Nvidia? A gaming GPU company?
On Monday, the multitrillion-dollar AI chipmaker announced its latest effort to weave AI advances into video games, but it quickly backfired.
This feature, DLSS 5, is said to be an enhanced version of the deep learning upscaling technology that Nvidia has been offering since 2018. That same year, the company called it “the most significant advance in computer graphics since the debut of real-time ray tracing.” However, the reaction to the shared demonstration footage was overwhelmingly negative.
Gamers and developers were outraged by the announcement, calling it a “sloppiness” and a “betrayal” of the game’s artistic intent. Memes parodying the flashy aesthetics of the AI feature are going viral, contrasting original characters and people with “DLSS 5” images that show the subject in an unrecognizable style. Some people have given it the harsh nickname “sloptracing,” a play on Nvidia’s ray tracing technology.
That reaction is guaranteed. The feature looks more like a beautified Snapchat filter than a blurry image that slightly sharpens it, giving your favorite game’s art style a sophisticated generative AI finish.
The effect is most noticeable when applied to the face. It’s no exaggeration to say that some of the iconic characters that appear in the demo, such as Leon Kennedy from the Resident Evil series, are literally being hunted down.
According to Nvidia’s announcement, DLSS 5 “introduces a real-time neural rendering model that injects photorealistic lighting and materials into pixels.” “It takes each frame’s game color and motion vectors as input and uses an AI model to infuse the scene with photorealistic lighting and materials anchored to the source 3D content.”
The AI model is “trained end-to-end to understand the semantics of complex scenes such as characters, hair, fabrics, and translucent skin.”
Nvidia CEO Jensen Huang spoke passionately about the technology’s impact, calling it gaming’s “GPT moment.”
“DLSS 5 is a GPT moment for graphics, fusing handcrafted rendering with generative AI to deliver a dramatic leap in visual realism while preserving the control artists need for creative expression,” he said in the announcement.
It’s a little hard to accept Huang’s promise to maintain creative expression, but looking at all the examples shared, DLSS 5 dramatically changes the aesthetics of games. More than that, it illustrates how generative AI can uniformly reinforce bland aesthetic standards and default to superior beauty standards. (Grace Ashcroft in the upcoming Resident Evil game has sunken cheeks, stronger cheekbones, and fuller lips.) The game no longer looks like a game, but like any other clip spewed out by a video-generating model and shared in AI circles with captions like “Hollywood is Cooking.”
Nvidia says DLSS 5 is coming this fall, but it appears to be limited to participating games including Resident Evil Requiem, Starfield, Hogwarts Legacy, and Assassin’s Creed Shadows. However, these are major titles and a sign that NVIDIA says its features are supported by some of the industry’s biggest publishers and developers, including Capcom, Bethesda, Ubisoft, and Warner Bros. Games.
Learn more about AI: Unity announces new product uses AI to cook entire games
