Bounding Volume Hierarchy Using Virtual Grid Bins
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ray tracing techniques face inefficiencies in calculating intersections between rays and triangles, particularly due to the computational expense of testing millions of ray/triangle intersections, which can be impractical for scenes with many thousands of triangles.
Innovation Solution
A lossless multi-level hierarchical top-down builder is developed to organize spatial data sets and create an acceleration structure, minimizing data movement by limiting hierarchical levels and targeting a balanced tree based on spatial density, thereby reducing the number of ray/triangle intersection tests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a traditional BVH with deep hierarchical structure is used to organize spatial data, then the bounding volume partitioning becomes more detailed and precise, but the data movement and traversal complexity increase significantly
Solution Approach 1:
The patent segments the scene into discrete bins organized in a wide shallow tree structure rather than using deep hierarchical subdivisions. Each bin contains spatially localized triangles, and the segmentation is performed by dividing the scene space into a grid of bins at each level, reducing traversal complexity while maintaining partitioning precision.
Solution Approach 2:
The patent introduces a new dimensional organization by using a wide shallow tree with multiple bins per node, effectively adding a horizontal dimension to the hierarchical structure. This allows the BVH to maintain detailed spatial partitioning without increasing vertical depth, thereby reducing data movement and traversal complexity.
2Reliability
If the number of hierarchical levels in the BVH is increased to improve spatial organization, then the ray traversal becomes more precise, but the computational time and data movement increase
Solution Approach 1:
The patent uses partial action by limiting the hierarchical depth while compensating with increased binning capacity at each level. Instead of creating many deep levels, the system creates a sufficient number of bins at shallower levels to achieve the necessary ray traversal accuracy, reducing unnecessary computational overhead.
Solution Approach 2:
The patent changes the structural parameters of the BVH by optimizing the balance between tree depth and bins per node. By adjusting these parameters to create a wide shallow structure, the system maintains ray traversal accuracy while minimizing the computational time and data movement associated with deep hierarchical structures.
3Productivity
If a bottom-up BVH organization is used to fit data sets on-chip, then the data access efficiency improves, but the adaptability to large data sets that exceed on-chip memory is limited
Solution Approach 1:
The patent implements a dynamic BVH organization that can adapt to different data set sizes. The wide shallow tree structure with configurable bins per node allows the system to optimize for on-chip memory when data sets are small while maintaining the ability to handle larger data sets that require external memory, providing both data access efficiency and scalability.
Solution Approach 2:
The patent creates a universal BVH structure that serves multiple functions: it provides efficient on-chip data access for smaller data sets while simultaneously supporting larger data sets that exceed on-chip capacity. The wide shallow tree with binning at each level achieves this multi-functionality by balancing local data organization with global scene representation.
4Adaptability or versatility
If a top-down BVH organization is used to handle large data sets, then the adaptability to different scene sizes improves, but the data movement and processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the scene into bins at each level of the top-down hierarchy, which simplifies the processing complexity compared to traditional continuous subdivision methods. This binning approach maintains scene size adaptability while reducing data processing complexity by organizing triangles into discrete, manageable groups.
Solution Approach 2:
The patent adds a dimensional aspect to the top-down organization by introducing multiple bins per node, creating a wide shallow structure. This dimensional change allows the system to handle large data sets with high adaptability while reducing the vertical traversal depth and associated processing complexity.
Data Source
AI summary
A method and system for ray tracing can include defining a bounding volume hierarchy (BVH) of a scene comprising several objects. Objects can be binned to the BVHs using bitmaps that represent the objects and bitmaps that represent the bounding volumes.