No, and it's the thing we care most about. Every frame is one you shot. Nothing is invented, no faces are made up. The AI does the legwork: reading the audio, separating people from backgrounds, timing words to speech. If your footage isn't real, the whole thing is pointless.
The film is live in beta: many clips in, one cut, graded, captioned film out, steered by chat. In the studio, looks, captions and captions behind the subject are live and free; templates like background replace preview on any video and download with Creator. Script editing and wardrobe change are being built.
You drop in up to 20 clips and, ideally, a song. Every clip gets watched and listened to, the song's beats get measured, and it writes a short story from the day: which moments, in what order, each cut landing on the beat, with the drop saved for the best shot. Then it grades the film with one of the real Looks and renders it, on your device when your hardware can. If the first cut isn't right, say so in plain words and it recuts.
Up to 20 clips and 10 minutes of footage per film, and a single clip can use the whole 10 minutes; out comes a short film, usually one to two minutes. Films are in beta and not part of the paid plan yet; limits may change while we finish them.
A complete finish, not a filter over the top: its own colour grade, halation, film grain and gate weave, tuned to belong to each other. There are seven Essentials, twenty more recipes, and a library of about 120 film stocks behind them: 140-odd in total, all free. Three dials are exposed if you want to push one further: halation, gate weave, and grain strength, size and colour.
The subject is separated from the background when your clip loads. Flip one switch and captions render behind them: text vanishes behind a shoulder and reappears on the other side, the way it would if someone had rotoscoped it by hand. No masks to draw, no keyframes to set.
It runs on ElevenLabs Scribe with word-level timestamps, and most clips need no edits at all. When something's wrong, click the word and fix it: text, timing and styling all live in the same view.
75+ with automatic detection, including code-switching like Hinglish and Punglish. Many can be captioned in native script or romanised, Hindi, Punjabi, Arabic, Urdu, Bengali and Tamil among them, and you switch between them in the transcript.
MP4, MOV and WebM, up to 4 GB per file. A film takes up to 20 clips, 10 minutes of footage in total, plus songs as MP3, M4A, AAC or WAV. A single studio clip can be up to 2 minutes. Export anywhere from 720p to 4K with no forced upscale. WebM uploads don't work in Safari; use MP4 or MOV there.
Not to upload a clip and edit captions. You'll need one to transcribe, export or make a film, so your quota and saved exports stay with you.
Free exports carry a small badge in the corner. Creator removes it. Nothing else about the finish or the quality changes between plans.
It's coming. For now the studio is built for desktop: picking a look works fine on a phone, but tuning individual words needs more precision than a phone screen gives you.
Something else? Ask us from the app, or just open the studio and start free.