Rendered Viewport Metrics for Lower-Overhead VR Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing streaming mechanisms for traditional video are not designed to efficiently stream VR video due to its significantly higher data requirements and dynamic viewport changes, leading to redundant viewport information communication.
Innovation Solution
A rendered viewport metric is employed that includes start and end times or durations for viewport entries, allowing a single entry to cover multiple frames until the viewport changes, reducing redundant data communication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If viewport information is signaled for each VR video frame, then complete viewport tracking is achieved, but redundant data communication occurs when viewport remains constant
Solution Approach 1:
The patent merges multiple frame viewport indicators into a single viewport entry that spans multiple frames when the viewport remains constant. This combines redundant information into one compact representation, reducing communication overhead while preserving complete viewport tracking accuracy across all frames.
Solution Approach 2:
The patent implements dynamic viewport signaling where the granularity of viewport information adapts based on viewport stability. When viewport is constant across frames, coarse-grained multi-frame indicators are used; when viewport changes, fine-grained per-frame indicators are applied, optimizing the balance between tracking precision and data efficiency.
2Adaptability or versatility
If traditional video streaming mechanisms are used for VR video, then existing infrastructure is utilized, but efficiency is insufficient due to higher data requirements and dynamic viewport changes
Solution Approach 1:
The patent modifies the signaling parameters of viewport information to suit VR video characteristics. By changing how viewport data is structured and transmitted - using multi-frame indicators with start time and duration fields instead of per-frame indicators - the system adapts traditional streaming mechanisms to efficiently handle VR video's higher data requirements and dynamic viewport behavior.
3Quantity of substance
If a single viewport entry covers multiple frames, then communication overhead is reduced, but viewport change detection timing must be optimized
Solution Approach 1:
The patent employs feedback mechanisms where the system monitors viewport stability across frames and dynamically adjusts the multi-frame indicator duration. When viewport changes are detected, the feedback loop shortens the indicator duration to ensure timely detection and signaling of the change, preventing excessive delay while maintaining compression benefits.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A mechanism implemented in a Dynamic Adaptive Streaming over Hypertext Transfer Protocol (HTTP) (DASH) network is disclosed. The mechanism includes receiving a DASH Media Presentation Description (MPD) describing media content including a virtual reality (VR) video sequence. The media content is obtained based on the MPD. The media content is forwarded to one or more rendering devices for rendering. A rendered viewports metric is determined that includes viewport information for the VR video sequence as rendered by the one or more rendering devices. The rendered viewports metric includes a plurality of entries with at least one of the entries indicating a viewport and a plurality of media samples of the VR video sequence applied to the viewport. The rendered viewports metric is transmitted toward a provider server.