DLSS (Deep Learning Super Sampling) is an AI-driven image upscaling technology created by NVIDIA.
DLSS employs a neural network trained on high-resolution images to upscale lower-res frames to near-native or superior visual quality.
DLSS 3.5, the most recent version, introduces Ray Reconstruction, which boosts image quality up to 4x over regular denoising techniques for real-time rendering.
DLSS and other aspects of neural graphics transform rendering by reducing rendering overhead, improving image quality, and providing creative headroom for developers.