NVIDIA has introduced DLSS 5 with neural rendering for photorealistic gaming.

At the GTC 2026 conference, NVIDIA unveiled DLSS 5, marking a shift in the DLSS technology from a performance-oriented feature set to image generation focused on maximum visual realism. The new version utilizes real-time neural rendering, adding photorealistic lighting and material detail to game frames. Support is planned for this fall.

DLSS 5 uses color data and motion vectors from each frame and applies an artificial intelligence model that enhances lighting and material response while remaining anchored to the original 3D scene. The system operates in real-time up to 4K resolution and provides a deterministic and stable result for each frame, which is a key requirement for games.

NVIDIA positions DLSS 5 as an update to the visual pipeline rather than just another step in frame generation. The model is trained to recognize scene elements such as skin, hair, fabric, and lighting conditions, using this information to enhance effects like subsurface scattering, fabric sheen, and light interaction with hair. Developers gain the ability to control effect intensity, color grading, and masking, with integration continuing through the existing NVIDIA Streamline platform used for DLSS and Reflex.

Support for the technology has already been confirmed by several major publishers and studios, including Bethesda, CAPCOM, Ubisoft, NetEase, Tencent, and Warner Bros. Games.

The first wave of games supporting DLSS 5 will include Starfield, Assassin’s Creed Shadows, Hogwarts Legacy, Resident Evil Requiem, Delta Force, NARAKA: BLADEPOINT, and The Elder Scrolls IV: Oblivion Remastered.

DLSS 5 works by passing each frame containing color data and motion vectors through an NVIDIA model that keeps the result anchored to the original 3D scene rather than creating separate image effects. Developers can precisely adjust the output using intensity control and color grading, allowing them to correct offset, contrast, saturation, and gamma to match the game's visual style. Masking allows excluding individual objects or areas of the image if developers wish to preserve the original appearance or avoid changes in selected areas.

Source: VideoCardz.com