Stronger than DLSS 4, NVIDIA has moved into fifth gear with what the company's engineers call neural rendering.
In the absence of new GeForce RTX 50 SUPER series graphics cards, or even the first hints of the GeForce RTX 60 series generation, NVIDIA is betting on software. At the very beginning of the year, DLSS 4.5 was presented as a prelude to CES 2026 in Las Vegas. At the time, NVIDIA emphasized the arrival of two major technologies: Dynamic Multi-Frame Generation and 6x Multi-Frame Generation. Two technologies that seem almost obsolete today, given the announcement made by Jensen Huang - co-founder and CEO of NVIDIA - at the opening of the GTC in San Jose just a few days ago.
Now, DLSS 5 has been announced... and presented with several impressive technology demos! The first essential point is that DLSS 5 is not here to replace previous versions, and the contributions of this new version are seen as complements to familiar technologies such as Super Resolution, Ray Reconstruction and Frame Generation. Jensen Huang emphasized that DLSS 5 may or may not be activated in addition to these technologies for even more successful rendering: DLSS 5 introduces a new real-time neural rendering model to apply " photorealistic lighting and materials to pixels " and enable " developers to offer a new level of photorealistic graphics, previously reserved for Hollywood visual effects. "
Jensen Huang adds: " The AI model is trained to understand the complex semantics of scenes, such as characters, hair, fabrics and skin, as well as ambient lighting conditions, all from the analysis of a single image. DLSS 5 then exploits this in-depth understanding to generate accurate images while preserving the structure of the original scene ". When it is released this autumn, DLSS 5 will already be supported by a number of games from NVIDIA's closest partners: we're talking about Starfield, Assassin's Creed Shadows, Hogwarts Legacy, Resident Evil Requiem, Delta Force, Naraka: Bladepoint, The Elder Scrolls IV: Oblivion Remastered.
While we wait to find out more about how this promising technology works, we'll leave you with the examples provided by NVIDIA on its GeForce Youtube page, which seem to be reserved for the most powerful configurations (GeForce RTX series 50 essential).
