BC7 Texture Partition Selection via Gradient Strength Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for selecting the best partition for BC7 texture encoding in computer graphics are inefficient, requiring exhaustive searches among 64 possible partitions, which is computationally costly and time-consuming.
Innovation Solution
A processor assembly identifies candidate partitions by computing gradient strengths in multiple directions, selecting the partition with the lowest absolute gradient strength, and using machine learning to predict the best compression mode based on per-channel ranges and error calculations, thereby optimizing the selection process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If exhaustive search among 64 possible partitions is performed for BC7 texture encoding, then compression quality is improved, but processing time and computational cost increase significantly
Solution Approach 1:
The patent segments the exhaustive search space of 64 partitions into multiple candidate partitions based on gradient strength analysis. By dividing the search into stages (computing gradient strengths, identifying candidate partitions, then selecting from candidates), the method maintains compression quality while reducing the computational scope at each stage.
Solution Approach 2:
The patent performs preliminary gradient strength computations and candidate partition identification before the final partition selection. This preliminary action filters out poor candidates early, so that the exhaustive search is only performed on a reduced subset of promising partitions, thereby reducing overall processing time while maintaining quality.
2Manufacturing precision
If exhaustive search among 64 possible partitions is performed for BC7 texture encoding, then compression quality is improved, but computational complexity increases significantly
Solution Approach 1:
The computational process is segmented into distinct stages: gradient strength computation, candidate partition identification, and final selection. This segmentation reduces the complexity of any single operation by breaking down the monolithic exhaustive search into manageable steps with decreasing complexity.
Solution Approach 2:
The patent performs partial exhaustive search only on candidate partitions rather than all 64 partitions. This partial action is sufficient to achieve high compression quality while avoiding the excessive computational complexity of a complete exhaustive search across all possible partitions.
3Measurement precision
If gradient strength computation is performed in multiple directions to identify candidate partitions, then partition selection accuracy is improved, but computational burden increases
Solution Approach 1:
The patent applies local quality analysis by computing gradient strengths in multiple directions (horizontal, vertical, diagonal) to identify regions with different compression characteristics. This local analysis improves partition selection accuracy by capturing directional variations in the texture data, while the results are used to limit the scope of subsequent exhaustive searches.
Solution Approach 2:
The gradient strength computation in multiple directions is performed as a preliminary filtering step before the final partition selection. This preliminary action identifies promising candidate partitions that warrant further exhaustive search, thereby reducing the overall computational burden while maintaining high selection accuracy.
Data Source
AI summary
Techniques are described for first generating a short list of candidate partitions for BC-7 texture compression using calculations of gradient strengths in multiple directions and then selecting one of the candidate partitions using pixel extents for further processing of the block.