By Nelson Schneider - 03/22/26 at 03:20 PM CT
It seems that the massive AI investments are starting to bear their first fruits in the realm of videogaming. nVidia recently revealed the brand-new, bleeding-edge 5.0 version of their DLSS upscaling software. DLSS has been around for a few years, mostly as a frame-generator or resolution upscaler, with functionality that “fudges” game visuals on low-end PCs or in games set to “low” in order to stabilize frame-rates and smooth-out low-end visuals on large, Ultra-High-Definition displays. DLSS5, however, aims to use AI for much deeper image analysis, allowing it to enhance the lighting and textures in a game to be closer to “photorealistic” than they actually are.
Unsurprisingly, in our idiotically-polarized world, the Internet has already drawn a line and chosen sides in a civil war over whether or not the results of DLSS5 look good or not.
For the record: They DO look good!
We have seen several examples of DLSS5 in action that were shown off by nVidia, and older games with notably ugly character models (like, say, “Oblivion”) look so much better, it’s not even funny. It’s mindboggling that the original ugly character models and poor environmental lighting ever passed muster for being “photorealistic” in the first place.
On the other side of the line, though, there are a number of poor arguments that are largely based on ideology rather than observation. Some people are *boo-hoo*-ing, *sob-sob*-ing over the idea that the artists who work on games will have their “original visions” destroyed by this upscaler – as if aiming for “photorealistic” and missing is an actual artistic choice. Some people claim that the results of DLSS5 look “yassified” or like Instagram filters, which must be the type of thing one can only notice when one is “Terminally Online.” Then there are the people who are best described as “AI Vegans,” in that they are vehemently opposed to any and all AI usage for any reason, and thus wouldn’t admit that DLSS5 looks good unless they were tricked by a series of blind comparisons. These same AI Vegans are the types who will immediately reverse their opinions on a game they like if it is discovered to have had any AI involved in its creation.
The only good anti-DLSS5 argument I’ve seen is the idea that developers who want photorealistic visuals in their games won’t have to try as hard. But how many of these developers are actually trying particularly hard as it is? My view on the availability of DLSS5 is that it will help boost the not-quite-photorealistic visuals we’ve had stagnating for nearly 3 Generations now to greater heights, while simultaneously allowing slightly older games to have significantly longer windows in which their photorealistic visuals hold-up to evolving standards. I don’t think the availability of this lighting and texture filtering tech will put videogame artists out of work if they are on teams that want to create stylized, painterly, or otherwise unique visuals. There is – and I think always will be – a huge demand for games with a unique look and style of their own that blends with their other elements to create a unique identity. Unique looking games stand out from the masses of wannabe photorealistic games already, and the ability for photorealistic games to age a bit more gracefully won’t change any of that.
Regardless, the topic of AI upscaling and other AI-powered graphical rendering tools will likely remain divisive for the foreseeable future. For now, we can all agree that the memes about DLSS5 are pretty great – even though the less-farcical examples are kind of a double-edged sword.
Comments
Nick - wrote on 03/22/26 at 03:49 PM CT
I may be a retro gamer purist when it comes to my NES games, but I also don't have any problems with things like 3dSen. Same goes for DLSS5. People can simply choose whether they want to use it or not. I don't understand why views are so divided. It's amazing technology.
