Automatic multi-camera remix from single video

In this paper we present a first of its kind automatic multi-camera video remix creation system from a single video, referred to as SmartView. We present a novel method which uses the fusion of multimodal content analysis and cinematic rules, for creating a multi-camera experience. Further, a playback metadata based model, which consists of playback instructions for metadata-aware media player, allows for providing a remix experience without editing the original video content. This approach produces a low footprint, which is suitable for on-device processing in resource constrained mobile devices. The research prototype demonstrates the feasibility of such a system with current off-the-shelf mobile devices. The SmartView creation process was seen to take less time than the video duration. 5 out of 9 test users found the fully automatic SmartView remix experience to be better than the conventional playback. The user customized SmartView remix was preferred over conventional playback.