Vertical crops that don't decapitate your facecam

Scroll any gaming hashtag on TikTok and you will find them: clips where the crop caught half a forehead, a chin floating at the top edge, or a facecam sliced clean out of existence. The gameplay is great. The clip is unwatchable. This post is about why naive vertical cropping fails, and how CLIPFARM's one-time facecam calibration fixes it for every clip you export.

The geometry problem nobody escapes

A stream goes out at 16:9. Short-form platforms want 9:16. Those two shapes barely overlap: a center-crop of a 16:9 frame to vertical throws away roughly two-thirds of the image. Whatever sat in the left or right third of your layout — which is where most streamers park their facecam — is gone, or worse, half-gone. That is where the decapitations come from: a crop window that doesn't know where your face is.

Why "smart" auto-detection isn't enough

The obvious fix is face detection: find the face, crop around it. It sort of works, until it doesn't. Face detection loses you when you lean out of frame mid-rage, when the game renders a character's face larger than yours, or when your cam has a border the detector mistakes for content. And a crop that follows a detected face drifts — the frame swims around while the viewer is trying to watch a clutch. Jittery framing reads as amateur even when the moment is a 98.

The boring fix that works: calibrate once

Your facecam does not move. It sits in the same corner of your layout every stream, because you put it there. CLIPFARM leans on that: you calibrate your cam region a single time — draw the box around where your camera lives in your stream layout — and the farm reuses that calibration on every clip from every VOD.

The output uses the layout short-form viewers already expect:

• Your facecam on top — full, stable, never cropped mid-face. The reaction is half the clip.
• The gameplay below — the action framed under your reaction, the way every successful gaming short is built.
• Rock-steady framing — nothing tracks, nothing swims. The crop was decided by geometry, not by a detector guessing frame-by-frame.

Change your overlay? Recalibrate in a few seconds and every future clip follows the new layout.

Why the reaction layout wins on short-form

There is a reason the cam-on-top format dominates gaming TikTok: the face is the story. Viewers who have never seen your stream don't yet care about the play — they care about the person reacting to it. A clip that preserves your full reaction converts scrollers into viewers; a clip with a cropped forehead converts them into scrolls. This is also why the farm's hype scoring weighs crowd noise: the moments worth clipping are the ones where the reaction happened.

Where this sits in the pipeline

Facecam-safe cropping is the render step of the same pass that detects and scores your moments — every clip in your dashboard comes out 9:16, captioned, scored and framed correctly, ready to download in full quality for 30 days. See the whole pipeline on the features page, the per-game detectors for Souls and COD, or just run your last VOD through the free trial — two VOD hours, and calibration takes less time than exporting a single manual crop.