Dotredlabs

BaddyAI Editor is a video-editing prototype for turning badminton match footage into player-focused, shareable highlights. Wide court recordings capture the whole game, but making a compelling clip requires selecting a moment, keeping the right player in view and adding visual emphasis. Our goal is to make that editing workflow accessible to everyday players and creators. The prototype integrates Meta SAM 3.1 through the Meta Model API to generate per-frame segmentation masks from a selected video clip. A browser-based editor uses those masks for player spotlights, glowing outlines, bystander blur and tracked player close-ups. Users can review and adjust player identities, choose among three overlay styles, preview the composition and export the edited video. A synchronized match overlay can display scores and shot statistics from supplied event data. The included sample data is illustrative; these statistics are not inferred by the segmentation model. The project focuses on the creative editing workflow: combining subject segmentation with editable framing and visual effects so players can tell the story of a rally in their own style.