Clip Studio
A long video in.
Clips out, in your name.
Paste a link or drop a file. The studio finds the passages that stand on their own, cuts them vertical, captions them word by word and puts your logo, name and handle on every one.
Included with a clipper or brand account. Up to 3 jobs at once, 8 a day.
Your page
@yourpage
The line that stops the scroll
clips that STAND alone
Your page
@yourpage
clips that STAND alone
How it works
001
Paste, upload, or pick a campaign
A YouTube link, an X or Reddit post, a Drive or Dropbox share, a direct file, or an upload up to 2 GB. From 20 seconds to 3 hours. On a live clipping campaign, the brand's own assets are one click away.
002
It finds the passages that stand alone
The sound is fetched first and transcribed once, word by word with times. A model reads the transcript for passages someone can hear cold, on mute, halfway down a feed: a claim that contradicts what people assume, a number that lands, a story with a turn in it. Each runs 15 to 75 seconds and starts on a sentence.
003
Clips come back in your branding
Only the seconds each passage needs are fetched as pictures. Every clip is rendered at 1080 by 1920 with your lockup, captions timed to the words with one word lit in your colour, and a hook line above the picture if you want one. Download and post.
Your kit
Logo, name, handle, one colour.
That is the whole kit. The lockup goes top-left the way the big theme pages wear it. One word in every caption lights up in your colour: a number first, then a name, then the word the line is about. Two layouts: the picture at full width on black with nothing covering it, or filling the frame, cropped to follow the speaker.
Run several pages? Make a kit for each and pick one per job.
Your page
@yourpage
the number was 40%
Studio Sample
@studio.sample
nobody tells you this
Honest about the picking
It finds what stands alone. It does not predict a hit.
The passages come back in the order they fall in the video, not ranked. A model reading a transcript is good at one thing here: finding a stretch that makes sense cold, without the rest of the video around it. Which of those travels depends on the page, the hook, the sound and the day, and nobody reading words alone can call that. So the studio gives you the candidates, and you choose.
Questions
- What does the studio take in?
- A link yt-dlp can read (YouTube, X, Reddit, Vimeo, a Google Drive or Dropbox share, a direct video file) or an upload up to 2 GB, from 20 seconds to 3 hours long. A clipper can also point it at a live campaign's own assets.
- How are the passages chosen?
- A model reads the timed transcript and returns the passages that stand alone: heard cold, on mute with captions, halfway down a feed. When no model key is configured the worker falls back to a rule of thumb (whole sentences, a number or a contrast word up front) and the job says so. The order is where they fall in the video, not a prediction of which will travel.
- What comes back?
- Up to twelve 1080 by 1920 MP4 clips of 15 to 75 seconds each, with word-timed captions (one word lit in your colour), your lockup top-left and, if you want it, a hook line above the picture. Download them and post them on your own accounts.
- What does it cost?
- Nothing on top of your account. Each account runs up to 3 jobs at once and 8 in a day.
Cut your first clips.
Make a kit, paste a link, and come back to a row of clips with your name on them.
Register and open the studio