Nvidia casts DLSS 5 as a major leap in AI-assisted graphics, while critics see a tool that could overwrite a game’s intended look. Its first release offers a more restrained test of that divide—but only for players with powerful new hardware.

The argument began with Nvidia’s March reveal, when the company called DLSS 5 “the company’s most significant breakthrough in computer graphics since the debut of real-time ray tracing in 2018.”¹ The technology uses AI to enhance game imagery, with developers able to choose how aggressively it affects individual elements of a scene.

But the early demonstrations quickly became the story. Examples were ridiculed for dramatically changing characters’ appearances, including what critics dubbed the “yassification” of Resident Evil Requiem protagonist Grace Ashcroft. The concern was not simply sharper lighting or richer textures; it was that an AI layer might turn authored faces into something closer to glossy caricatures.¹

On September 3, Nvidia will put DLSS 5 into its first official game, NBA 2K27, at 9 p.m. PT on PC. For now, the feature is officially limited to RTX 50-series desktop and laptop GPUs, plus GeForce Now. Games highlighted in March—including Requiem, EA Sports FC, Starfield and Hogwarts Legacy—still do not support it, and Nvidia has not provided a timetable for broader adoption.¹

The new basketball-game examples appear less radical: added shadows and textures make players look more realistic rather than visibly remaking their faces. Nvidia says creators can target details such as foliage, food or silverware while leaving characters untouched.¹

Yet the hardware trade-off remains severe. Nvidia estimates a 50 to 60 percent performance hit; its 1080p RTX 5060 example also relies on a top-tier CPU and 6x frame generation—one rendered frame for every six displayed. DLSS 5’s debut, then, is less a victory lap than a tightly controlled first audition.¹