Anti-aliasing via Analytical Visibility Resolution

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current anti-aliasing techniques in 3D graphics, such as multi-sampled anti-aliasing (MSAA), are inefficient in terms of bandwidth and power usage, especially in regions of high complexity, as they store multiple samples per pixel even for interior triangle edges, leading to redundant shading and high memory requirements, and fail to effectively address small primitives falling between sample points.

Innovation Solution

An anti-aliasing method that renders the majority of the screen at one sample per pixel using conservative rasterization to identify and shade only pixels requiring accurate anti-aliasing, with the addition of a fixed-function fragment compositing unit for analytical visibility resolution, allowing for high-quality anti-aliasing of triangle edges and intersections with reduced bandwidth.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If multi-sampled anti-aliasing (MSAA) is used to reduce aliasing artifacts, then anti-aliasing quality is improved, but bandwidth and power consumption increase significantly

Engineering Contradiction:
Improveanti-aliasing qualityVSAvoidbandwidth
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies different sampling strategies to different regions of the image: full multi-sampling is applied only to pixels containing silhouette edges or interior silhouette edges, while interior pixels use single sampling. This local differentiation resolves the contradiction by concentrating bandwidth resources only where aliasing artifacts actually occur, rather than uniformly across the entire image.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the anti-aliasing process into distinct stages: silhouette edge detection, pixel classification, selective multi-sampling for edge pixels, and single sampling for interior pixels. This segmentation allows the system to achieve high anti-aliasing quality at edges while minimizing bandwidth consumption in interior regions.

Inventive Principle:
Principle #1Segmentation

2Manufacturing precision

If multiple samples per pixel are stored for the entire screen, then anti-aliasing is improved, but memory requirements become prohibitively high

Engineering Contradiction:
Improveanti-aliasing qualityVSAvoidmemory requirements
Core Design Contradiction:
Manufacturing precisionVSVolume of stationary object

Solution Approach 1:

The patent stores multiple samples only for pixels containing silhouette edges or interior silhouette edges, while storing single samples for interior pixels. This local differentiation dramatically reduces memory requirements while maintaining high anti-aliasing quality where it is most needed at geometric discontinuities.

Inventive Principle:
Principle #3Local quality

3Manufacturing precision

If MSAA performs anti-aliasing of all triangle edges, then edge quality is improved, but redundant shading occurs along shared edges

Engineering Contradiction:
Improveedge qualityVSAvoidredundant shading
Core Design Contradiction:
Manufacturing precisionVSLoss of energy

Solution Approach 1:

The patent identifies and processes only silhouette edges and interior silhouette edges with multi-sampling, while using single sampling for shared edges between triangles. This eliminates redundant shading operations on shared edges while maintaining high edge quality where aliasing artifacts actually manifest.

Inventive Principle:
Principle #3Local quality

4Manufacturing precision

If high sampling rates are used, then anti-aliasing quality is improved, but bandwidth and memory become prohibitive

Engineering Contradiction:
Improveanti-aliasing qualityVSAvoidbandwidth
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies high sampling rates selectively only to pixels containing silhouette edges or interior silhouette edges, while using single sampling for interior pixels. This local differentiation achieves high anti-aliasing quality at critical edges while keeping overall bandwidth consumption manageable.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP2854108B1Anti-aliasing for graphics hardware
Publication Date: 2019.12.11 INTEL CORP
  • EP2854108B1 patent drawingFigure 1~2
  • EP2854108B1 patent drawingFigure 3~4
  • EP2854108B1 patent drawingFigure 5

AI summary

Visibility may be analytically resolved rather than using point-sampling, thereby entirely avoiding geometric aliasing and the need to store multiple samples per pixel. By relying on existing techniques for shading, i.e., by shading once per fragment and focusing on visibility, visual results may be equivalent to multi-sampled anti-aliasing (MSAA) using an infinite sampling rate in some embodiments.