NVIDIA introduces DLSS 5 generative AI graphics technology for photorealism

Nvidia Chief Executive Jensen Huang introduced DLSS 5 during the company's keynote at Nvidia GTC on Monday. The new version of the chipmaker's AI graphics technology is designed to make video games more realistic while using less compute power. It combines traditional 3D graphics data with generative AI models that can predict and fill in parts of an image. This allows Nvidia's GPUs to produce detailed scenes and lifelike characters without rendering every element from scratch. Huang said the system fuses controllable 3D graphics, the ground truth of virtual worlds and structured data with generative AI and probabilistic computing. One is completely predictive while the other is probabilistic yet highly realistic, he said. The executive noted that combining those ideas allows developers to create content that is beautiful, amazing as well as controllable. He described the concept of fusing structured information and generative AI as one that will repeat in industry after industry. Structured data is the foundation of trustworthy AI, Huang said. Gaming accounts for a smaller share of Nvidia's revenue today than in the past. Huang presented DLSS 5 as an example of a broader computing shift that could extend into enterprise computing. He cited platforms such as Snowflake, Databricks and BigQuery as structured datasets that future AI systems could analyze. In the future these data structures will be used by AI, which will be much faster than humans, Huang said. Future agents will use both structured databases and the unstructured generative database that represents the vast majority of the world.