pyflink.multimodal.codec.extract_video_frames

extract_video_frames(video_input, bridge_client: pyflink.fn_execution.file_system_bridge.FlinkFileSystemBridgeClient, *, frame_selector: str = 'keyframe', sample_interval_ms: Optional[int] = None, max_frames: Optional[int] = None, container_options: Optional[dict] = None, read_chunk_size: Optional[int] = None, max_cached_blocks: Optional[int] = None, read_ahead_blocks: Optional[int] = None)

Decode selected frames from a full video or metadata-only video segment.

video_input accepts a URI string, VideoRef, or a Row/tuple matching VIDEO_STRUCT_TABLE_TYPE. Segment handling is metadata-only: the same URI is opened and frames outside start_time_ms / end_time_ms are skipped.