US20260261747
2026-09-03
Electricity
H04N21/8541
A novel system and method facilitate the automated selection of video streams capturing the same real-world event from multiple viewpoints. The system evaluates each video stream's contribution to the narrative and penalizes redundancy. By calculating a composite arbitration score for each stream, it selects the most contextually relevant stream for output. Transition timing between streams is controlled by predicting changes in the event's narrative state.
The disclosure pertains to automated video production systems, focusing on dynamic selection among simultaneous video streams. It introduces temporal narrative modeling and contextual comparison to choose an output stream, enhancing traditional systems that rely solely on independent stream quality metrics.
Existing systems often independently rank video streams based on quality or event detection, which can result in redundant outputs when streams show similar content. These systems typically react to current stream characteristics without considering the event's temporal progression, potentially disrupting narrative continuity. This invention addresses these limitations by integrating narrative modeling and contextual analysis.
The system improves video stream selection by modeling the narrative state of the event and comparing contextual contributions of each stream. A computing platform assesses each stream's marginal contribution and applies a redundancy penalty for similarity with the current output stream. The system dynamically adjusts its logic based on narrative phases, ensuring the output stream maintains narrative continuity while avoiding redundancy.
The system comprises modules for narrative state determination, marginal contribution computation, redundancy suppression, and transition control. It handles asynchronous streams from various devices, allowing real-time or post-capture arbitration. The process ensures editorial decisions consider inter-stream comparisons and evolving narrative states, rather than isolated stream evaluations.