Random-Access Neural Compression of Material Textures

TL;DR

This paper introduces a neural compression method for material textures that significantly increases detail, surpasses traditional image codecs in quality, and enables real-time random access decompression suitable for GPU applications.

Contribution

A novel neural compression technique that compresses multiple textures and mipmaps together, achieving higher detail, better quality than existing codecs, and real-time random access.

Findings

Achieves 16x more texels with low bitrate compression.

Outperforms AVIF and JPEG XL in image quality.

Enables real-time, random access decompression on GPUs.

Abstract

The continuous advancement of photorealism in rendering is accompanied by a growth in texture data and, consequently, increasing storage and memory demands. To address this issue, we propose a novel neural compression technique specifically designed for material textures. We unlock two more levels of detail, i.e., 16x more texels, using low bitrate compression, with image quality that is better than advanced image compression techniques, such as AVIF and JPEG XL. At the same time, our method allows on-demand, real-time decompression with random access similar to block texture compression on GPUs, enabling compression on disk and memory. The key idea behind our approach is compressing multiple material textures and their mipmap chains together, and using a small neural network, that is optimized for each material, to decompress them. Finally, we use a custom training implementation to achieve practical compression speeds, whose performance surpasses that of general frameworks, like PyTorch, by an order of magnitude.