Text to Video AI Generator

What happens after you paste the script

The script is read once and split into scenes at natural beats, roughly one scene per sentence or clause depending on pacing. Each scene gets its own image or video prompt, written against the visual style you picked so all of them belong to the same video.

The narration is generated in a single pass that returns the timing of every word. That timing is then the authoritative clock: caption highlights and scene changes are both derived from it, rather than estimated from character counts. It is the reason the cuts land on sentence boundaries instead of a beat late.

Visuals are generated per scene, composed with the transition and music settings you chose, and rendered. Two to five minutes for most videos.

Writing a script that renders well

Short declarative sentences produce clean scene boundaries. Long sentences with several clauses tend to produce one scene holding on screen while a lot is said, which reads as slow.

Concrete nouns give the image generator something to work with. A sentence about a specific object in a specific place renders far better than an abstract statement, because the prompt written for that scene has something to depict.

Avoid stage directions in the script. Anything you write becomes narration and will be read aloud, including a note to yourself about what the shot should look like.

Still images or generated video clips

Image output generates a still per scene and moves the camera slowly across it. It is cheap, about 100 credits a minute, and it is the right default for narration-led scripts where the words carry the video.

Animated output generates a real video clip per scene, with motion and native sound. It is 35 credits a second in HD, so a minute is 2,100 credits, and it is worth it when the visual is the reason someone watches rather than the narration.

Animated videos accept a script of up to 1,200 characters. Image videos take considerably longer scripts.

Voices and captions

There are 36 voices with previews. Whichever you pick is stored with the video, so a recompile later sounds like the original rather than the current default.

Captions are word-synced, in two styles with a choice of highlight colour. They require a voiceover, since there is nothing to sync to otherwise. You can also generate a video with no narration at all, in which case the timing comes from the scene lengths you set.

Changing your mind afterwards

The editor lets you rewrite the narration and recompile. When the words change, the voiceover and the captions regenerate together in one aligned pass and the scene boundaries are re-mapped to the new audio, so nothing falls out of sync.

When the words did not change, the existing audio and captions are reused and you only pay for the render. Regenerating a single scene's visual works the same way, at the cost of that one scene.

Formats and output

Image videos render in 9:16, 16:9, 1:1 and 4:5. Animated videos render in 9:16, 16:9 and 1:1, at Standard 480p, HD 720p or Full HD 1080p.

The output is an ordinary MP4 with no watermark on any plan, and you hold full commercial rights to it.