DLSS 4.5 Ray Reconstruction Set for August Launch, Bringing Significant Ray Tracing Quality Improvements

Nvidia announced the next-generation neural rendering technology “DLSS 4.5 Ray Reconstruction” at Computex 2026. Scheduled for release in August 2026, this update revolutionizes the quality of denoising in ray tracing and path tracing, potentially redefining the graphical fidelity of PC gaming.

The Culmination of DLSS Evolution in 2026

In 2026, Nvidia has been steadily advancing its DLSS (Deep Learning Super Sampling) technology. The first major update of the year was DLSS 4.5 Super Resolution, unveiled at CES 2026. This update introduced a more advanced and computationally intensive transformer-based AI architecture, enabling superior image quality even at low input resolutions.

Following this, DLSS 4.5 Multi Frame Generation was released, featuring 5x and 6x frame generation and a dynamic mode that optimizes frame rates for smoother gameplay. Now, the DLSS 4.5 Ray Reconstruction update has been revealed, marking the final piece in this year’s series of innovations.

What is DLSS Ray Reconstruction?

To understand DLSS Ray Reconstruction, it’s essential to revisit its foundational mechanism. Traditional ray tracing relies on manually tuned denoising filters. However, these filters struggle with accurately predicting pixels in areas where light rays haven’t been cast.

DLSS Ray Reconstruction replaces these manual denoisers with a neural rendering model. It estimates the appearance of pixels in noisy regions where light rays are absent while simultaneously upscaling the resulting image to high resolution. By leveraging temporal and spatial input data provided by the game engine, this technology generates sharp, stable, and high-fidelity output images.

Three Key Advances Enabled by

Second-Generation Transformers

The DLSS 4.5 Ray Reconstruction update achieves its advancements through three primary improvements:

1. Enhanced Transformer Architecture: The adoption of an improved transformer architecture significantly increases the volume of input data that can be processed within the same computational budget, compared to the previous generation. The number of parameters has also been expanded, enabling more accurate predictions of complex lighting patterns and reflections.

2. Improved Spatial Awareness: Borrowing advancements from DLSS 4.5 Super Resolution, the precision of input data processing has been enhanced. This improvement allows for more accurate lighting, more stable visuals, and clearer motion representation.

3. Expanded Training Dataset: By training the model on a more diverse range of gaming scenarios, the efficiency of utilizing input data from games has improved. Additionally, developers now have more granular control over temporal accumulation behavior, offering increased flexibility to fine-tune final image quality.

Full Compatibility Across All RTX Generations

A particularly noteworthy aspect of this update is its compatibility. Unlike DLSS 4.5 Super Resolution, which faced significant performance drops on older hardware, DLSS 4.5 Ray Reconstruction maintains 100% compatibility with all GeForce RTX GPUs.

This includes earlier RTX series GPUs like the RTX 20 series and RTX 30 series, which were the pioneers of ray tracing hardware. For users of these older GPUs, DLSS 4.5 Ray Reconstruction represents one of the few latest updates they can fully benefit from.

This decision likely stems from the observation that the computational load of Ray Reconstruction is less intensive than that of Super Resolution, and that ray tracing itself is supported by older GPU architectures. Nvidia’s commitment to supporting its extensive user base of legacy hardware highlights its dedication to maintaining a healthy ecosystem.

Real-World Game Improvements

Nvidia showcased the impact of DLSS 4.5 Ray Reconstruction on visual quality using Capcom’s “Pragmata” as a demonstration. In the game, laser traps were rendered with more dynamic detail, and issues like lingering artifacts when lasers were deactivated were eliminated.

These improvements are tangible benefits stemming from enhanced lighting accuracy and responsiveness. They directly address the longstanding challenge of making ray tracing not just visually stunning but also stable, a feat achieved through the power of neural networks.

Industry Implications and Future Prospects

The debut of DLSS 4.5 Ray Reconstruction signifies the growing role of neural rendering in real-time graphics. The shift from manually tuned denoisers to machine learning-based models has become an irreversible trend.

By integrating cutting-edge AI technology like transformer architecture into its graphics pipeline, Nvidia is further solidifying its technical edge over competing technologies such as AMD’s FSR and Intel’s XeSS.

As the August launch approaches, key questions remain: which game titles will first support this new technology, and how well will older GPUs handle the update? These will be critical areas to watch in the coming months.