But… the example given shows that they place the video frame behind the window an make the front window transparent except for the round play button. This apparently offloads the frame… so why not just do the same for rounded corners?
The selfie example has an explanation under it. Summary: because the video subsurface is square and needs to be clipped to produce rounded corners. Otherwise you would have corners of the video surface protruding from under the corners.
What am I missing?