Rights accessRestricted access - publisher's policy
Current schemes for texture compression fail to exploit spatial coherence in an adaptive manner due to the strict efficiency constraints imposed by GPU-based, fragment-level decompression. In this paper we present a texture compression framework for quasi-lossless, locally-adaptive compression of graphics data. Key elements include a Hilbert scan to maximize spatial coherence, efficient encoding of homogeneous image regions through arbitrarilysized texel runs, a cumulative run-length encoding supporting fast random-access, and a compression algorithm suitable for fixed-rate and variable-rate encoding. Our scheme can be easily integrated into the rasterization
pipeline of current programmable graphics hardware allowing real-time GPU decompression. We show that our scheme clearly outperforms competing approaches such as S3TC DXT1 on a large class of images with some degree of spatial coherence. Unlike other proprietary formats, our scheme is suitable for compression of any graphics data including color maps, shadow maps and relief maps. We have observed compression rates of up to 12:1, with minimal or no loss in visual quality and a small impact on rendering time.
CitationAndújar, C.; Martinez, J. Locally-adaptive texture compression. A: Congreso Español de Informática Gráfica. "Congreso Español de Informática Gráfica 2009". San Sebastián: 2009, p. 105-114.
All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder. If you wish to make any use of the work not provided for in the law, please contact: email@example.com