Nvidia reveals the two DLSS 5 AI filter sliders

When Nvidia dropped the news about DLSS 5 back in March, the reaction was predictably mixed. Many observers felt that the demonstrations suggested nothing more than applying a simple ‘yassifying’ AI filter over existing game scenes. The initial skepticism was understandable, given the technical framework behind the technology.

However, as industry experts delved into the specifics at SIGGRAPH 2026, Nvidia clarified how this sophisticated system actually operates. They confirmed that every AI frame is built upon traditional 2D frames and motion vectors, essentially confirming that the magic remains an advanced filtering technique rather than a fundamental change to geometry.

To address developer concerns about control, Nvidia revealed that they are providing two main sliders for customizing DLSS 5. While some might have expected more granular options, these controls focus on manipulating the visual quality rather than altering the scene itself.

These sliders are dedicated to ‘structural intensity’ and ‘tone intensity’. The structural slider is designed to modulate the high-frequency details of the frame, affecting elements like ambient occlusion, subsurface scattering, and reflections. It is crucial to note that this adjustment uplifts the image without actually changing the underlying geometry.

The tone intensity slider handles the low-frequency detail, which governs expressive elements such as lighting and the overall mood or color palette of the scene.

This distinction gives developers powerful tools for artistic fine-tuning. For instance, a game designer could independently adjust the structural intensity to add realistic depth—making materials like subsurface scattering pop—while leaving the tone intensity untouched to preserve the original artistic lighting.

This level of separation means that developers can achieve highly specific results, such as giving background objects a more realistic rendering while maintaining a lighter touch on character details. This fine control is particularly valuable because implementing complex effects like subsurface scattering would traditionally be prohibitively expensive in conventional rendering pipelines.

Nvidia also confirmed that the underlying model is designed to preserve artistic intent through internal buffers, ensuring that creative vision remains intact even when processing is applied.

Despite these capabilities, the development of DLSS 5 is ongoing. Nvidia is actively working with partners to incorporate developer feedback and is committed to building additional controls as the technology matures. We will have to wait and see what further customization options are rolled out before the full feature set launches this fall.