music videos, skits. Change the setting and the action to fit.