How it works

From script to finished reel

You paste a script and press Generate. Here is everything that happens next, in order, and what the AI is and is not allowed to do.

Make 3 reels free

1. A content check

Your script is checked first. Scripts with hate, sexual content, scams, medical misinformation or song lyrics are stopped before anything is drawn, and the credit comes back automatically. See the content rules.

2. The AI writes a storyboard, not a video

An AI model (Anthropic's Claude) reads your script and any images you added, and writes a storyboard: a plan listing each scene, its layout, its narration, the on-screen text, the character and its mood, and roughly how long it lasts. It chooses from our library of scene templates (title cards, quizzes with countdowns, list reveals, myth vs fact, stat counters, comparisons, character speech, image showcases and calls to action).

The AI decides the shape for each script: how many scenes, in what order, and how long the reel is. A quiz, a story and a promo come out with different shapes. Only a few rules always apply: open strong in the first two seconds, one idea per scene, captions short enough to read, and a call to action only when your script has one.

Your script is treated as text, never as instructions. The AI is told not to follow anything written inside it, and its output is a plan that we check against strict limits. Nothing the AI or you write is ever run as code.

3. Voice

An AI narrator reads the script one caption chunk at a time, so each caption appears exactly when its words are spoken.

4. Drawing every frame

Our render engine draws all 30 frames of every second at 1080×1920: backgrounds, characters with blinks and idle motion, text, countdowns and reveals. Captions and key text are kept clear of the areas where TikTok, Reels and Shorts put their buttons. The same storyboard always produces exactly the same frames.

5. Music, sound effects and mixing

A music bed and sound effects (taps, dings, whooshes, pops) are generated for the reel, lowered under the voice and mastered to -14 LUFS, the loudness short-video apps aim for.

6. Encoding and delivery

The scenes are joined and encoded as an H.264 MP4 with AAC audio. You get the video, an SRT caption file, a cover image and a suggested post caption with hashtags. Downloads stay available for 30 days.

How long it takes

Drawing, voicing and encoding take about 28 seconds per 30 seconds of video on our render machine, plus a few seconds for the storyboard. When the queue is empty, most reels are ready in one to three minutes. The progress page shows each step and your place in the queue.

What you control

The script, up to five images, one character image and an optional instruction of up to 300 characters, such as "make it funny", "keep it under 30 seconds", "use a calm male voice" or "end with: follow for more". Instructions steer the choices the AI makes; they can't break the content rules or the 10 to 90 second length. There are no other settings to tune.

Try it on a real example

Every reel here was made by the app from the script shown on its page. No hand edits.

Read next

Paste your script and see it animated

3 free reels · no card · free reels carry a small watermark