Nvidia has unveiled a new technology called Neural Texture Compression (NTC) that could radically change the way games use video memory. According to the company, the new system can reduce VRAM usage by up to seven times without any noticeable loss of image quality.
This technology uses artificial intelligence to compress textures much more efficiently than traditional methods. Instead of storing large high-resolution textures directly in memory, the system stores compressed neural data and reconstructs textures in real time using artificial intelligence. This allows games to maintain high visual quality while using much less video memory.
During the demonstration, Nvidia showed an example where a scene that would normally require 6.5 GB of VRAM was rendered using only 970 MB of memory with NTC activated. The visual difference was minimal, suggesting that the technology could be extremely useful for modern games with high-resolution textures and detailed environments.
Traditional texture compression methods reduce file size but often decrease image quality or introduce visual artifacts. Neural texture compression works differently: it trains a neural network to recreate textures from compressed data. When the game runs, the GPU reconstructs the texture using AI instead of loading a complete texture file into memory.
This approach not only reduces VRAM usage but also game file sizes. If textures are stored in a compressed neural format instead of large image files, games themselves could become much smaller without sacrificing visual quality.
Nvidia also stated that once developers implement the NTC technology, game installation sizes could be significantly reduced. Modern AAA games often occupy between 100 and 200 GB of memory, largely due to high-resolution textures. Neural compression could greatly reduce these sizes.
This technology could be crucial for players with limited SSD storage space and for consoles where storage space is often a major limitation.
The new technology will be part of DLSS 5, Nvidia’s next version of graphics technology based on AI, set to be released this fall. DLSS already improves performance by using AI for image upscaling. With DLSS 5, Nvidia expands the use of AI beyond image upscaling to memory optimization and texture management.
If widely adopted by game developers, neural texture compression could become one of the most important graphics technologies in the coming years. Reduced VRAM usage translates to better performance on mid-range GPUs, smaller games, and more detailed worlds without requiring expensive hardware.
Overall, Nvidia’s NTC technology, especially when combined with DLSS 5, could change how games handle textures, memory, and performance in the future, making high-quality graphics more efficient and accessible to a larger number of players.
