Video Gen Editor

Text to Video AI: Turn a Script Into Video

Write a sentence or paste a script, and text to video AI builds the scene: the characters, the setting, the camera moves and, on many models, the sound. Try the same prompt on a few leading video models and keep the take that looks most like the one in your head.

Free sign-up · see the cost before you generate

How to turn text into a video

  1. 1

    Write your prompt

    Describe one scene: who's in it, what happens, where, how the camera moves and the mood. For a script, go one shot at a time.

  2. 2

    Choose the format

    Pick a model, the shape (16:9, 9:16, 1:1 and more) and the length. If the model can generate audio, turn it on.

  3. 3

    Generate and refine

    Check the price and press Generate. Not quite right? Change a few words or try another model.

From script to video, one shot at a time

A video model makes one continuous shot per generation, usually 5 to 20 seconds, and up to 30 with Seedance 2.5. So to turn a full script into a video, break it into shots the way a storyboard artist would: a wide establishing shot, a close-up, a reaction. Give each shot its own prompt, copy your character descriptions word for word between them, and cut the clips together in your editor.

To keep a character or product looking the same from shot to shot, add a reference image. Seedance 2.5, Wan 3.0 and MiniMax H3 all use reference images to hold faces, outfits and objects steady.

How to write a text to video prompt

A good prompt answers a handful of questions. Here's one built up piece by piece:

  • Who or what: a few specific details go a long way (“a red-haired cyclist in a yellow rain jacket”).
  • What happens: one clear movement (“rides through a puddle, water splashing”).
  • Where and when: “a narrow Amsterdam street at dusk, wet cobblestones”.
  • The camera: “low-angle tracking shot”, “slow dolly-in”, “drone flyover”.
  • Style and light: “cinematic, 35mm film grain, neon reflections”.
  • Sound, on models with audio: “rain, distant traffic, a bicycle bell”.

Which model suits your text

Test the idea cheaply first, with LTX-2.5 Fast or a 360p run on Gemini Omni Flash, then run the prompt that works on a higher-end model. Seedance 2.5 handles long, multi-part instructions well and holds a scene together. Wan 3.0 can also read a document or a web page as context, which helps when you're turning an article into an explainer. Gemini Omni Flash 1.1 goes up to 4K when you need every detail.

Would you rather start from a picture? Use the AI image to video generator. Need to change a clip afterwards? Open it in the AI video editor.

What you can make from text

  • Faceless YouTube and Shorts

    Visuals for narrated videos, one prompt per line of your script.

  • Explainers

    Show a concept, a process or a place without hunting for stock footage.

  • Ads and promos

    Quick concept ads. For product-led ads, see the AI UGC video generator.

  • Storyboards and pre-viz

    Preview the scenes of a film or a pitch before the shoot.

Frequently asked questions

Is it free?

Signing up is free, and so is looking around: every model, setting and tool is open before you pay anything. Generating a video costs credits, and the price depends on the model, the length and the resolution. The estimate sits next to the Generate button, so you always see it before you commit. A quick 3-second test at 360p on Gemini Omni Flash costs about $0.09.

What is text to video AI?

It's a video model that works from a written description. You describe the scene and it creates every frame: the motion, the lighting and the camera movement.

Can I turn a whole script into a video?

Yes, shot by shot. Each generation is one continuous clip, so split the script into scenes, generate each one, and edit them together. Keep character descriptions identical, or use a reference image, so people look the same throughout.

Does the AI add sound?

Several models can, including Seedance 2.5, Wan 3.0, LTX-2.5 and FLUX 3 Video. They generate ambience and sound effects in the same pass as the picture. Describe the sounds you want in the prompt.

How long does it take to generate a video?

Anywhere from under a minute to a few minutes, depending on the model, the length and the resolution. Fast models like LTX-2.5 Fast come back quickest.

More AI video tools