Punch-In

You have a moment from a 1920×1080 video and you need it as a 1080×1920 post. Two options: letterbox the wide frame with bars, or punch in — crop a tall slice from the middle and scale it up. Almost everyone punches in, because a letterboxed clip announces itself as a repost before the audio starts, and viewers on TikTok and Reels have learned to read the bars that way. A center punch-in on 1080p applies about 1.78× digital zoom. On 4K source the overhead is far lower and the output noticeably cleaner, which is a real and underrated advantage of picking sources that publish in high resolution.

Punch-ins do double duty as uniquification. Changing the crop changes the pixel hash of every frame, which weakens visual fingerprint matching. That is genuine protection against false-positive claims on material that qualifies as commentary, or where you have a licensing arrangement — but it is not the whole job. Audio fingerprints do not care about your crop; the waveform is identical to the source. On channels with active Content ID enrollment, pair the crop with a small audio change — a pitch shift of ±2–3 cents, or speed in the 0.97×–1.03× range — or expect the claims to keep landing.

Two modes cover most work. A static crop locks the frame position for the whole clip: fast, predictable, and correct for a seated speaker who does not move. A tracking crop follows the subject, which is what you want when someone paces, gestures wide, or walks across the shot. Tracking has a failure mode worth knowing about — if it loses the subject during a fast edit, the crop snaps, and a snap is more distracting than a slightly off-center static frame. If you are hand-editing 30 clips a day, static crops with a quick review pass will outlast tracking every clip individually. If the tool does the tracking for you, that tradeoff stops mattering: AutoClip picks the layout per clip and keeps the speaker centered, so movement-heavy footage does not have to be triaged by hand.