It’s Still A Little Early To Say, But Nvidia’s DLSS 5 Is More Uncanny Than Impressive
March 17, 2026
NVIDIA is ready to change the video game industry forever with DLSS 5, their new mainline rendering technology that they unveiled Monday afternoon. DLSS 5 will feature the most comprehensive use of AI in computer graphics yet, leveraging the power of neural rendering to provide photorealism well beyond what most games are capable of on their own. NVIDIA still has some kinks to work out of the technology, but DLSS 5 will come sooner rather than later, currently aiming for a Fall 2026 release.
What NVIDIA is doing with DLSS 5 can, in earnest, be considered revolutionary. Neural rendering has been a focus of NVIDIA for some time now, but in no way were there any indications that the technology would be viable for public use this quickly. To be fair, it still actually isn’t, as NVIDIA used two RTX 5090s to run their DLSS 5 demos at 4K. Looking ahead, however, DLSS 5 is supposed to be operable on a lone graphics card, especially by the time it has been publicly released.
To get into the specifics of how DLSS 5 works, it takes every frame as an input, reads the game’s colors and motion vectors, interprets the semantics of the 3D scene, and then finally generates lighting and material enhancements that remain anchored and stable as you play. All of this happens in real time, with NVIDIA promising smooth and consistent visuals.
The ability to add further complexity to objects at a geomety level makes DLSS 5 very enticing. Enhanced lighting is not something that should be underestimated. Hair, clothing, foliage, and skin, anything that ordinarily interacts with light in an unconvincing manner stands out significantly more with DLSS 5.
Unfortunately, the technology is very AI-forward when it comes to adding detail to human characters. Many faces, as a result, look uncanny. I would argue that some even look unprofessional with how apparently artificial they are. Grace Ashcroft in Resident Evil Requiem barely even resembles herself with DLSS 5 on. I just don’t understand how NVIDIA can look at all the different comparison images and actually be proud of their work. In concept, the technology is fantastic for the industry, but what NVIDIA aims to deliver is something that most players want no part of.
I understand the desire to create games that feel as photorealistic as possible, but DLSS 5 only accomplishes that in a very limited sense. Sure, most of the DLSS 5 demos manage to look more realistic, but they also look more like something generated by AI. Some won’t mind that, but it’s incredibly off-putting for a lot of people. Immersion depends a lot on the visuals, and I don’t see how DLSS 5 can be considered an improvement if it makes things look fake.
NVIDIA has every right to feel pissed about the public’s immediate reaction to DLSS 5, but let’s be real, the company has consistently been out of touch with its consumer base when it comes to AI. People don’t care if DLSS 5 is or isn’t using generative AI for post-production effects, what people care about is that it ultimately ends up looking AI-generated. Even if it’s the developers themselves who are in charge of refining DLSS 5 to their specific liking, the technology innately feels like a bastardization of art.
Sorry, NVIDIA, but DLSS 5 takes things too far. I see great promise in neural rendering, but, as of right now, it seems too invasive and uncanny. Maybe investors don’t mind that, but gamers most certainly do.
