Ray Traversal via Virtual Grids for Bounding Volume Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The computational expense of ray tracing in computerized rendering, particularly due to the need for millions of ray/triangle intersection tests, is not efficiently addressed by existing acceleration structures like BVH, especially for large datasets that exceed on-chip storage capacity.
Innovation Solution
A hardware-based bounding volume detection circuit processes a target grid to identify the first intersected bounding volume by a ray, utilizing plane intersect descriptors and ray direction bits to determine the closest intersecting plane, thereby reducing the number of unnecessary intersection tests through a hierarchical grid structure and efficient grid addressing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional BVH structures are used for ray tracing, then the number of ray/triangle intersection tests is reduced, but the structure becomes impractical for large datasets exceeding on-chip storage capacity
Solution Approach 1:
The patent divides the scene into a hierarchical grid structure where space is partitioned into coarse grids and fine grids. This segmentation allows the acceleration structure to be distributed across multiple memory tiers (on-chip and off-chip), enabling large datasets to be handled by processing only relevant grid portions during ray traversal while maintaining high intersection test reduction efficiency
Solution Approach 2:
The patent introduces a hierarchical grid dimension overlaying the traditional BVH structure. By organizing bounding volumes within grid cells at multiple hierarchy levels, the system adds a spatial indexing dimension that enables efficient memory access patterns and allows the acceleration structure to scale to large datasets without requiring all data to reside in on-chip memory
2Manufacturing precision
If all triangles are tested for ray intersection to ensure rendering quality, then rendering accuracy is maintained, but computational expense becomes impractical with millions of tests
Solution Approach 1:
The patent performs preliminary organization of bounding volumes into the hierarchical grid structure before ray tracing. This pre-processing groups triangles by spatial location across multiple grid levels, enabling the system to quickly identify and test only those triangles whose bounding volumes intersect with the ray, thereby maintaining rendering quality while dramatically reducing the number of intersection tests required
Solution Approach 2:
The hierarchical grid structure acts as an intermediary between the ray and the triangles. Instead of directly testing rays against all triangles, the grid hierarchy serves as a filtering mechanism that guides rays to relevant bounding volumes and triangles, ensuring that high-quality rendering is achieved through selective testing rather than exhaustive testing of all primitives
Data Source
AI summary
Circuitry for identifying a closest intersected bounding volume in a target grid in a 3-dimensional scene includes circuitry that generates grid coordinates of candidate x-, y-, and z-planes in the target grid that belong to bounding volumes and which are closest to the ray. Comparator circuitry determines the closest plane among the candidate x-, y-, and z-planes to the ray. Circuitry generates an identifier of the bounding volume using the grid coordinates of the closest plane. The allocation of resources for some components comprising the circuitry is based on the number of planes that define the target grid, and for other components of the circuitry is based on the number of bounding volumes being considered.