A wide video, made vertical. The face stays in.
Drop a 16:9 video. The faces are found, a calm 9:16 window follows them, and you get a vertical MP4 for TikTok, Reels and Shorts.
Runs in your browser. Nothing is uploaded, no signup.
Watch an example
13 seconds · a made-up example
The studio's Clips mode turns your own long recordings into vertical shorts, captions and brand included.Join the beta
How it works
Short and checkable: every limit links to the platform's own page, read on 28 Oct 2026.
Faces, five times a second
Each sampled frame is cut into overlapping squares (left, middle, right) and searched by Google's MediaPipe face detector, so faces that fill only a small part of a wide frame are still found. The detector runs in your browser.
One face, chosen with care
The window follows the biggest face, weighted by how sure the detector is. Between two faces of similar size it stays on the one it was already on, so it doesn't swing between speakers. One-off wrong detections are dropped by a median over five samples.
A calm camera operator
The window holds still while the face moves inside its middle half, then glides after it at most 0.6 window widths a second. When the face jumps far and stays there (a cut), the window cuts too instead of panning. With no face it holds where it was.
Made at the source's quality
Every frame is redrawn from the window and encoded as H.264 at 1080 × 1920 when the source is 4K, else at the window's own size (608 × 1080 for a 1080p video), with the sound kept.
What it doesn't do
- It follows faces: a product shot, a landscape or a person seen from the back keeps the last position.
- One face at a time: two people talking side by side get the bigger one.
- The frames are re-encoded by your browser; a long 4K video can take a few minutes.
- Nothing is sent or saved. The video stays in this tab.
Questions
How do I turn a horizontal video into a vertical one without losing the person?
Drop it here and press Find the faces: five frames a second are searched, a 9:16 window follows the face like a calm camera operator, and you get a vertical MP4. A plain centre crop cuts people off as soon as they move.
Does it follow several people?
It follows one face at a time: the biggest, and between two of similar size the one it was already on, so it doesn't jump from one speaker to the other at every word.
Is my video uploaded?
No. The face finder (Google's MediaPipe, about 14 MB, downloaded once) runs in your browser, and the new video is encoded there too.
Sources
- 1Google AI Edge — MediaPipe Face Detector (BlazeFace short range, Apache-2.0)the face finder of the auto-reframe, run in the browser · read 28 Oct 2026
More free tools
All 25 tools- Make the videoShort video analyzerWhy does this short work?
- Make the videoScript timerHow long will this script run?
- Make the videoAuto subtitlesNo subtitles yet? Make them here.
- Make the videoVideo to textNeed the words of a video as text?
- Make the videoBurn subtitlesSubtitles that stay on TikTok and Reels?
- Make the videoSubtitle checkerCan people read your subtitles in time?
Long recordings, cut into vertical shorts.
Vertiscale's Clips mode finds the moments of your own long recordings and makes each one a vertical short, with captions and your brand.

