What is an AI video prompt generator?

A tool that turns an idea into the text you give a video model. This one asks for the parts of a shot (subject, action, setting, camera, look, sound) and writes them as one prompt in a fixed order, sized for the model you pick.

Does it use AI to write the prompt?

No. It arranges the words you type and does not add, translate or rewrite them. That keeps the prompt yours and means nothing is sent anywhere, but it also means the page cannot invent detail for you: what you leave empty stays out of the prompt.

Is the prompt tested on the model?

No. A prompt from this page has not been generated. What is checked is what can be checked without generating: the length against the model's limit, and the durations, shapes and starting point the model accepts on Cicadas. For prompts with recorded results, each model page shows the prompts behind its sample clips.

Which models can I write for?

Every video model that is open on Cicadas and has its own page; the table under the tool lists them with their durations, shapes, sound and prompt limit. The text is ordinary prompt text, so you can also paste it into another service that runs the same model.

How is an image to video prompt different?

The image already shows the subject and the style, so describing them again can pull the clip away from your picture. Choose My image and the page writes the motion, the camera and the sound, and leaves the look out. A few words naming the subject are enough.

How do I write sound and speech into a prompt?

Use the Sound field for what is heard (rain, footsteps, no music) and the two speech fields for who speaks and the exact line; the line is written in quotation marks. On Cicadas some models add sound to every clip and others have an Audio switch that can change the price; the page tells you which applies to the model you picked.

Is it free, and is anything saved?

The prompt generator is free and needs no account. Your brief is saved in this browser only, so it is still there when you come back, and it is not sent to Cicadas. Generating a video on a model page uses credits; the price is shown there before you start.

AI Video Prompt Generator: One Brief, Fitted to the Model

Free AI video prompt generator: fill in subject, action, camera, look and sound, pick a model, get a prompt within its limits and open it in the generator.

Free
no sign-up
runs in your browser

Write a prompt for one model

Runs in your browser. It arranges what you type; it does not call a model, rewrite your words or use credits.

Model

Start from

SubjectWhat happensOne or two actions fit a short clip. More than that and models tend to skip some.Where and when · optionalCamera · optional

Look · optional

Sound · optional

Who speaks · optionalSpoken line · optional

DurationAspect ratio

Prompt for Veo 3.1 Fast

0 of 1794 characters

Duration: 8 seconds
Aspect ratio: 16:9
Sound: sound off

These three are chosen on the generator, not written in the prompt. Only the prompt text is carried over.

Not generated or tested: this is your brief in a fixed order. Generating on the model page uses credits.

What each model accepts

From the settings Cicadas sends to each model. Prompt limits reserve room for the model page’s built-in scene instructions. Pick a row to write for that model.

Model	Seconds	Aspect ratios	Sound	Starts from	Prompt limit
page	4, 6, 8	16:9, 9:16	optional, credits ×1.5	text or image	1794 characters
page	5, 10	16:9, 9:16, 1:1	optional, credits ×1.5	text or image	2353 characters
page	4, 5, 10, 15, 30	21:9, 16:9, 4:3, 1:1, 3:4, 9:16	optional, same price	text or image	2000 characters
page	4, 5, 10, 15	21:9, 16:9, 4:3, 1:1, 3:4, 9:16	optional, same price	text or image	2000 characters
page	5, 10	16:9, 9:16, 1:1, 4:3, 3:4	always	text or image	2000 characters
page	5, 10	16:9, 9:16, 1:1	optional, same price	text or image	2000 characters
page	5, 10, 15	16:9, 9:16	always	text or image	2000 characters
page	5, 10	16:9, 9:16, 1:1	always	text or image	2000 characters
page	6, 8, 10	16:9, 9:16	optional, same price	text or image	1500 characters
page	5, 10, 15, 20	16:9, 9:16, 1:1	optional, same price	image only	2000 characters
page	5, 10, 16	16:9, 9:16, 1:1, 4:3, 3:4	optional, same price	text or image	2000 characters
page	5, 8, 10, 15	16:9, 9:16, 1:1, 4:3, 3:4, 21:9	optional, credits ×1.34	text or image	680 characters

About AI video prompt on Cicadas

Fill in who does what, where, how the camera moves, how it looks and what is heard. Pick the model you will use and the page arranges your brief into one prompt, checks it against that model's prompt limit, and shows the durations and shapes the model accepts. One click carries the prompt to that model's generator on Cicadas. It is free and runs in your browser. It arranges your own words in a fixed order; it does not call a language model, and a prompt it writes has not been generated or tested.

Why use Cicadas for AI video prompt

A shot brief, not a blank box

Eight short fields: subject, what happens, where and when, camera, look, sound, who speaks and the spoken line. Camera and look have presets you can click and then edit. The prompt is always written in the same order: who does what, the setting, the camera, the look, then sound and speech.

Fitted to the model you picked

Models differ in ways a prompt has to respect. On Cicadas, Veo 3.1 Fast makes 4, 6 or 8 seconds in 16:9 or 9:16; Seedance 2.5 goes up to 30 seconds in six shapes; PixVerse v6 accepts a prompt of at most 680 characters where most accept 2,000; FLUX 3 only animates an image. The page limits your choices to what the model takes and tells you when the prompt is too long.

From an image, the prompt changes

When you start from your own image, the picture already sets the subject and the look. Switch to My image and the prompt keeps the motion, the camera and the sound, and leaves the look out. The table under the tool shows which models start from text, from an image, or both.

How to use AI video prompt

01
Pick a model and whether you start from text or from your own image.
02
Fill in the subject and what happens; add setting, camera, look, sound and a spoken line if you want them. The prompt updates as you type.
03
Copy the prompt, download it as a text file, or click Open to carry it to that model's generator.

Example prompts in this order

Arranged by the generator on this page, in the same order. They have not been generated or tested on any model.

“A ceramic robot with a watering can waters a row of tomato plants, then looks up. Setting: inside a glass greenhouse at dawn. Camera: slow push-in at eye level. Look: soft morning light, realistic materials. Sound: water trickling, birds outside, no music. The robot says: "Good morning, little ones."”
“A barista in a denim apron pours steamed milk into a cup and draws a leaf. Setting: a small cafe counter by a rainy window. Camera: overhead top-down shot. Look: documentary, natural light. Sound: milk hissing, rain on glass, quiet room.”
“The woman in the red coat turns toward the camera and smiles as snow starts to fall. Camera: slow orbit around the subject. Sound: wind, distant traffic.”

AI video prompt FAQ

Tested on Cicadas · reviewed 2026-10-08

Tested on 2026-10-08 in Chrome, in English and Chinese; not tested in Safari or Firefox. The tool was used the way a visitor would: fill the brief, click the camera and look presets, switch between Veo 3.1 Fast, Seedance 2.5, Kling 3.0 Pro, MiniMax H3 Max, FLUX 3 and PixVerse v6, copy, download and reload. The prompt came out in the fixed order, the durations and shapes on offer changed with the model, a 30-second choice fell back to 5 seconds on a model without it, FLUX 3 refused a text-only start, the look was left out and reported when starting from an image, and a 1,097-character prompt was refused for PixVerse v6 (680) and accepted for Kling 3.0 Pro (2,500). Open in Kling 3.0 Pro opened that page with the same text in its prompt box, on both language versions, and the stored copy was removed after one use. No request other than page loads was sent while writing. What was not tested: whether a prompt in this order gives a better clip on any model. No video was generated for this page, and the model facts come from the settings Cicadas submits with, not from a new measurement.

Updated Oct 8, 2026 · How we test

Scene inspiration · this tool writes prompts

Free prompt builder · No video, no creditsNew

Seedance Prompt Builder

Build Seedance prompts free in your browser: seven inputs become one English Seedance 2.5 or Seedance 2.0 prompt to copy or download. No video is generated.

Free · runs in your browser

Tested 2026-09-21

Kling 3.0 Pro · Recorded Cicadas sample

One English prompt → private MP4New

AI Text to Video Generator

Turn one English prompt into a private MP4 with AI text to video. Pick Kling for a silent 5-second clip or Veo for text to video with audio, 6 seconds at 720p.

Kling 3.0 Pro · from 140 credits

Tested 2026-09-30