Wan Prompt Generator

About the “Wan Prompt Generator” template

Builds an English prompt in Wan's own syntax: a separate negative_prompt, timed multi-shot and quoted lines, tuned to Wan 3.0, 2.7, 2.6–2.5 or open-weights 2.2. It is one of the templates in AI Video.

What it will ask you (19)

1. Wan version you will use
2. Generation mode
3. General idea of the video (e.g. a neighbourhood newsagent closes her kiosk in the rain)
4. Main character or subject, original and fictional, with concrete traits (if there are several, describe each)
5. Main movement or action (one clear action per shot)
6. Scene and atmosphere (place, era, weather, background objects)
7. Shot type / framing
8. Camera movement
9. Lighting
10. Visual style (pick up to 2)
11. Length in seconds (Wan 3.0: 2–30 s; wan2.7: 2–15 s; for other versions, check the maximum)
12. Aspect ratio
13. Resolution
14. Split the video into several timed shots (Wan 3.0, wan2.7 and wan2.6 only)
15. Sound (Wan 2.2 does not generate sound)
16. Exact spoken text, if any (short lines; say who says each one)
17. How the voice sounds: emotion, tone, pace, timbre and accent (e.g. tired, low, slow)
18. Language of the dialogue or voice-over
19. Things you do NOT want to appear (comma-separated; they go into the negative_prompt field)

Example result

What the AI returned with this prompt and some sample answers.

# Wan prompt · a baker sets bread rolls in the shop window (wan2.7)

SUMMARY AND ASSUMPTIONS
  wan2.7, text to video, multi-shot with two shots and one spoken line.
  One unique label for the character, no pronouns. The line is the user's own.

PROMPT FOR WAN
  Early morning inside a small neighbourhood bakery, realistic cinematic look, warm
  interior light and steam on the window.
  Shot 1 [0–3 s] medium shot, the baker in the flour-dusted apron slides a tray of
  round bread rolls onto the window shelf, camera pushes in slowly.
  Shot 2 [3–6 s] close-up, the baker in the flour-dusted apron wipes both hands on a
  cloth, smiles and says, "Fresh out of the oven." Cheerful, warm, unhurried voice
  with a soft timbre.
  Sound: tray sliding on wood, a quiet oven hum. No background music.

NEGATIVE_PROMPT
  text, logos, watermark, distorted hands, flicker, extra people

SETTINGS IN THE TOOL
  - Version: wan2.7 · Mode: text to video
  - Duration: 6 s · Format: 16:9 · Resolution: 720p
  - Sound: on (to be checked).
  - prompt_extend: on by default, and it rewrites the prompt; check it and
…

How to use it

1. Answer the questions in the form above.
2. Generate the prompt and copy it, or open it straight in your AI.
3. Paste it into ChatGPT, Claude or Gemini. If the result doesn't fit, ask the AI to adjust it or generate it again with other answers.

Frequently asked questions

What do I get with “Wan Prompt Generator”?

A complete prompt tailored to your answers, ready to paste into ChatGPT, Claude or Gemini. Builds an English prompt in Wan's own syntax: a separate negative_prompt, timed multi-shot and quoted lines, tuned to Wan 3.0, 2.7, 2.6–2.5 or open-weights 2.2.

How long does it take?

You answer 19 guided questions; most have options to pick from, so it usually takes a couple of minutes.

Is it free?

Yes. Fill in the form, sign in with Google and copy the prompt, at no cost.

Do you store what I type?

Yes, in your history: we keep your answers and the generated prompt, linked to your account, for 12 months. You can see and delete them in “My account”. For health, diet, fitness and finance templates we store nothing.