Multi-View Video Preloading via MPD Viewpoint Ordering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for transmitting multi-view video in virtual reality technologies face high bandwidth consumption due to the need to download and buffer multiple viewpoints, leading to increased system load and reduced client Quality of Experience (QoE).
Innovation Solution
A system and method that records distance and direction information between viewpoints in a Media Presentation Description (MPD) for transmission, allowing for preloading of the video with the closest viewpoint information, reducing system load and enhancing QoE by allocating spare bandwidth for efficient video playback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all viewpoint information is downloaded and buffered to ensure complete multi-view video playback, then video playback reliability is improved, but bandwidth consumption increases significantly
Solution Approach 1:
The system performs preliminary action by downloading and buffering only the nearest viewpoint video before actual playback occurs. The viewpoint ordering information is used to predict which viewpoint will be played next, and only that specific viewpoint is preloaded into the buffer, rather than buffering all viewpoints in advance.
Solution Approach 2:
The system extracts only the essential viewpoint information needed for playback by using viewpoint ordering information to identify and download only the nearest viewpoint video, separating the necessary content from the complete multi-view dataset, thereby reducing unnecessary bandwidth consumption.
2Reliability
If multiple viewpoint videos are buffered to enable smooth viewpoint transitions, then playback quality is improved, but system load increases
Solution Approach 1:
The system applies local quality by focusing buffering resources on only the nearest viewpoint based on viewpoint ordering information, rather than uniformly buffering all viewpoints. This localized approach ensures high playback quality for the current and upcoming viewpoints while reducing overall system load.
Solution Approach 2:
The system performs preliminary buffering of only the nearest viewpoint video before playback, using viewpoint ordering information to predict future playback needs. This preliminary action ensures smooth transitions without requiring the system to maintain multiple viewpoint videos in buffer simultaneously.
3Loss of energy
If viewpoint ordering information with distance and direction data is added to MPD to enable efficient preloading, then bandwidth efficiency is improved, but data structure complexity increases
Solution Approach 1:
The viewpoint ordering information is integrated into the existing MPD structure, making the MPD serve multiple functions: it still provides media presentation description while now also containing viewpoint ordering information for efficient preloading. This multi-functionality approach avoids creating separate data structures.
Solution Approach 2:
The system adds only the necessary viewpoint ordering information (distance and direction data) to the MPD without including all possible viewpoint metadata. This partial action approach provides sufficient information for efficient preloading while minimizing the increase in data structure complexity.
Data Source
AI summary
The present invention relates to a system and a method of preloading a multi-view video. According to an embodiment of the present invention, load of a system is reduced and client's Quality of Experience (QoE) is maximized. In addition, the sense of direction, distance, and space for virtual reality are felt in the same manner as those for the real environment, so that a virtual reality service is provided realistically, thereby further improving immersion and interest in the virtual reality service.