Lossless Fragment Color Compression for Multi-Sample Pixels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multi-sample anti-aliasing techniques face challenges in reducing memory bandwidth requirements while maintaining image quality, as they often require storing and accessing large amounts of mapping bits per pixel, leading to increased memory footprint and computational costs.
Innovation Solution
The implementation of lossless fragment color compression techniques that use metadata to specify the number of fragment colors stored in each pixel, reducing the need for sample-to-fragment mapping data and minimizing memory footprint, by storing fragment coverage metadata within the pixel itself rather than as separate metadata.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional multi-sample anti-aliasing techniques store separate sample-to-fragment mapping data, then fragment compression can be achieved, but memory footprint and bandwidth requirements increase due to storing large amounts of mapping bits per pixel
Solution Approach 1:
The patent combines the sample-to-fragment mapping data with the fragment color data by storing mapping bits within the same memory location as the fragment colors. This merging eliminates the need for separate mapping data structures and reduces overall memory footprint while maintaining the ability to perform fragment compression
Solution Approach 2:
The patent makes the fragment color memory serve multiple functions: storing both fragment color data and sample-to-fragment mapping information. This multi-functionality reduces the total memory required by eliminating dedicated mapping data storage, thereby reducing memory footprint without sacrificing compression capability
2Reliability
If fragment colors are replicated to each sample marked in the fragment's sample mask, then image quality is maintained, but memory bandwidth requirements increase significantly
Solution Approach 1:
The patent extracts only the essential mapping information (sample-to-fragment mapping bits) and stores it compactly within the fragment color memory rather than replicating full fragment color data to each sample. This extraction approach maintains the ability to reconstruct full-color images while significantly reducing memory bandwidth requirements
Solution Approach 2:
The patent changes the representation parameters by storing compact mapping bits (e.g., 2 bits per sample) instead of full fragment color values for each sample. This parameter change enables the system to maintain image quality through selective access while reducing memory bandwidth consumption by a factor of 4x or more
3Quantity of substance
If coverage information is stored in a separate control plane array, then fragment compression is achieved, but memory access complexity and bandwidth requirements increase
Solution Approach 1:
The patent merges the coverage information (mapping data) with the fragment color data by storing mapping bits within the same memory structure as the fragment colors. This consolidation eliminates separate control plane arrays and simplifies memory access operations while maintaining fragment compression benefits
Data Source
AI summary
Described herein are technologies related to facilitate lossless compression for multi-sample color data of computer graphics that maximizes the apparent quality of pixels while avoiding a corresponding burden on memory and processor bandwidth.