For the vast majority clips, Split View will be the optimal choice for a reframe algorithm. This process will add approximately 5 minutes onto your workflow run time.
Split View:
Dynamically determines who is speaking, tracks their face and body, and can handle multiple simultaneous speakers
Centered:
Will take a horizontal video and place it in a vertical frame with padding. You may customize the position and the amount of zoom.
Face Tracking: Deprecated. This model uses face detection with no active speaker detection. Good for single speakers but still recommended to use Split View.
Custom:
For enterprise customers who will have custom styling. Defaults to a “Streamer” template for Twitch streams, centering faces on top of the main content.