v1.50 – v1.99

v1.72.0 (2026-10-13)

Fixed

• [VOD] Improved character identification and reduced false identifications from photographs
• [VOD] Improved translations of short descriptive fields and prevented recoverable translation failures from omitting an output language
• [LIVE] Prevented some live analyses from failing before media arrives
• [VOD] Reduced unnecessary scene splits within continuous conversations and montages connected by music
• [VOD] Prevented missing or reordered character-name answers from assigning names to the wrong characters
• [VOD] Improved subject appearance times in shot details for short clips
• [VOD] Some valid videos that previously failed analysis now complete successfully
• [VOD] Reduced repeated subject entries across motion and camera cuts in shot details while retaining distinct people and objects
• [VOD] Prevented false scene splits at fades to black when audio continues without a gap
• [VOD] Improved scene boundaries near the beginning of videos and prevented some audio-processing failures from stopping scene analysis
• [VOD] Improved tobacco detection for small or distant smoking depictions surrounded by visual effects