Video Processing
Run AI detection and tracking to generate MP4E metadata for your video.
Overview
Processing is the step that analyzes your uploaded video to produce the object registry + tracking timeline used by overlays, rules, and interactions.
Processing is server-side
Studio starts processing via an API call and then listens to a live progress stream while the backend detects and tracks objects.
Start Processing
Open a project, then run processing from the editor UI (the project panel provides controls to start processing). When processing starts, Studio opens a modal with real-time progress.
Live Progress
The processing modal shows:
- Frame progress and percent complete
- Objects found so far
- Processing FPS, elapsed time, and estimated remaining time
- Live detection preview (bounding boxes on a canvas)
Processing Options
Studio supports optional processing parameters to focus detection and improve results:
Processing request body (optional)
1// Studio calls POST /api/video/:id/process with an optional JSON body:2{3 "filter_terms": ["shoe", "bag"], // optional: focus detection on specific concepts4 "tracking_quality": "high", // optional: e.g. "standard" (default) / higher-quality presets5 "save_polygons": true // optional: store per-frame polygons (useful for replacement zones)6}When to enable save_polygons
If you plan to use replacement zones or need high-fidelity shapes (not just bounding boxes), enable polygon saving so the engine/player has richer per-frame geometry.
Reprocessing
Reprocessing regenerates detection/tracking and may overwrite parts of the metadata. Studio prompts for confirmation when reprocessing an already-processed project.
Pro workflow
Export Metadata JSON before a big reprocess so you can restore edits or selectively merge changes after.