AMD's neural texture compression could reduce storage and VRAM requirements

AMD reveals a new neural texture compression system aimed at reducing storage and VRAM requirements, similar to Nvidia's method from last year.

: AMD announced a neural texture compression system to reduce game data usage at the Eurographics Symposium on Rendering. The technology, similar to one proposed by Nvidia, aims to compress textures using neural networks, decreasing install sizes and VRAM usage. More details will be shared at the symposium next month.

AMD has introduced a new texture compression system, 'Neural Texture Block Compression,' aimed at reducing storage and VRAM requirements for games. This technology utilizes neural networks to compress game textures, thus decreasing download sizes and easing the data burden on graphics cards.

Despite similarities to Nvidia's 'Neural Texture Compression' method proposed last year, AMD's system's integration and usage requirements for developers remain unclear. Nvidia's approach promised higher image quality at lower file sizes but has yet to be widely adopted in games, possibly due to development challenges.

AMD's new system could significantly benefit users with slower internet or smaller SSDs by reducing the size of game installations, which can often exceed 100 GB. More details on the compression method and its implementation will be revealed during AMD's presentation at the upcoming Eurographics Symposium in London.