Blurry to HD in seconds, NVIDIA's disruptive move over the past 25 years

IT之家 March 17 report: At the 2026 GPU Technology Conference (GTC) held today (March 17), NVIDIA CEO Jensen Huang delivered a keynote speech where he showcased DLSS 5 technology and announced that the technology will launch this fall.

Jensen Huang said: After NVIDIA invented programmable shaders 25 years ago, we are redefining computer graphics once again. DLSS 5 is the “GPT moment” for graphics—combining hand-crafted rendering with generative AI—delivering a huge leap in visual realism while maintaining the control that artists need for creative expression.

Jensen Huang said that DLSS 5 is the most disruptive breakthrough in the field of computer graphics since real-time ray tracing technology was introduced in 2018. The core of DLSS 5 lies in introducing a brand-new real-time neural rendering model, which can completely break down the barriers between traditional rendering and the real physical world, helping game developers achieve photo-realistic visuals in their works—visuals that until now could only be achieved by Hollywood VFX.

Jensen Huang emphasized that DLSS 5 is not only a redefinition of computer graphics, but also the “GPT moment” for the graphics industry. IT之家 has included the official demo comparison image below:

The technology combines traditional hand-crafted rendering with cutting-edge generative AI. In real-time operation, DLSS 5 extracts each frame’s color and motion vectors from the game as input data, and then uses AI models to inject realistic lighting and materials into the scene in real time. At resolutions up to 4K, it can maintain real-time performance, ensuring players get a smooth and responsive interactive experience.

While generative AI video models in the past could produce convincing images, they often only run offline, and the content generated each time lacks coherence and controllability. To address this pain point, NVIDIA conducted end-to-end deep training on the AI model for DLSS 5.

Now, all it takes is analyzing a single game frame, and the AI can understand complex scene semantics. Whether it’s the sheen of character hair, the texture of fabric, or the subsurface scattering effect of light beneath skin, the AI can reproduce it perfectly. In addition, it can accurately handle complex lighting conditions such as front lighting, backlighting, or overcast skies, while maintaining a high degree of consistency across different frames.

To ensure creative artists have freedom, DLSS 5 also provides developers with fine-grained control functions for intensity, color grading, and masking. The technology has already received an unprecedented level of support in the industry. Top global game developers, including B 社 (Bethesda), CAPCOM, Ubisoft, Tencent, and NetEase, have all announced plans to integrate it.

Players will be able to personally experience this revolutionary technology this fall in many blockbuster games, including Starfield, Assassin’s Creed: Shadows, and Resident Evil: Armageddon.