NVIDIA has introduced DLSS 3.5, the latest version of its renowned Deep Learning Super Sampling technology. This iteration comes with a significant breakthrough called “Ray Reconstruction,” a notable advancement aimed at refining the realism of gaming graphics through advanced rendering techniques.
DLSS 3.5 address the problem on how hand-tuned denoisers would introduce inaccurate lighitng effects due to using bad pixel from previous frame, low quality global illumination and inconsisten reflections due to not enough data available for interpret accurate lighitng.
The Ray Reconstruction employs AI-powered ray tracing, which enhances the accuracy of how light interacts with various surfaces. It is trained with 5X more data than DLSS 3, DLSS 3.5 recognizes different ray-traced effects to make smarter decisions about using temporal and spatial data, and to retain high frequency information for superior-quality upscaling.
It result in improves visual quality due to the fact that AI is able to identify what to keep and what to remove as noise, rather than denoiser algorithm that generally blends adjacent pixels togethers which results in blurry, undefined reflections, on top of potentially confusing the actual reflections with artifacts from previous frames
DLSS 3.5 will ba available to RTX card, however user of RTX 40 series will be able to combine Supersampling, Frame Generation, and Ray Reconstruction, RTX 30 and 20 series can add Ray Reconstruction and Super Resolution.
NVIDIA DLSS 3.5 will be available in this September with Cyberpunk 2077, Cyberpunk 2077: Phantom Liberty, October in Alan Wake 2, and in future, Portal with RTX, Chaos Vantage, D5 Render, and NVIDIA Omniverse.