Nvidia’s DLSS 3.5 update revolutionizes ray tracing | ENBLE

Nvidia's DLSS 3.5 update revolutionizes ray tracing | ENBLE

Nvidia Introduces DLSS 3.5: A Game-Changing Update for Ray Tracing


Nvidia, the renowned technology company, has recently announced a major update to its Deep Learning Super Sampling 3 (DLSS 3) feature. The new DLSS 3.5 is set to launch this fall, coinciding with the highly anticipated release of Cyberpunk 2077: Phantom Liberty. This update brings something truly groundbreaking to Nvidia’s RTX feature: Ray Reconstruction.

Ray Reconstruction is a revolutionary addition to Nvidia’s RTX technology. It allows for enhanced ray tracing quality without sacrificing performance. In fact, in some cases, it can even improve performance. Nvidia emphasizes that Ray Reconstruction is primarily an image quality improvement over traditional ray tracing methods, rather than a performance booster.

DLSS 3.5 will initially be available in Cyberpunk 2077: Phantom Liberty, Portal RTX, and the upcoming Alan Wake 2 (which has been delayed to late October). Unlike Frame Generation, another DLSS feature exclusive to RTX 40-series graphics cards, Ray Reconstruction works on all RTX GPUs. Nvidia has confirmed that Ray Reconstruction will be a separate setting in the graphics menu, giving users the option to enable or disable it.

Under the DLSS 3.5 brand, Nvidia has bundled four different features: Super Resolution, Deep Learning Anti-Aliasing, Ray Reconstruction, and Frame Generation. While Super Resolution, Deep Learning Anti-Aliasing, and Ray Reconstruction will work with all RTX GPUs in DLSS 3.5 games, Frame Generation is exclusive to RTX 40-series graphics cards like the RTX 4070.

Ray Reconstruction, like other DLSS features, utilizes AI powered by the Tensor cores on RTX graphics cards. However, its functioning is more intricate compared to other DLSS features we’ve seen so far.

How DLSS 3.5 Works


The primary goal of Ray Reconstruction is to address the loss of detail during the denoising process when ray tracing is enabled. In ray-traced games, only a limited number of samples are taken per pixel, resulting in noise similar to grainy film or noisy digital images. Denoising is the solution to clean up the image after sufficient samples have been taken.

Nvidia explains that denoising, while effective, has its limitations. Spatial denoising, which uses nearby pixels to fill in missing details and enhance the image, fails to capture fine details. This often leads to a blurring effect in games with ray tracing.

Temporal denoising, another method that compares two frames to fill in missing details, can cause instability. Fast-moving objects may exhibit ghosting, while stationary objects may have shimmering reflections as the pool of samples changes.

Ray Reconstruction aims to overcome these limitations and make ray tracing more accurate by bypassing the denoising process altogether. It functions as an AI-powered denoiser that adapts to different lighting conditions, such as hard or soft reflections, global illumination, and shadows.

What sets Ray Reconstruction apart is its ability to recognize the scene it is processing. Nvidia claims that this feature was trained on five times more data than DLSS 3, enabling it to utilize more data from the game engine and maintain the necessary detail for high-quality upscaling.

Nvidia has showcased the capabilities of Ray Reconstruction through demos of Portal RTX and Cyberpunk 2077. These demonstrations highlight the significantly improved lighting conditions with Ray Reconstruction enabled. Remarkably, Ray Reconstruction does not impact performance negatively and can even slightly improve performance in demanding denoising scenarios.

While we eagerly await the release of DLSS 3.5 this fall, only time will tell if Ray Reconstruction lives up to the hype. If Nvidia’s claims hold true, this update has the potential to revolutionize ray tracing technology. Until then, we can only anticipate the exciting possibilities that DLSS 3.5 will bring to the gaming world.