Nvidia has just announced its next version of the deep learning super sampling technique, DLSS 3.5.
Since its launch, DLSS has gone through multiple iterations. DLSS 1.0 was the first version of Nvidia’s upscaling technology and was improved further to produce a much more stable and almost native image with DLSS 2.0.
Its successor, DLSS 3.0, is only supported by the current generation hardware and comes with frame generation to improve performance in intensive scenarios like ray tracing or path tracing.
DLSS 3.5 focuses on ray reconstruction or (RR). This version of deep learning super sampling will improve the visual quality of ray tracing features, including RT Global Illumination, RT Shadows, RT Ambient Occlusion, and RT Reflections.
Apart from the other two ray tracing titles, this is the first time Nvidia has confirmed ray tracing for Alan Wake 2. The game is being developed by Remedy Entertainment and will include support for the new DLSS 3.5 at launch.
Control was one of the earliest showcases of ray tracing and was also developed by the same studio.
According to Nvidia, its new ray reconstruction upscaling technology is trained with up to 5 times more data compared to DLSS 3. This should help it produce a superior image while ensuring slightly better performance.
Most scenarios are unlikely to show additional performance gains, but Nvidia’s official slide shows a difference of about 8% between DLSS 3 and DLSS 3.5. Nonetheless, the latter builds on the foundations of DLSS 3 and DLSS 2.
It remains to be seen if DLSS 3.5 will be exclusive to RTX 4000 graphics cards like frame generation. Since both technologies are entirely different from one another, there is a possibility that ray reconstruction may be available to older RTX graphics cards.
Thank you! Please share your positive feedback. 🔋
How could we improve this post? Please Help us. 😔