Twitch clips to TikTok vertical without facecam cropping

Twitch footage is 16:9 and TikTok is 9:16, and the overlap between those two rectangles is about a third of your frame. Converting a clip well means deciding, once, what survives: your face, the action, and the words. Here are the three ways streamers do it, why two of them fail, and the layout rules that keep the platform's own UI from eating your work.

Start from the VOD, not the clip page

Before the how, a word on the what. Working from the full VOD instead of a 60-second Twitch clip (Twitch Help) means every moment of the stream is available, not just the ones somebody pressed a button on, and you control the in and out points instead of inheriting a stranger's guess. It also means the conversion runs on full-quality source. If the VOD is your raw material anyway, you may as well let the farm find the moments too; that pipeline is in how to auto-clip Twitch VODs.

Why not just upload it horizontal

The lazy alternative is uploading the 16:9 clip as-is and letting TikTok letterbox it. The result has dead bars above and below, tiny gameplay, captions the size of a legal disclaimer, and it fills maybe a third of the screen the viewer paid full attention for. Every platform treats letterboxed uploads as low-effort, and the viewer's thumb agrees. Vertical is not a style choice; it is the admission price.

Option 1: center crop, the default failure

Cutting a vertical slice out of the middle of the frame is what every quick converter does, and it fails in a predictable way: streamers park their facecam in a corner, exactly where the slice is not. The result is gameplay with no human in it, or worse, half a human, the floating chins and cropped foreheads all over every gaming hashtag. A center crop does not know where your face is, and it cannot.

Option 2: face-tracking crops, smart in theory

The obvious fix is detecting your face each frame and cropping around it. In practice the detection loses you when you lean out mid-rage, locks onto a game character's face during a cutscene, and the crop drifts frame to frame, so the whole shot swims while the viewer is trying to watch a clutch. Jittery framing reads as amateur even when the moment is incredible. Your camera does not move during a stream, so the crop should not move either.

Option 3: stacked layout, cam on top

The format that dominates gaming short-form sidesteps the geometry instead of fighting it: take two crops of the same source, put your full facecam across the top of the vertical frame, fill the rest with the gameplay crop, and lock both. Nothing tracks, nothing drifts, and your reaction rides above the action for the whole clip. It also survives muted playback, because the viewer reads your face and the captions before ever unmuting. The OBS-side setup that makes this easy, cam size, placement, keeping overlays off it, is covered in vertical crops that don't decapitate your facecam.

Doing the conversion by hand

In CapCut or Premiere the stacked layout is a few minutes of work per clip: duplicate the layer, crop one copy to the cam and one to the action, position both, add captions, export at 1080x1920. Doable, and worth doing once to understand the format. The problem is the multiplication: 10 clips a stream, four streams a week, and the "few minutes" is suddenly the evening you were supposed to spend streaming.

Doing it automatically

Clip Farm's version is deliberately boring. You calibrate your cam region once, drawing a box around where the camera lives in your layout, and every clip from every VOD reuses that calibration: facecam full in frame on top, gameplay below, word-level captions placed in the safe zone, exported at 1080x1920. Change your overlay someday and recalibration takes seconds. What else ships with each clip is on the features page, and the full download-edit-post pipeline, manual and automated, is in the downloader-to-editor workflow.

One note on judging the output: trial exports come out at 720p with a watermark, so you can check the crop, the captions and the framing on your own footage before paying anything beyond the $1. Paid plans export clean 1080p. Judge the geometry on the trial, because that is the part that does not change between plans.

Safe zones: the part TikTok covers up

A clean crop can still get eaten by the platform. On a 1080x1920 frame:

• Bottom 15 percent: the caption, sound info and your username live there. Keep nothing important in it.
• Right edge: the like, comment and share rail sits on top of it. No text, no HUD elements that matter.
• Middle band: captions belong here, roughly the vertical center, where nothing overlaps them.

Shorts and Reels cover similar zones, so one safe layout serves all three platforms.

Captions are part of the conversion

Reframing the video is only half the job; the words need reframing too. A stream's audio assumes a viewer who chose to be there, so it rambles. Short-form captions fix that in two ways: word-level burned-in captions make the clip watchable muted, which is how most people will first see it, and the caption on the post itself becomes the hook that stops the thumb. Skip the burned-in kind and you have converted the shape but lost the audience.

Layout habits that pay off later

Whether you crop by hand or let a tool do it, three OBS habits make every future conversion easier: keep the cam rectangular with no alerts or chat boxes crossing it, because whatever overlaps the cam gets baked into the top of every clip; make the cam big enough to survive being scaled to the top of a 1080-wide frame, around a quarter of your stream width; and stop moving it between streams. A fixed layout means one crop setup works for every clip from every VOD, whether that setup is yours or the farm's calibration.

One file, three platforms

The same 1080x1920 file works everywhere short-form lives; upload natively to each platform rather than cross-linking, and never upload one platform's watermarked export to another, because watermarked reposts get suppressed. If you want the cost math on automating the whole conversion, it is VOD-hour pricing: a $1 trial for one hour on your own stream, Starter $13/mo for 10, on the pricing page.

Whatever route you pick, the acceptance test is the same: scroll your posted clip on a phone like a stranger. Full face visible, captions readable muted, nothing hiding under the platform UI. Pass all three and the crop did its job.

FAQ