Texture Compression with Hierarchical Palette and its Cache Design

• Main Authors: Jia-HaoYang, 楊家豪
• Other Authors: Chih-Hung Kuo
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/43438132828572706685

碩士 === 國立成功大學 === 電機工程學系 === 103 === To improve the quality of compressed texture, this thesis proposes new texture compression algorithms and a texture cache strategy. We propose a hierarchical palette to increase the color candidates without increasing the index bits. The new texture compression algorithms can provide better image quality. We adopt direction and locality of extracted texel for the texture cache strategy when texture mapping. By predicted direction, the pre-fetch block can effectively improve the cache hit rate. The average gain with our texture compression scheme is better than DXT1 about 2.3 dB PSNR. For the proposed texture cache strategy improves about 15.72% Miss rate.