A vertical crop that looks clear at the beginning can cut off the subject halfway through an AI zoom. Review the caption area and whichever part of the frame carries the action. Use more than the start or end still to choose the crop.
Locate the most constrained part of the move
For a short portrait of a cook, the start includes the face and a mixing bowl. A later push-in reveals hands at the edge. Place the source in the Compose editor and use the intended vertical framing where supported. Mark where the hands approach the frame edge and where captions appear. Keep the most important action visible in both the full clip and any shorter cut.
Try a stable reframing first. Constant adjustment can create a second movement that fights the original push. If the shot requires a major shift in subject emphasis, consider cutting to a different scale.
Test the final file in a phone preview
Watch at full speed and pause briefly at the point marked earlier. Check the face, hands and caption contrast. If the bowl leaves frame during a spoken instruction, insert an accurate detail shot or use a different crop for that beat.
Export one short test for each destination format. A caption safe in one interface can still overlap controls in a different placement.
Check whether the edit uses on-screen platform text or baked caption graphics. They occupy different positions once the file is posted. Leave the action in a region that can remain readable if the social team decides to change the caption styling.

