Depth-Ranged Occlusion Surfaces for Low-Latency Artificial Reality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional artificial reality environment reconstruction techniques face challenges in rendering virtual objects quickly enough to prevent latency and sensory dissonance due to sudden changes in user perspective, requiring expensive computing resources that are impractical for compact HMDs.
Innovation Solution
Generate and pose two-dimensional occlusion surfaces at varying depths representing physical objects within specific depth ranges, allowing for efficient occlusion rendering and rapid adjustment of views by resampling surfaces rather than re-rendering entire scenes, utilizing a two-stage computing system to distribute processing tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional artificial reality environment reconstruction techniques are used to render virtual objects, then rendering accuracy is improved, but computational resources and latency increase
Solution Approach 1:
The patent segments the rendering process by separating occlusion surface generation from full scene rendering. Only occlusion surfaces at predetermined distances are rendered, while other elements are handled through resampling techniques. This segmentation reduces the computational burden and latency associated with rendering complete environments while maintaining accuracy for critical occlusion elements.
Solution Approach 2:
The patent performs preliminary action by pre-generating occlusion surfaces at predetermined distances before full rendering occurs. These pre-computed occlusion surfaces are then used to quickly determine visibility relationships, avoiding the need to render entire scenes from scratch when perspective changes occur, thus reducing latency while preserving rendering accuracy.
2Measurement precision
If traditional artificial reality environment reconstruction techniques are used, then rendering quality is improved, but computing resources required increase
Solution Approach 1:
The patent extracts only the essential occlusion information from the full environment at predetermined distances, separating this critical data from the complete scene rendering process. By extracting and processing only occlusion surfaces rather than entire environments, the system maintains rendering quality for visibility determination while significantly reducing computing resource consumption.
Solution Approach 2:
The rendering workload is segmented into occlusion surface generation at specific distances and resampling operations. This segmentation allows the system to focus computational resources on generating accurate occlusion surfaces while using less intensive resampling techniques for other elements, thereby maintaining rendering quality while reducing overall computing resource requirements.
3Measurement precision
If full scene re-rendering is performed for perspective changes, then view accuracy is improved, but processing time increases
Solution Approach 1:
Occlusion surfaces are pre-generated at predetermined distances before perspective changes occur. When the user's viewpoint changes, these pre-computed surfaces are quickly resampled rather than re-rendering the entire scene, maintaining view accuracy for occlusion relationships while dramatically reducing processing time compared to full scene re-rendering.
Solution Approach 2:
The patent uses resampling to create copies of occlusion surfaces at new perspectives rather than re-rendering from scratch. This copying approach through resampling maintains view accuracy by preserving the original occlusion geometry while significantly reducing processing time compared to complete re-rendering of the scene.
Data Source
AI summary
A method includes generating a depth map of a real environment as seen from a viewpoint that comprises pixels having corresponding depth values of one or more physical objects. A first two-dimensional occlusion surface is generated representing at least a visible portion of the one or more physical objects located within a first predetermined depth range defined relative to the viewpoint. A second two-dimensional occlusion surface is generated such that a minimum depth of a second predetermined depth range is greater than a maximum depth of the first predetermined depth range. The first and second occlusion surfaces are posed in a three-dimensional coordinate system. The visibility of a virtual object is determined relative to the one or more physical objects by comparing a model of the virtual object with the first and second occlusion surfaces, and an output image is generated based on the visibility of the virtual object.