Problem
16:9 interview footage is auto-reframed to 9:16 using face detection, but the crop follows the interviewer during a reaction shot and leaves the actual speaker partially outside frame.
Solution
Root Cause / Diagnostic:
Automated reframe algorithms rely on generic computer vision tracking bounding boxes around dominant facial contours rather than analyzing multichannel audio tracks. When an interviewer nods or gestures emphatically during a guest's dialogue, the motion tracking priority shifts to the moving non-speaker, cutting the actual speaker out of the vertical frame. Automated visual reframe without audio-gated speaker prioritization consistently fails during conversational turn-taking.
Actionable Step-by-Step Fix:
1. Disable Automated Motion Tracking on Two-Shots: Switch the Auto Reframe motion preset to "Manual / Point Tracker" or nest individual speakers onto dedicated video tracks before reframing.
2. Split Video Cuts at Audio Turn Transitions: Blade the video track precisely where conversational turns occur, locking crop coordinates independently to the active audio source.
3. Verify Headroom and Center Anchors: Apply a 9:16 safe margin overlay in the monitor panel and manually set horizontal pan offsets to hold the speaker centered during dialogue.
Pro Creator Tip:
Separate audio dialogue tracks into distinct channels and use waveform silence detection to blade video cuts before applying vertical crops, ensuring visual cuts precisely track spoken dialogue.
Automated reframe algorithms rely on generic computer vision tracking bounding boxes around dominant facial contours rather than analyzing multichannel audio tracks. When an interviewer nods or gestures emphatically during a guest's dialogue, the motion tracking priority shifts to the moving non-speaker, cutting the actual speaker out of the vertical frame. Automated visual reframe without audio-gated speaker prioritization consistently fails during conversational turn-taking.
Actionable Step-by-Step Fix:
1. Disable Automated Motion Tracking on Two-Shots: Switch the Auto Reframe motion preset to "Manual / Point Tracker" or nest individual speakers onto dedicated video tracks before reframing.
2. Split Video Cuts at Audio Turn Transitions: Blade the video track precisely where conversational turns occur, locking crop coordinates independently to the active audio source.
3. Verify Headroom and Center Anchors: Apply a 9:16 safe margin overlay in the monitor panel and manually set horizontal pan offsets to hold the speaker centered during dialogue.
Pro Creator Tip:
Separate audio dialogue tracks into distinct channels and use waveform silence detection to blade video cuts before applying vertical crops, ensuring visual cuts precisely track spoken dialogue.