Delta-PVS Visibility Precomputation for 3D Streaming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for from-region visibility precomputation in 3D graphics, such as the BSP/portal-sequence method, face challenges with high computational cost, overestimation of visible sets, and inefficiency in handling complex scenes, especially outdoor and open environments, due to their reliance on closed-cell and portal relationships which do not effectively account for freestanding occluders and occluder fusion.
Innovation Solution
The method determines conservative linearized umbral event surfaces in the primal space, estimating their deviation from exact surfaces and adaptively refining them to provide a more precise and efficient from-region visibility precomputation, employing polygonal approximations to quadric visibility event surfaces and using delta-PVS storage to reduce storage and data transfer requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If BSP/portal-sequence method is used for from-region visibility precomputation, then visibility determination can be achieved, but computational cost becomes excessively high and visible sets are overestimated
Solution Approach 1:
The patent segments the visibility determination problem by introducing viewcells (spatial partitions) and organizing geometry into scene graphs with hierarchical structures. This segmentation allows the system to compute visibility in smaller, manageable units rather than processing entire scenes at once, reducing computational overhead while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary visibility precomputation during an offline phase, calculating viewcell-to-scene-graph-cell visibility relationships in advance. These precomputed results are stored and reused during runtime, eliminating the need for expensive real-time visibility calculations and significantly improving rendering performance.
2Adaptability or versatility
If BSP/portal-sequence method is used, then closed-cell visibility can be determined, but it fails to handle freestanding occluders and occluder fusion effectively
Solution Approach 1:
The patent introduces dynamic occlusion culling techniques that adapt to different scene configurations. The system dynamically identifies and processes freestanding occluders and occluder fusion cases during runtime, adjusting visibility calculations based on the actual spatial relationships rather than relying on static portal sequences.
Solution Approach 2:
The patent uses viewcells as intermediary spatial partitions that mediate between the camera and scene geometry. These viewcells provide a flexible framework for handling various occluder types by establishing visibility relationships at the cell level, allowing the system to accurately represent complex occlusion scenarios including freestanding occluders and occluder fusion.
3Reliability
If complete scene data is delivered for streaming, then all visible content can be displayed, but data transfer requirements increase and low-latency connections are needed
Solution Approach 1:
The patent performs preliminary visibility precomputation to determine which scene geometry and texture data are visible from each viewcell before streaming. This allows the system to deliver only the necessary data subsets corresponding to currently visible regions, dramatically reducing data transfer volumes while ensuring complete delivery of all visible content.
Solution Approach 2:
The patent applies local quality optimization by delivering high-detail geometry and textures only for currently visible regions (near the camera), while using lower-detail representations for distant or partially visible regions. This selective data delivery reduces overall data transfer requirements while maintaining visual quality where it matters most.
4Manufacturing precision
If high-detail geometry and textures are streamed, then visual quality is improved, but storage requirements and data transfer costs increase
Solution Approach 1:
The patent implements local quality optimization by associating different levels of geometry and texture detail with specific viewcells and scene graph nodes. The system streams high-detail assets only for regions that are currently visible or about to become visible, while using compressed or low-detail representations for other regions, thereby maintaining high graphics quality where needed while minimizing storage and transfer requirements.
Solution Approach 2:
The patent uses dynamic level-of-detail (LOD) management that adapts the quality of streamed graphics data based on the viewer's current position and orientation. As the camera moves through different viewcells, the system dynamically adjusts which detail levels are streamed, ensuring high visual quality for visible content while reducing data transfer for invisible or distant content.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method determines a set of mesh polygons or fragments of the mesh polygons visible from a view region having a plurality of view region vertices, the mesh polygons forming polygon meshes. The method includes determining at least one supporting polygon between the view region and the polygon meshes. The method further includes constructing at least one wedge from the at least one supporting polygon, the at least one wedge extending away from the view region beyond at least the polygon meshes. Further, the method includes determining one or more intersections of the wedges with the mesh polygons. Also, the method includes determining the set of the mesh polygons or fragments of the mesh polygons visible from the view region using the determined one or more intersections of the at least one wedge with the polygon meshes.