Texture Image Storage Access Reduction via Clear Color Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in reducing power consumption when accessing and storing texture images, particularly in portable devices, due to the increased number of storage accesses required for uncompressed texture images, which leads to higher power usage and limited battery life.
Innovation Solution
The technique involves analyzing texture images to determine a clear color, reorganizing pixel color values from raster-scan to tiled order, and compressing adjacent blocks of pixels with similar colors, allowing for fewer storage accesses and reduced power consumption by avoiding unnecessary retrievals and optimizing storage page access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If uncompressed texture images are used, then image quality is improved, but the number of storage accesses increases
Solution Approach 1:
The patent extracts only the non-clear pixels from the texture image and stores them separately, while clear pixels are handled through a different mechanism (clear color buffer). This reduces the amount of data that needs to be accessed from main storage, thereby reducing the number of storage accesses while maintaining image quality.
Solution Approach 2:
The patent segments the texture image data into clear pixels and non-clear pixels, storing them in different locations. This segmentation allows the system to optimize access patterns by only retrieving necessary pixel data from storage, reducing overall access frequency while preserving complete image quality.
2Measurement precision
If uncompressed texture images are used, then image quality is improved, but power consumption increases
Solution Approach 1:
By extracting and separately storing only the non-clear pixels, the patent reduces the volume of data that must be transferred from storage to processing units. This extraction strategy directly reduces power consumption associated with data movement while preserving full image quality, which is critical for portable devices with limited battery capacity.
Solution Approach 2:
The segmentation of texture data into clear and non-clear portions enables differentiated handling and storage. This segmentation reduces the overall data transfer requirements and processing overhead, thereby lowering power consumption while maintaining uncompressed image quality for the non-clear portions that require detailed representation.
3Quantity of substance
If compressed texture formats are used, then storage space is reduced, but application complexity increases
Solution Approach 1:
The patent extracts clear pixels from the texture image and stores them in a simplified format (clear color buffer with metadata), separate from the main texture data. This extraction approach reduces storage space requirements while avoiding the complexity of general-purpose compression algorithms, as the clear pixel storage uses a straightforward metadata-based representation.
Solution Approach 2:
By segmenting the texture data into clear and non-clear portions with different storage strategies, the patent achieves space efficiency without uniformly applying complex compression. The clear portions use compact metadata storage, while non-clear portions use selective storage, reducing overall complexity compared to applying a single compression scheme to all texture data.
Data Source
AI summary
Various embodiments are generally directed to techniques for reducing storage access bandwidth requirements in retrieving a texture image from a storage for applying textures to rendered objects by rendering the texture image itself into the storage to reduce the storage space in which the texture image is stored and to arrange portions of the texture image to be retrieved with fewer accesses. A device to render images includes a processor component; a color analyzer to determine a clear color of a texture image stored as source texture data; and a rendering routine to render the texture image into a storage as reduced texture data, the rendering routine to selectively store in the reduced texture data pixel color values retrieved from the source texture data that are associated with pixels of the texture image not colored with the clear color. Other embodiments are described and claimed.