Earlier this week, Nvidia unveiled DLSS 5, but rather than iterating and improving on the existing tools to boost performance for lower-end rigs even further, it yassifies characters with an uncanny AI filter.
Considering how demanding DLSS 5 is, which goes against the entire ethos of the technology, it isn’t clear who this is for. The angry mob who caked Aloy in make-up because she had puffy cheeks? Even developers were caught off guard after seeing their games fed through the deep learning super sampling shredder. But Nvidia CEO Jensen Huang insisted that everyone is “completely wrong” about DLSS 5, assuring that it “fuses the controllability of the geometry and textures […] with generative AI” and that it’s not just a filter.
NVIDIA’s DLSS 5 Doesn’t View Video Games As Art, And That’s A Big Problem
The reveal of Nvidia’s DLSS 5 has the internet in uproar.
Except, it is. As reported by IGN, GeForce Evangelist Jacob Freeman said in a statement to YouTuber Daniel Owen that “Yes, DLSS 5 takes a 2D frame plus motion vectors as input” and “is trained end to end to 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.”
Devs Probably Won’t Have As Much Control Over DLSS 5 As Nvidia Promises
In so many words, DLSS 5 looks at a flat image and works from there. It isn’t baked into the actual systems of a game. So, when Nvidia promises that developers will have “detailed controls” to ensure that their art direction stays in-tact, that’s not exactly true.
Considering how the tech works, developers will presumably only be able to make surface-level adjustments, and considering that DLSS 5 is essentially a generative AI makeover — like feeding a screenshot into Nano Banana and saying ‘make this photorealistic’—the results will vary for everyone, anyway.
However, even with all the backlash to DLSS 5, Nvidia remains stalwart in its defense of the technology, and its push evermore into AI tools, so it’s hard to imagine one of the world’s richest companies buckling to the pressure and rolling back its plans. Instead, it seems dead-set on continuing to train and improve the model, streamlining DLSS 5 so that it doesn’t require an expensive supercomputer to run, but I’m still not sure who this is for. People who want hyperrealistic NPCs with Bethesda’s janky animations?
There’s A Website That Will “DLSS 5 Anything” So You Can Find Out How NVIDIA Thinks You Should Look
There’s a website that will apply NVIDIA’s DLSS 5 filter to anything, but beware what you let it see, you almost definitely won’t like the results.
