Dense Geometry Interpolation with Shared Vertices and Topology

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image synthesis techniques in ray tracing are inefficient in terms of storage and processing due to the duplication of vertex information in compressed data structures, leading to increased memory usage and computational complexity.

Innovation Solution

The implementation of a dense geometry format that supports interpolation, where unique vertices are stored once and shared topology information is used across multiple interpolation points, allowing for the generation of intermediate geometry using interpolation parameters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If vertex information is stored multiple times in compressed data structures to represent different interpolation points, then geometry can be generated for multiple interpolation points, but memory usage and data duplication increase

Engineering Contradiction:

Improvesupport for multiple interpolation pointsVSAvoidstorage space for vertex information

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent segments the geometry data into two distinct parts: unique vertex information stored once and topology information that can be shared across multiple interpolation points. This segmentation allows the system to store vertex data only once while still supporting multiple interpolation points through shared topology references, directly resolving the contradiction between versatility and storage space.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The topology information is designed to be universal and can be shared across multiple interpolation points. The same topology data structure serves multiple purposes by referring to the same unique vertices for different interpolation points, enabling one set of data to perform multiple functions and eliminating redundant storage.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If vertex information is duplicated for each interpolation point, then geometry can be processed for different interpolation points, but computational complexity increases

Engineering Contradiction:

Improveinterpolation point supportVSAvoidprocessing complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

By segmenting data into unique vertices and shared topology, the processing complexity is reduced. Instead of processing duplicate vertex information for each interpolation point, the system only processes unique vertex data once and then shares it through topology references, simplifying the computational workload while maintaining support for multiple interpolation points.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent merges the processing of multiple interpolation points into a single unified structure. By combining unique vertex data with shared topology information, the system processes geometry for multiple interpolation points more efficiently, reducing computational complexity through data sharing rather than processing separate duplicate datasets.

Inventive Principle:

Principle #5Merging (Combining)

3Reliability

If acceleration structures are rebuilt for each frame, then geometry can be accurately represented, but processing time increases

Engineering Contradiction:

Improvegeometry representation accuracyVSAvoidprocessing time for acceleration structure rebuilding

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-defining the data structure with unique vertices and shared topology information before processing. This preliminary organization of data allows acceleration structures to be maintained across frames without rebuilding, as the shared topology can be efficiently reused, thereby reducing processing time while maintaining accuracy.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

Instead of discarding and rebuilding acceleration structures for each frame, the patent recovers and reuses the shared topology information across frames. The unique vertex data and topology references are preserved and reused, eliminating the need for complete rebuilding and reducing processing time while maintaining geometric accuracy.

Inventive Principle:

Principle #34Discarding and recovering

Data Source

PatentUS20250391058A1Interpolated geometry in dense geometry format encoding

Publication Date: 2025.12.25 ADVANCED MICRO DEVICES INC

AI summary

A geometry compression format is described. The compression format eliminates the need to store duplicate vertex information by storing unique vertices in each compressed data structure. Different triangles can refer to the same vertex using an index value, meaning that even if the same vertex is used multiple times in the compressed data structure, the entirety of the vertex information (e.g., positional information) does not need to be stored multiple times. An improvement can be made to the dense geometry format to support interpolation. More specifically, in an example, geometry for two different interpolation points can be provided in a compressed data structure encoded using the dense geometry format. Using an interpolation parameter, new interpolated geometry can be derived from the two different interpolation points. This interpolated geometry represents an intermediate point between the two interpolation points, where the similarity to either interpolation point is dependent on the interpolation parameter.