Predictive Bitrate Selection for 360-Degree Video Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 360-degree video streaming systems experience jarring resolution drops when the viewer rapidly changes their field of view, and important artistic details may be lost due to the use of low-resolution video streams, as they only load high-quality streams within the current viewport.
Innovation Solution
Implementing author-defined and predictive metadata to pre-fetch high-resolution video frames based on the viewer's likely future orientation, using vectors and probabilistic models to anticipate and prepare for viewport changes, allowing for smoother transitions and improved video quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If the client requests only high resolution video stream frames within the current field of view, then bandwidth usage is reduced and buffering requirements are minimized, but resolution drops occur when the viewer rapidly changes their field of view
Solution Approach 1:
The system performs preliminary actions by predicting future viewport locations before the viewer actually looks in those directions. The server analyzes current viewport position, heading rate, and video content characteristics to pre-fetch and prepare high-resolution video frames for anticipated future viewports, ensuring quality is maintained when the viewer turns their view
Solution Approach 2:
The system dynamically adjusts the field of view parameters based on real-time viewer behavior. It continuously monitors viewport position, calculates heading rate, and adapts the prediction model to change the requested field of view size and position, allowing the system to respond flexibly to rapid viewer movements while optimizing bandwidth usage
2Reliability
If the client pre-fetches video frames to maintain quality during viewport changes, then video quality consistency is improved, but buffering costs and system resource consumption increase
Solution Approach 1:
The system applies local quality by selectively pre-fetching high-resolution video frames only for the specific future viewport locations that are predicted to be viewed, rather than uniformly pre-fetching all possible frames. This targeted approach maintains video quality where needed while minimizing unnecessary buffering of frames that will not be displayed
Solution Approach 2:
The system changes parameters dynamically by adjusting the field of view size and position based on predicted viewport movement. By calculating optimal field of view parameters considering heading rate and video characteristics, the system determines precisely which frames to pre-fetch, reducing overall buffering requirements while maintaining quality during transitions
3Device complexity
If the system sends a single request for frames and downloads them sequentially, then system complexity is reduced, but the ability to respond to rapid viewport changes deteriorates
Solution Approach 1:
The system performs preliminary actions by sending multiple frame requests in advance based on predicted viewport positions. Instead of waiting for sequential frame downloads, the server anticipates future viewport locations and pre-requests the necessary video frames, enabling rapid response when the viewer turns their view without increasing operational complexity
Solution Approach 2:
The system dynamically adjusts request parameters by continuously monitoring viewport position and calculating optimal field of view settings. This dynamic approach allows the system to adapt request patterns to match actual viewer behavior, improving response speed to viewport changes while maintaining manageable system complexity through automated parameter adjustment
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Predictive pre-fetching of streams for 360 degree video is described. User view orientation metadata is obtained for a 360 degree video stream that includes data for a plurality of viewports. Data corresponding to one or more high-resolution frames for a particular one of the viewports is pre-fetched based on the user view orientation metadata and those frames are displayed. The high resolution frames are characterized by a higher resolution than for remaining viewports.