Video Summarizer & Storyboard Extractor
Audio, Video & Media Understanding
Sample keyframes across long video feeds and assemble a structured summary storyboard.
In strict compliance with MultimediaToolKit engineering standards, this module does not fabricate fake progress bars, synthetic mock data, or pretend to execute cloud pipelines. The production specifications below detail the exact target architecture scheduled for worker container deployments.
Blueprint Role & Lifecycle
Media Understanding (Slide 7)
Directly mapped from the authoritative PPT specification. Formats and protocols conform to the unified pipeline: IDEA → UNDERSTAND → CREATE → POLISH → DELIVER.
Worker & Processing Engine
SceneDetect + Keyframe Extractor + Vision LLM
Target execution environment: Worker Background Runtime. Designed to run as non-blocking background tasks or standalone Redis queue workers.
Target REST API Route
/api/v1/video/summarizeRegistered in the FastAPI v1 specification. Asynchronous task state is stored with standard status transitions: queued → running → succeeded / failed.
Supported Formats & Payloads
Storage handled by the SHA-256 StorageService with path traversal sanitization and 100MB upload limits.
Currently Operational Tools Ready to Execute
The following core modules are fully operational today with genuine test coverage:
Media Extractor
Native FFmpeg video-to-audio MP3 extraction & download.
Storyboard Studio
Structured multi-scene directorial generation with JSON export.
Projects & Assets
Workspace management, asset storage and file persistence.
Job Monitor
Live task tracking, progress monitoring, and output downloads.