Nvidia Introduces DLSS 5 Generative Neural Rendering for Video Games

Nvidia has introduced DLSS 5, a new version of its Deep Learning Super Sampling technology that uses generative artificial intelligence to modify video game graphics in real time. Unlike previous versions that focused on increasing resolution and generating frames to improve performance, DLSS 5 utilizes generative neural rendering to alter the appearance of lighting, materials, skin, hair, and shadows. The system uses data from the game engine to interpret surfaces and apply learned visual phenomena, such as subsurface scattering, which simulates how light penetrates translucent materials like human skin. Nvidia describes this as one of its most significant graphical advances since the 2018 introduction of real time ray tracing.

The reception of the technology has been polarized. A center leaning source describes the initial reveal as a failure, noting that early images of characters from Resident Evil Requiem looked like they had been processed with low quality beauty filters. This source reports that CEO Jensen Huang initially dismissed gamer complaints as nonsense before the company attempted to reframe the technology as a tool for developers to use according to their artistic intent. Conversely, a center right leaning source frames the technology as a profound shift in image construction, emphasizing the technical leap from reconstructing pixels to creating a final visual appearance based on real world data. While some tech journalists at Gamescom were reportedly convinced by updated versions of the software, the technology remains a subject of intense debate among the gaming community following a leak of the software from the files of NBA 2K27.