Nvidia Unveils DLSS 4.5 at CES for Better Gaming

Nvidia unveiled an upgraded version of its Deep Learning Super Sampling technology, DLSS 4.5, during its presentation at the Consumer Electronics Show in Las Vegas on Wednesday. The announcement, aimed at enhancing gaming performance on RTX graphics cards, introduces improvements to image upscaling and a new frame generation system designed to deliver smoother visuals and higher frame rates.

At the core of DLSS 4.5 is a second-generation transformer model for Super Resolution, which Nvidia described as providing a deeper understanding of scenes through expanded training data and five times the compute power of the original model introduced in DLSS 4. This upgrade eliminates previous limitations, such as detail clipping and shadow crushing in high-contrast areas, by processing data in linear space for more accurate lighting, finer edges, and improved motion clarity. The technology also enhances temporal stability, reduces ghosting, and improves antialiasing, resulting in sharper images overall. Nvidia noted a minor performance trade-off compared to the prior version, but emphasized gains in visual quality across supported titles.

A standout addition is Dynamic Multi Frame...