Nvidia has clarified the controls available to developers for its upcoming DLSS 5 technology, revealing two primary sliders that will govern the AI filter's application. Initially announced in March, DLSS 5 drew mixed reactions due to early demonstrations that suggested it functioned as an AI filter applied to game scenes. Nvidia confirmed that the technology processes traditional 2D frames and motion vectors, reinforcing this perception.
At SIGGRAPH 2026, the company elaborated on these controls. Developers will have access to sliders for 'structural intensity' and 'tone intensity.' Nvidia emphasizes that 'structural intensity' does not alter game geometry but rather "uplifts the image" by modulating high-frequency details such as ambient occlusion, subsurface scattering, and reflections. The 'tone intensity' slider, conversely, adjusts low-frequency details, influencing elements like lighting and the overall mood of the scene.
These two sliders offer developers a degree of control, allowing them to fine-tune the visual output. Additionally, developers can selectively mask different objects within a game scene to apply varying levels of DLSS 5 intensity independently. For example, background elements could receive a more pronounced effect while characters are treated with a lighter touch.
Nvidia suggests that this targeted approach can enhance realism without compromising the scene's established tone. The company highlighted that features like subsurface scattering, which can significantly impact visual fidelity, are computationally expensive to achieve with traditional rendering methods but can be enhanced by DLSS 5. \n
