16:9 to 9:16 Reframing
keeping the shot alive in vertical

The cheapest fix is upstream. Recording in 4K gives you a 1216×2160 crop window with no upscaling, which means vertical deliverables that look native rather than soft. Framing subjects centrally with headroom, and keeping graphics and product out of the outer thirds, makes reframing nearly automatic.

If you cannot control capture, prioritise 4K on the primary camera only. It solves most of the problem for the shots that carry the video.

First, stacked split-screen: both speakers in vertically stacked windows. It works for interviews and podcasts and keeps reaction visible. Second, alternating punch-ins driven by who is speaking, which reads more like a scripted edit and needs clean audio to trigger cleanly. Third, a wide letterboxed centre band with captions above and below — acceptable for a beat or two, weak as a default.

What to avoid is a static centre crop on a two-shot. You get two half-faces and a viewer who leaves in the first second.

Automatic reframing tools track a subject reasonably well in single-speaker shots and badly everywhere else. Use them for the first pass, then scrub for three specific failures: the pan that starts before the subject moves, the jitter when two faces are in frame, and the moment the tracker locks onto a background person.

Manual keyframes on the ten to fifteen percent of shots that fail is faster than reframing everything by hand and produces a materially better result than trusting the tool.