Nvidia, a significant player in the PC graphics arena, recently made a bold claim: the future of gaming graphics lies in AI rendering, not in traditional raster rendering. The statement came from Nvidia’s Vice President of Applied Deep Learning Research, Bryan Catanzaro, during a round table discussion with Digital Foundry about the company’s new DLSS 3.5 Ray Reconstruction technology.

Catanzaro’s Perspective

Catanzaro not only emphasized the importance of AI rendering but also highlighted how it outperforms raster rendering. For example, he compared the technology in Big Hero 6, reportedly the first CGI movie to fully utilize path tracing, with that in the video game Cyberpunk 2077. Catanzaro found it remarkable that such advanced technology is now running in real-time at 4K resolutions.

He further explained that traditional methods, often referred to as “brute force,” which re-render every frame 120 times a second at 2160p, are wasteful. According to Catanzaro, these methods ignore many correlations that exist in the output of any rendering process. Therefore, it is possible to be smarter about this, reusing computation and consequently delivering superior image quality benefits.

The Evolution of DLSS

Deep Learning Super Sampling (DLSS) has gone through several iterations. Starting with DLSS 1.0, which was not well-received, to the current DLSS 3.5, each version has built upon the lessons from the previous ones. DLSS 3.5 now introduces Ray Reconstruction, which according to Catanzaro, makes scenes in games like Cyberpunk 2077 look even better than they do with native rendering. The AI can make smarter decisions about rendering, which improves both performance and realism.

Implications for the Industry

One of the challenges that arise from Nvidia’s focus on AI-rendering is how to compare it with other GPUs, especially those from competitors like AMD. Traditional raster rendering or even conventional ray tracing allows for direct comparisons between AMD and Nvidia GPUs in terms of visual quality output. However, an Nvidia GPU running Ray Reconstruction generates visuals that are fundamentally different from any GPU that doesn’t support the feature.

This leads to another issue: AMD, Nvidia’s main competitor, is already lagging behind in this space. AMD’s FidelityFX Super Resolution (FSR) technology has been playing catch-up, and it still doesn’t offer features comparable to Nvidia’s DLSS.

What’s Next for AMD?

AMD faces a significant challenge. It has to respond not just to Nvidia’s DLSS but also to the new Ray Reconstruction feature. AMD has been attempting to close the gap with features like Fluid Motion Frames in its upcoming FSR 3, which is still under development. Whether AMD can catch up in this rapidly evolving field remains to be seen.

Conclusion

Nvidia’s assertion that the future of gaming graphics lies in AI rendering, particularly DLSS 3.5, has broad implications. The technology has not only continued to evolve but has also posed challenges for traditional methods of graphical comparison and for competitors like AMD. As AI rendering technologies continue to advance, they could indeed become the new standard in the gaming industry. Therefore, Nvidia’s vision of the future appears to be taking a strong lead over competitors still focused on producing more powerful hardware.

Also Read: