Shadow Ray Start Node Determination via Off-Screen Rasterization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current shadow rendering techniques in computer graphics are inefficient due to the need to traverse hierarchical data structures from the root node, leading to a high number of unnecessary ray intersection tests during shadow ray tracing.

Innovation Solution

The method involves organizing primitives in a hierarchical data structure with bounding volumes and rasterizing these volumes to an off-screen render target, allowing the GPU to determine a non-root node as the start point for traversal, thereby reducing the number of intersection tests by utilizing shader units to quickly identify intersecting pixels and nodes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the GPU traverses the hierarchical data structure from the root node to determine shadow ray-primitive intersections, then the traversal process is systematic and complete, but the number of unnecessary ray intersection tests increases significantly

Engineering Contradiction:

Improvecompleteness of shadow ray intersection detectionVSAvoidtime for ray intersection tests

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-rasterizing bounding volumes to an off-screen render target before performing shadow ray tracing. This creates a depth buffer that encodes spatial information about the scene geometry, allowing the system to quickly determine which nodes in the hierarchical data structure could potentially intersect with shadow rays without performing full traversal from the root node. The depth buffer is prepared in advance and can be rapidly queried during shadow rendering.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent extracts the essential spatial information from the complete hierarchical data structure by projecting bounding volumes onto a 2D off-screen render target. This extraction process creates a condensed representation (depth buffer) that contains only the necessary information for determining potential ray intersections, eliminating the need to traverse and test against every node in the full hierarchical structure.

Inventive Principle:

Principle #2Taking out (Extraction)

2Reliability

If the GPU performs ray intersection tests on all nodes in the hierarchical data structure, then no potential intersections are missed, but the computational expense increases

Engineering Contradiction:

Improveaccuracy of shadow determinationVSAvoidcomputational expense of intersection tests

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent introduces an intermediary data structure - the off-screen render target containing rasterized bounding volumes and depth buffer - that mediates between the hierarchical data structure and the shadow ray tracing process. This intermediary encodes spatial relationships in a format that can be quickly queried, allowing the system to determine which hierarchical nodes warrant further intersection testing without examining all nodes, thus reducing computational expense while maintaining accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Ease of manufacture

If the GPU uses traditional root-to-leaf traversal of the hierarchical data structure, then the algorithm is simple to implement, but traversal efficiency is low

Engineering Contradiction:

Improvesimplicity of traversal algorithmVSAvoidshadow rendering performance

Core Design Contradiction:

Ease of manufactureVSProductivity

Solution Approach 1:

The patent replaces the mechanical sequential traversal process with a more efficient approach using shader units to query the pre-rasterized off-screen render target. Instead of systematically walking through the hierarchical structure node by node, the system uses parallel shader execution to rapidly test multiple potential intersection candidates simultaneously, substituting the traditional mechanical traversal with a parallelized query-based approach that leverages the GPU's parallel processing capabilities.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS9818221B2Start node determination for tree traversal for shadow rays in graphics processing

Publication Date: 2017.11.14 QUALCOMM INC

AI summary

At least one processor may organize a plurality of primitives of a scene in a hierarchical data structure, wherein a plurality of bounding volumes are associated with a plurality of nodes of the hierarchical data structure. The at least one processor may rasterize a representation of each of the plurality of bounding volumes to an off-screen render target in the memory. The at least one processor may determine, based at least in part on a pixel in the off-screen render target that maps to a ray in the scene, a non-root node of the hierarchical data structure associated with the pixel as a start node to start traversal of the hierarchical data structure. The at least one processor may traverse the hierarchical data structure starting from the start node to determine whether the ray in the scene intersects one of the plurality of primitives.