The face is tiny
On a 1080p stream the facecam is often a tenth of the frame. Cropped to vertical and blown up, it is either cut off or the size of a stamp, and the reaction is the reason anybody watches.
A stream is wide, with the streamer in a small box in one corner. A plain vertical crop either keeps the game and loses the face, or zooms onto the face and loses the game. The layout that works on TikTok stacks them: the camera in the top third, the gameplay filling the rest.
Three things go wrong when a stream is cropped like any other video.
On a 1080p stream the facecam is often a tenth of the frame. Cropped to vertical and blown up, it is either cut off or the size of a stamp, and the reaction is the reason anybody watches.
A crop that follows the streamer drifts toward the corner and cuts the action in half. Viewers see a sliver of the screen and a lot of background.
Plenty of streams switch layouts, or cut between two streamers' cameras by who is talking. A layout set once for the whole clip is wrong half the time.
In an editor like CapCut it is four steps. SnapDub does the same thing on its own.
Not every moment wants it. If the clip is the streamer talking straight to camera, a full-screen face is stronger than a split. If the clip is a play with no reaction worth seeing, the game should have the whole frame. The split is for the moments where the reaction and the gameplay are both the point.
Coming from Twitch? See Twitch to TikTok. And for how the crop decides who to follow in the first place, here is the test.
Yes, CapCut on a phone has layers and masks, but it is slow work on a small screen, and every scene change in the stream means another fix.
Then there is nothing to lift, and the clip should simply follow the action. SnapDub only uses the split when it actually finds a camera on screen.
Yes. When a video cuts between two people's cameras, each stretch gets its own layout with the right camera on top.
Nobody can promise that. It is the layout viewers already know from the biggest stream clippers, and it keeps both the reaction and the game readable on a phone.
SnapDub finds the moments, finds the facecam, and builds the layout for every clip.