Nvidia relaunches DLSS 5 ahead of 3 September debut
Nvidia is relaunching DLSS 5, its "neural rendering" AI technology for video games, ahead of a launch on 3 September, after an earlier preview drew criticism when a character's AI-altered face was widely mocked as "AI slop". The company is now stressing that the tool "preserves artistic intent" and gives developers full control, positioning it less as blanket image generation and more as a sophisticated post-processing filter that reworks lighting, shadow and detail without changing the underlying shapes in a scene.
Unlike typical generative AI, which can invent objects or produce inconsistent results, Nvidia says DLSS 5 is deterministic, so games will render the same way every time, though it remains GPU-intensive and can roughly halve frame rates. Developers get granular control via sliders and three visual presets ("default", "natural" and "cinematic"), and can choose to apply the effect selectively, for example to skin, hair, foliage or objects, while excluding faces via automatic or manual masking tools. The first title to use the technology will be NBA 2K27.
- Nvidia relaunches DLSS 5 neural rendering AI ahead of 3 September launch
- Tool reworks lighting/detail per frame but doesn't invent new shapes
- Developers get sliders, presets and masking to control where AI is applied
AI Art Business Companies Culture Entertainment Gaming Horror Music Technology
Read the full article at the source →
Originally published by The Verge as “Nvidia’s DLSS 5, explained”.