Bethesda has committed to “further adjusting the lighting and final effect” of Nvidia’s controversial DLSS 5 visuals on Starfield, after fans strongly criticized the game’s AI-like faces when the controversial new technology was applied.
Countless fans have left comments slamming the “AI-generated slop” visuals demonstrated by Nvidia in Starfield and a range of other blockbuster games, including Oblivion Remastered, Assassin’s Creed Shadows, and Resident Evil Requiem, calling it an “AI filter” that players “don’t want.“
Nvidia announced DLSS 5 yesterday, and dubbed “the GPT moment for graphics.” (For more about DLSS and why it matters to gaming, check out IGN’s handy guide).
As part of the announcement, Nvidia acknowledged the challenges that AI video models have faced in the past, but claimed its solution was to tie the model to the color and motion vector data taken from the game engine – just like with Frame Generation – to keep the output grounded in the original scene.
The model is then trained to “end-to-end understand complex scene semantics such as characters, hair, fabric and translucent skin, along with environmental lighting conditions like front-lit, back-lit or overcast – all by analyzing a single frame,” then use that information to generate images.
However, despite comments from Bethesda boss Todd Howard hailing DLSS 5’s impact as “amazing,” the results have failed to impress some fans, who have hit back at the photorealistic lighting and facial details, forcing Bethesda to seemingly walk back some of that excitement soon afterwards.
“Appreciate your excitement and analysis of the new DLSS 5 lighting here. This is a very early look, and our art teams will be further adjusting the lighting and final effect to look the way we think works best for each game,” the company said. It also stressed that “this will all be under our artists’ control, and totally optional for players.”
The latter comment stems from complaints from players that DLSS 5 essentially overlays AI-generated graphics over the original, with some calling it “an insult to your own artists” and pleading, “please don’t subject your art teams to this.”
DLSS debuted back in 2018 with the RTX 2080, and it was initially just the Deep Learning Super Sampling that it’s named after. The idea was to take a game, render it at a lower resolution, and then upscale it using AI back to your native resolution. DLSS has evolved a lot in the years since, adding features like Frame Generation, Reflex, and now its AI model that injects new lighting and materials to make the scene more “photorealistic.”
Opinion on the use of AI in games continues to divide studios and their fans, with some vehemently against its use, while others claim it’s an inevitable part of the future. Rockstar co-founder and former Grand Theft Auto writer Dan Houser recently likened AI to mad cow disease, but the CEO of Genvid — the company behind choose-your-own-adventure interactive series like Silent Hill Ascension — has claimed “consumers generally do not care” about generative AI in games, and stated that: “Gen Z loves AI slop.”
Vikki Blake is a reporter for IGN, as well as a critic, columnist, and consultant with 15+ years experience working with some of the world’s biggest gaming sites and publications. She’s also a Guardian, Spartan, Silent Hillian, Legend, and perpetually High Chaos. Find her at BlueSky.
