Nvidia Stands Firm on DLSS 5, Defending Its Integration of Generative AI in Gaming
Earlier this week, Nvidia unveiled DLSS 5, a cutting-edge generative AI upscaling technology that aims to help developers achieve their artistic visions, even if it means giving popular characters some bizarre looks, like Grace from Resident Evil Requiem. This has sparked significant backlash from the gaming community.
During the recent GPU Technology Conference, CEO Jensen Huang addressed the criticism when questioned by an outlet. He firmly stated, “Well, first of all, they’re completely wrong,” emphasizing that developers retain full control over the technology.
“DLSS 5 combines control over geometry, textures, and all aspects of the game with generative AI,” Huang explained. “It’s not just post-processing; it’s generative control at the geometry level.”
He asserted that developers, including those at Capcom and Bethesda, can “fine-tune the generative AI” to align with their artistic style. While DLSS 5 introduces a new layer of AI-driven fidelity, it doesn’t strip away “artistic control.”
When a reporter raised concerns that DLSS 5 might lead to games appearing worse or more uniform, Huang reiterated, “They’re completely wrong… you can fine-tune the generative AI to your artistic style… if you want cartoon, toon shader, made of glass… it’s up…”
Huang’s comments seem to address the widespread interpretation of the technology as a “slop filter,” which adds an uncanny hyper-fidelity to games based on generic training models rather than the original artists’ intentions. This perception is partly fueled by demo videos showcasing before-and-after comparisons that make characters in games like Starfield and Hogwarts Legacy look like exaggerated versions of themselves. Huang insists that the adjustments are made at the developer level, not through a simple post-processing algorithm.
However, it remains to be seen if this reasoning will sway fans who are currently unsettled by the early results of the technology. The question looms: is DLSS 5 merely a cheap slop-face filter, or is it a genuine integration of generative AI into the core geometry that shapes a game’s appearance and feel?
While DLSS 5 may enhance more generic elements like environmental lighting, the tools inherently possess a degree of randomness and unpredictability. The primary concern among gamers so far is whether Nvidia will provide an option to disable this feature entirely.