NVIDIA releases second-generation light reconstruction technology incorporated into DLSS 4.5 feature set

[ Gearbest Technology News]On June 1, NVIDIA announced the launch of its second-generation Ray Reconstruction technology as part of the DLSS 4.5 feature set. Ray reconstruction was first introduced with DLSS 3.5 in 2023, aiming to solve problems such as distortion of ray tracing shadows and reflections and inaccurate details when DLSS 2 super-resolution is turned on. This function introduces an AI reconstruction module into the rendering pipeline to replace traditional temporal and spatial filters to better restore the lighting effects of ray tracing.

NVIDIA

In 2025, NVIDIA will upgrade the early light reconstruction AI model based on convolutional neural network (CNN) to a new Transformer architecture, improving the detailed understanding and temporal consistency of complex scenes. The subsequently released DLSS 4.5 feature set included the fourth-generation Transformer model, but the light reconstruction module was not updated simultaneously at that time. The launch of the second generation of light reconstruction marks the completion of the DLSS 4.5 feature set. The new version adopts the second-generation Transformer model and conducts targeted training based on significantly expanding the amount of training data. It aims to achieve more accurate shadow contours, more natural reflection performance and more stable temporal consistency, and reduce image quality problems such as flicker, noise and ghosting.

In addition, DLSS 4.5 ray reconstruction integrates a more efficient AI denoiser, which can achieve cleaner ray tracing images at lower noise processing costs. The new denoiser combines stronger spatial perception capabilities with a deeper understanding of the spatial relationships between different pixels, object boundaries and lighting changes, helping to improve overall lighting accuracy and local detail performance. In scenes with highly reflective materials, complex shadow overlapping, and multiple light sources, the new version of light reconstruction is expected to significantly reduce artifacts and improve visual stability.

With the second generation of light reconstruction officially joining DLSS 4.5, NVIDIA's DLSS function line is evolving towards a unified AI rendering framework based on Transformer. From the early CNN model to the Transformer model, functions such as light reconstruction and super-resolution gradually rely on the same deep learning architecture and larger-scale training data, making the synergy of different modules better. For game developers, when using the latest DLSS version, they can obtain higher-quality ray tracing output and more stable picture performance with less manual adjustments.