Footage in. A finished video out.
Drop in your footage, choose a length and a shape, and get back a finished video: cut, captioned, framed, graded and animated, with no timeline and no manual steps.
- Length
- 15 seconds to 1 hour
- Shape
- 9:16 · 1:1 · 16:9
- Speech
- Mandarin, English, and the two mixed
- Captions
- Word by word; Chinese in Taiwan Traditional characters
- Runs on
- macOS on Apple silicon, Windows, Linux
An example edit: 60 seconds at 9:16, made by Fineframe 0.2.0 with no manual steps
Footage: “Why NASA and IBM made a new AI model that predicts the Sun’s damaging flares” by IBM Research, licensed under CC BY 3.0, shortened, reframed and graded, with its sound processed and music, captions, name cards and keyword highlights added. Music: “3 am West End” by statusq, CC0 1.0. Neither IBM Research nor the people in it endorse Fineframe.
How an edit is made
Analyse
Finds the shots, where people speak and pause, every word with its timing, the faces and who is speaking, and sounds such as laughter and applause.
Plan
Claude picks the best takes, removes fillers and dead air, and fits the edit to the length you asked for.
Render
Cuts, framing on whoever speaks, colour, voice, music, animated captions, motion graphics and loudness.
Check
Claude reviews every cut and fixes what is wrong, in up to three passes.
What it does
Captions, word by word
Burned in and lit word by word in one of five styles, with a subtitle file that matches the picture.
Framing that follows
A virtual camera frames every clip in the edit’s shape and follows whoever is speaking.
Motion graphics
Once the cut is known, Claude writes the edit’s labels, lower thirds and full-frame graphics, each checked against the speech.
B-roll
Cutaways to the web pages the speaker names and, with a free Pexels or Pixabay key, to stock footage.
Studio voice
Each source’s speech is denoised and processed by a chain chosen from its own measurements.
Fillers and dead air
Hesitations and stutters go by default; the aggressive setting removes verbal tics too.
Several shorts
Up to 10 shorts from one long video, each with a title, a description and hashtags.
Links as footage
Paste a public video’s link, from YouTube or another site yt-dlp reads, and it is imported at up to 1080p.
Brand kits
Your logo as a watermark, up to three colours, your fonts and your own caption style, on any edit.
Music
Tracks you supply, fitted to the edit and lowered under speech.
Change the edit
Delete words from the transcript, fix a caption, change the look or the captions, or tell Claude what you want and plan it again.
In your language
The app is in English and Traditional Chinese, following your system’s language.
Your footage stays on your computer
Analysis and rendering run on your computer, and no video file is uploaded. Claude plans and checks the edit on your own Anthropic key: it receives the transcript and details of each file, and pictures of the frames around each cut.
Speech is heard on your computer, or on OpenRouter when you choose it or your computer has too little memory. Everything the app sends, and where.
Your key, your bill
Fineframe uses your own Anthropic API key, and Anthropic bills you directly: about US$0.20 to 0.60 for an edit of a minute or less, and US$1 to 1.20 for one of five minutes. The app estimates the cost before each edit and shows it after.
The app is free during its public beta. When paid plans open, it is a subscription: US$15 a month or US$120 a year. See pricing.