The system integrates into game engines in a similar way to earlier DLSS features like super resolution and frame generation. While the core geometry, textures, and materials of a game remain unchanged, DLSS 5 applies a neural network to reinterpret lighting and shading across the scene. The model is designed to recognize different types of surfaces—such as skin, hair, water, and metal—and treat them differently to produce more realistic visual responses.
In demonstrations shown at the event, characters displayed more detailed lighting effects, including improved subsurface scattering in skin and more natural hair rendering. Environmental scenes also showed changes in how shadows and ambient light behave, particularly around objects and foliage. These effects aim to make environments appear more grounded and visually consistent compared with traditional rendering methods.
Nvidia says the system works across different rendering pipelines, including standard rasterized games as well as titles using ray tracing or path tracing. According to the company, higher-quality rendering inputs allow the AI model to produce better lighting and material responses.
The technology remains under development. Nvidia described the demonstrations as a “snapshot” of the current state of the project, which has been in development for about three years. Some visual artifacts were visible in early builds, and further optimizations are expected before the technology launches.
Another open question involves the computational cost of the system. Nvidia’s live demos used two RTX 5090 GPUs, with one running the game itself and the other dedicated to processing the DLSS 5 lighting model. The company says the final implementation is designed to run on a single GPU, though performance optimizations and memory requirements are still being refined.
DLSS 5 is expected to appear as a new option in PC game graphics menus alongside existing DLSS features. In the demonstrations, it was integrated with frame generation technology, effectively meaning every frame’s lighting is produced by the AI system.
Nvidia said developers have responded positively to the technology and that multiple games already plan to support it when the feature launches. The company emphasized that DLSS 5 will complement existing rendering systems rather than replace them entirely, and developers will have tools to adjust how the system interacts with their games.
With DLSS 5, Nvidia is pushing machine learning deeper into the graphics pipeline, using AI to simulate lighting effects that previously required more advanced hardware or years of incremental rendering improvements.
This analysis is based on reporting from Digital Foundry.
Image courtesy of Nvidia.
This article was generated with AI assistance and reviewed for accuracy and quality.
