NVIDIA Neural Texture Compression Cuts VRAM 6.5GB to 970MB
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- NVIDIA's Neural Texture Compression (NTC) reduces GPU VRAM usage by up to seven times, as demonstrated in a GTC 2026 technology session.
- In a Tuscan Villa scene comparison, NTC cut VRAM consumption from 6.5 GB to 970 MB with virtually no difference in texture appearance across both exterior and interior shots.
- NTC replaces traditional block-compressed formats like BC5, BC6, and BC7 with small trained neural networks that produce the desired pixel format and texture look at a fraction of the size.
- In an interior tableware scene, NVIDIA's NTC outperformed downscaled BCn textures while operating within the same 970 MB VRAM budget.
- NVIDIA has trained NTC neural networks for nearly every game material, positioning the tech for real-world game deployment.
- NTC output can either match the base texture's appearance for major VRAM savings or render a more realistic layer on top of the base texture that games already use.
Why it matters: A 6.5 GB-to-970 MB compression ratio — roughly 85% VRAM savings on a single scene — directly lowers the memory barrier for high-fidelity textures, meaning games running on mid-range GPUs could soon deliver detail currently locked behind flagship cards. Game developers gain a tool to either free up VRAM for denser worlds or boost material quality at the same memory cost, with NVIDIA's per-material neural networks already trained for deployment.
