MSAA Upscaling With Pretrained Blending Weights for Low-Power GPUs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing graphics processing methods for upscaling images on low-power devices face challenges due to high computation requirements and memory bandwidth, while simple algorithms fail to deliver high-quality results, and proprietary hardware solutions are not applicable to general graphics acceleration hardware.
Innovation Solution
Utilizing multi-sample anti-aliasing (MSAA) intermediate buffers to reconstruct higher resolution images by leveraging pre-trained blending weight vectors and micro-geometry information, which are trained offline to minimize run-time computation complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If existing graphics processing methods are used for upscaling images, then high-quality image output can be achieved, but computation requirements and memory bandwidth increase significantly
Solution Approach 1:
The patent pre-trains blending weight vectors offline using high-quality reference images and MSAA intermediate buffers. This preliminary action transfers complex computation from runtime to training time, enabling low-power devices to perform simple weighted sums during actual upscaling while maintaining high image quality
Solution Approach 2:
The patent segments the upscaling process into distinct components: MSAA intermediate buffer generation, surrounding subsample pattern determination, blending weight vector selection from lookup tables, and final weighted sum computation. This segmentation allows each component to be optimized independently, reducing overall computational requirements
2Power
If simple upscaling algorithms are used, then computation requirements are reduced, but image quality deteriorates
Solution Approach 1:
The patent introduces MSAA intermediate buffers as an intermediary that captures micro-geometry information at runtime with minimal computational overhead. These buffers serve as a bridge between the simple rendering process and the quality-enhanced upscaling output, enabling high-quality results without complex runtime computation
Solution Approach 2:
Complex blending weight vectors are pre-computed and stored in lookup tables during offline training. This preliminary action replaces complex runtime algorithms with simple table lookups and weighted sums, maintaining high image quality while minimizing computation requirements on low-power devices
3Productivity
If proprietary hardware solutions are implemented, then upscaling performance improves, but adaptability to general graphics acceleration hardware decreases
Solution Approach 1:
The patent designs a upscaling approach that works with standard GPU hardware components already present in general graphics acceleration systems. By utilizing existing MSAA infrastructure, lookup tables, and blending operations, the solution achieves high performance without requiring proprietary hardware, ensuring broad adaptability across different platforms
Data Source
AI summary
Processing circuitry (e.g., GPU) obtains a first MSAA intermediate buffer from a rendering of a first image, the first MSAA intermediate buffer includes a plurality of subsamples for each pixel of the first image. The processing circuitry determines, for a first pixel of the first image, a first surrounding subsample pattern based on first subsamples in a surrounding region of the first pixel. Further, the processing circuitry obtains, based on the first surrounding subsample pattern, a first blending weight vector for a first location among a plurality of locations that define output pixel locations within the first pixel for upscaling the first pixel to a plurality of output pixels in an upscaled image of higher resolution than the first image. The first blending weight vector includes first blending weights respectively for first candidate subsamples associated with the first pixel. The processing circuitry generates the upscaled image accordingly.