9:16 without the guessing.
Never's reframe converts landscape recordings to vertical by measurement: it detects your face across sampled frames, finds the true hair line against the background, and locks a crop that keeps your head exactly where the format wants it — inside the safe zones, every frame, without a human nudging it.
How does the measured reframe work?
Sample frames across the scene, detect the face, then find where the hair actually ends — not the detector's box, which cuts foreheads, but the measured boundary against the background. From that, compute the one crop that places the head at the format's standard position, and hold it for the scene. Static, deliberate framing — the way a human operator frames an interview — rather than a camera that chases.
The measurement matters because detectors lie a little: face boxes run tight, hair and hats extend the true subject, and dark hair on a dark wall defeats simple edges. Measuring the median background and subtracting it finds the real boundary even then. When cutaways cover part of a scene, B-roll hides the crop change entirely.
Who needs this?
Everyone repurposing landscape recordings into clips: podcasts shot on wide cameras, webinars, screen-and-face recordings. The reframe is applied automatically during Harvest, and the same measurement drives where captions and graphics sit in the 9:16 frame.
Try it on your own footage — free
Read the transcript
9:16 without the guessing. Never's reframe converts landscape recordings to vertical by measurement: it detects your face across sampled frames, finds the true hair line against the background, and locks a crop that keeps your head exactly where the format wants it — inside the safe zones, every frame, without a human nudging it.
Frequently asked questions
Why isn't center-cropping enough for 9:16?
Because speakers aren't centered and don't stay still. A fixed center crop loses the subject the moment they lean; a naive face-follow jitters. Measuring the composition once per scene and locking it gets the stability of a human reframe with the accuracy of detection.
What are safe zones and why do they matter here?
Platform UI overlays the top ~200px and bottom ~300px of a vertical frame. A reframe that ignores them puts your eyes under the username or your captions under the like rail. Never composes inside the clear band — verify any frame with the safe-zone checker.
Does reframing lose quality?
A 9:16 crop keeps roughly a third of a 16:9 frame, so 4K sources reframe to full 1080×1920 sharpness while 1080p sources upscale slightly. Record in 4K if vertical clips are a primary output.