Catanzaro, Nvidia’s VP of Applied Deep Learning Research, asserted that native resolution gaming no longer represents the pinnacle of graphical fidelity.
In a recent roundtable discussion video by Digital Foundry, Nvidia’s Bryan Catanzaro, CD Projekt Red’s Jakub Knapik, and Jacob Freeman delved into the transformative impact of Nvidia’s DLSS 3.5 technology in Cyberpunk 2077. They explored the broader advantages of AI-generated upscaling and rendering, signaling a shift away from traditional native resolution gaming.
Catanzaro, Nvidia’s VP of Applied Deep Learning Research, asserted that native resolution gaming no longer represents the pinnacle of graphical fidelity. He emphasized that the gaming graphics industry is steering towards greater reliance on AI-driven image reconstruction and graphics rendering in the future.
This revelation came in response to a query from Pedro Valadas of PCMR, inquiring about Nvidia’s prioritization of native resolution performance. Catanzaro underscored that relying solely on brute force for graphics fidelity, a method increasingly hindered by the plateauing of Moore’s Law, is no longer a viable strategy.
Consequently, Catanzaro advocated for the integration of smarter technologies like DLSS to enhance graphics fidelity and counteract the limited generational performance gains observed in today’s graphics hardware.
In the context of Cyberpunk 2077, both Valadas and CD Projekt Red’s Knapik affirmed that achieving full path-tracing in the game would have been unattainable without DLSS technologies. They emphasized DLSS’s role in image upscaling and frame generation, noting how it provided substantial performance headroom for modern GPUs.
This empowered Cyberpunk 2077 to execute real-time, realistic light simulation, resulting in a markedly more lifelike and intricate visual output compared to conventional rendering methods.
Addressing the future of machine-learning in gaming graphics, Catanzaro envisaged a scenario where DLSS and AI could eventually supplant traditional rendering altogether. He emphasized that the industry is realizing the potential for learning complex functions from large datasets with AI, surpassing the efficacy of constructing algorithms from scratch, a process inherent to “traditional rendering techniques.”
Nvidia provided a glimpse into this future at NeurIPS 2018, showcasing a driving demo entirely rendered using AI.
Catanzaro’s remarks suggest that the familiar landscape of gaming may be in its twilight years. With the perceived “death” of Moore’s Law, AI manipulation appears poised to be the driving force behind 3D graphics advancement in the foreseeable future.
This subtly signals Nvidia’s intent to prioritize AI performance in upcoming graphics cards, underpinning AI-driven frame rendering technologies like DLSS. As the industry ushers in this new era, the integration of AI into gaming promises a revolution in visual fidelity and realism.