3D Image Depth Adjustment for VR Sweet Spot Degradation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current virtual reality applications face challenges in maintaining high-quality 3D experiences when users move outside the limited 'sweet spot' due to significant degradation in image quality and introduction of artefacts, particularly due to lack of de-occlusion data, which existing solutions fail to optimize in terms of user experience, data rate, complexity, and resource usage.
Innovation Solution
An image generating apparatus that determines view poses for each eye and adjusts depth values based on the difference between the current and reference poses, synthesizing output images by view shifting the reference images, allowing for improved user experience and reduced degradation by modifying depth values to maintain a consistent 3D perception even when the user moves outside the sweet spot.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If view shifting is performed based on original depth values when user moves outside the sweet spot, then image quality degrades significantly and artefacts are introduced, but if depth values are modified to reduce the difference from reference depth, then image quality degradation is reduced and user experience is improved
Solution Approach 1:
The patent modifies depth values dynamically based on the difference between current and reference view poses. When the user moves outside the sweet spot, the system adjusts the depth parameters to reduce the difference from reference depth values, thereby maintaining image quality while allowing greater user movement freedom. This parameter adjustment prevents the severe quality degradation that would otherwise occur with large pose differences.
2Loss of energy
If a limited sweet spot is enforced to maintain high image quality, then data rate is reduced, but user movement freedom is restricted
Solution Approach 1:
The system dynamically adjusts depth values based on the user's current view pose relative to the reference pose. Instead of enforcing a fixed sweet spot boundary, the patent continuously modifies depth parameters to maintain image quality across a wider range of movements. This dynamic adaptation allows users to move more freely while keeping data transmission efficient, as the system only transmits necessary depth adjustments rather than complete scene data.
3Reliability
If depth values are heavily modified to maintain image quality outside the sweet spot, then user experience is improved, but computational complexity increases
Solution Approach 1:
The patent applies depth value modification selectively based on the specific view pose difference. Rather than uniformly processing all pixels or applying complex algorithms throughout the scene, the system focuses computational effort on adjusting depth values in regions where the pose difference impacts image quality. This localized approach maintains consistent user experience while reducing overall processing complexity compared to global scene re-rendering.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An apparatus comprises a determiner (305) which determines a first-eye and a second eye view pose. A receiver (301) receives a reference first-eye image with associated depth values and a reference second-eye image with associated depth values, the reference first-eye image being for a first-eye reference pose and the reference second-eye image being for a second-eye reference pose. A depth processor (311) determines a reference depth value, and modifiers (307) generate modified depth values by reducing a difference between the received depth values and the reference depth value by an amount that depends on a difference between the second or first-eye view pose and the second or first-eye reference pose. A synthesizer (303) synthesizes an output first-eye image for the first-eye view pose by view shifting the reference first-eye image and an output second-eye image for the second-eye view pose by view shifting the reference second-eye image based on the modified depth values. The terms first and second may be replaced by left and right, respectively or vice verse. E.g. the terms first-eye view pose, second-eye view pose, reference first-eye image, and reference second-eye image may be replaced by left-eye view pose, right-eye view pose, reference left-eye image, and reference right-eye image, respectively.