Detection, audio and scene analysis across every frame of Harry Potter and the Prisoner of Azkaban — derived from object recognition, transcript alignment and shot segmentation.
All three series normalized to 0–100. Hover to scrub the runtime.