Multi-Face Video Handling: Priority Selection and Overrides

When a scene contains more than one face, Bitcut needs to decide which one to follow. It uses a priority system to pick the most likely subject, and locks onto that face for the duration of the scene. You can override this choice if the algorithm picks the wrong person.

How Face Priority Works

For each scene, Bitcut scores every detected face based on several factors:

  • Face size — larger faces (closer to camera) get higher priority, since they're usually the main subject
  • Position — faces near the center of the frame score higher than those at the edges
  • Consistency — faces visible for more of the scene are preferred over those that briefly appear
  • Edge penalty — faces partially cut off at frame edges get lower priority

The face with the highest combined score becomes the tracking target for that scene. When a scene change occurs, priority is recalculated for the new scene — so different scenes can follow different people.

One face per scene. Within a single scene (between camera cuts), Bitcut locks onto one face and ignores the others. This prevents the camera from bouncing between speakers, which would look distracting. The lock only breaks at scene changes.

Overriding the Selection

If Bitcut follows the wrong person, you can manually adjust the crop to center on the face you want. Drag the preview to reposition the vertical frame onto the correct subject. Bitcut creates a keyframe at that position, effectively overriding the automatic selection for that moment.

For a persistent override across an entire scene, add a keyframe near the start and another near the end of the scene, both centered on your preferred subject.

Common Scenarios

  • Interview (two people, alternating) — if the camera stays on a two-shot, Bitcut picks the larger face. Override manually to follow the speaker you want.
  • Panel discussion — scene changes (camera switching between panelists) let Bitcut naturally re-select the active person for each shot.
  • Presenter with audience — the presenter is typically larger and more central, so they get priority automatically.
  • Crowd or group shots — with many small faces, tracking is less precise. Consider a static center crop for these sections.
Split long two-shots. If you have a long two-person shot and want to switch between them, trim the clip at the points where you want to change the tracked face. Each clip segment gets its own face tracking, letting you follow a different person in each.