Parallel Grid Population for Ray Tracing Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ray tracing methods using grids are inefficient due to the high time and processing resources required for populating and repopulating grids with a large number of objects in three-dimensional scenes, especially in applications with dynamically changing scenes.
Innovation Solution
The method involves dividing a grid into distinct portions and assigning objects to processors for parallel population, allowing each processor to determine which grid portions the objects are bounded by and populate accordingly, enabling efficient and rapid grid population.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a grid is populated with a large number of objects in a three-dimensional scene, then ray tracing can avoid high-cost intersection tests by using low-cost cell intersection tests, but the time and processing resources required to populate and repopulate the grid become excessively slow
Solution Approach 1:
The patent divides the grid into multiple portions and assigns each portion to a different processor. Objects are also divided and assigned to processors, allowing parallel population of grid portions. This segmentation enables simultaneous processing of multiple objects and grid portions, dramatically reducing grid population time while maintaining the efficiency benefits of grid-based ray tracing.
2Reliability
If a single processor populates the grid sequentially with thousands or millions of objects, then the grid structure can be maintained for ray tracing, but the rendering process becomes very costly in terms of time and processing resources
Solution Approach 1:
The grid is segmented into multiple portions that can be populated independently by different processors. Each processor maintains the integrity of its assigned grid portion while working in parallel with other processors, ensuring overall grid structure integrity is preserved despite the distributed population approach.
Solution Approach 2:
Objects are pre-assigned to specific processors based on their spatial location before grid population begins. This preliminary assignment ensures that each processor knows which objects to place in which grid portions, eliminating the need for complex coordination during population and maintaining grid integrity through predetermined object-grid portion mappings.
Data Source
AI summary
Parallel population of a grid with a plurality of objects using a plurality of processors. One example embodiment is a method for parallel population of a grid with a plurality of objects using a plurality of processors. The method includes a first act of dividing a grid into n distinct grid portions, where n is the number of processors available for populating the grid. The method also includes acts of dividing a plurality of objects into n distinct sets of objects, assigning a distinct set of objects to each processor such that each processor determines by which distinct grid portion(s) each object in its distinct set of objects is at least partially bounded, and assigning a distinct grid portion to each processor such that each processor populates its distinct grid portion with any objects that were previously determined to be at least partially bounded by its distinct grid portion.