Nvidia’s DLSS 5 is like motion smoothing for video games, but worse

Foto: The Verge AI
Nvidia presented DLSS 5 — a new graphics upscaling technology that it called "the biggest breakthrough in computer graphics since ray tracing was introduced in 2018". Reality proved disappointing. Instead of improvement, the AI algorithm transforms character faces into something resembling "AI slop" — a homogeneous, characterless appearance. Resident Evil Requiem protagonist Grace gained features characteristic of artificial images: unnaturally smooth skin, uniform facial features and synthetically-looking hair. Characters from Hogwarts Legacy and even real footballer Virgil van Dijk suffered similarly. The problem extends beyond aesthetics. Games are already grappling with a wave of layoffs and studio closures, and the potential for AI to replace artists' work generates justified industry concerns. The technology will be available in autumn and will be optional, but the fact that giants like Nvidia, Bethesda and Capcom support it signals a dangerous direction. DLSS 5 is another symptom of AI aesthetics infiltrating mainstream entertainment — from social media to video games.
Nvidia announced its latest upscaling technology yesterday — DLSS 5 — describing it as "the most significant breakthrough in computer graphics since the debut of real-time ray tracing in 2018". It sounds impressive, until you actually see it. The reality is decidedly less glamorous: the faces of characters from demonstration materials look like they were pulled straight from an AI generator. Grace from Resident Evil Requiem underwent a metamorphosis worthy of Tilly Norwood films, and students from Hogwarts Legacy look like they've been filtered through Instagram. Even Virgil van Dijk, a very real and famous footballer, had distorted features and became someone completely different. This is not a breakthrough — it's a warning.
Nvidia claims the technology "saturates pixels with photorealistic lighting and materials", but everyone focused on something completely different: that all faces suddenly look identical. They have the same smooth skin, uniform features, gleaming eyes, perfectly formed hair, and that characteristic HDR lighting that emphasizes every contour. It's exactly the same aesthetic signature that every AI-generated artwork on the internet carries — from YouTube thumbnails to Instagram posts. DLSS 5 is simply the first so visible example of how this homogeneous, dull algorithmic look is seeping into video games.
What's particularly frustrating is the fact that Nvidia and its partners seem to be completely aware of what they're doing. Todd Howard from Bethesda talks about how DLSS 5 allows "artistic style and details to shine without the limitations of traditional real-time rendering". Jun Takeuchi from Capcom claims it's "an important step forward in raising visual fidelity". Words that sound like modern technology, but in practice mean replacing carefully designed characters with generic AI versions. It's exactly like motion smoothing on televisions — a feature nobody asked for, but one that changes the experience in a way many people find degrading.
Read also
Why it looks so bad — and why everyone knows it
DLSS 5 is not a simple filter. It's an advanced neural network that analyzes images and generates new pixels based on what it has "learned" from datasets containing millions of faces. The problem is that these datasets are inherently homogeneous — they contain averaged, typical features that the algorithm then reproduces in every new image. The result? Everyone looks like everyone. Nobody is distinctive, nobody is memorable, everyone is interchangeable.
The artistic decisions of game designers — those subtle differences in eye shape, jawline, facial expression that make characters recognizable — are ignored. It's as if an artist spent months on a portrait, and then someone blurred it and said: "Don't worry, the algorithm will fix it". Of course the game looks worse. Of course the faces look weird. Of course everyone is upset.
Bethesda quickly issued a statement that what we're seeing is a "very early preview" and that art teams will continue working on lighting and effects. It sounds like an apology wrapped in technical jargon. But even if the final version is somewhat better, the fundamental problem remains: this technology changes what games can show. It changes how characters can look. And it changes it toward greater homogeneity, not less.
Where we've seen this before — and why it should concern us
AI-generated faces didn't appear suddenly. Over the last few years we've watched as this aesthetic "AI slop" (as internet users remarkably aptly named it) has gradually seeped into mainstream media. YouTube thumbnails are full of AI-generated faces. Brand marketing campaigns use AI models. Some films are already experimenting with AI-generated actors. And now video games.
Interestingly, the gaming industry was one of the last bastions where artists still had significant control over how characters looked. Character designers, concept artists, animators — these were the people who decided every detail. These people were proud of their work. Now Nvidia proposes replacing that work with an algorithm that doesn't understand nuance, character, or artistic intent.
This is particularly painful in the context of what's happening in the gaming industry. The last two years have brought a wave of layoffs and studio closures. Thousands of artists, programmers, and designers have lost their jobs. And now they see how their work — the work they did for months — could be replaced by a tool that costs a fraction of what it costs to employ a human. This is not a technical breakthrough. This is a threat to an entire category of professionals.
Indie developers are already rebelling — but is it enough?
The indie response to DLSS 5 was quick and decisive. Memes, critical statements, and some even started using "AI free" as a marketing label. This is not an accident — it's a conscious choice. Independent developers understand that their player, their audience, wants something authentic, something created by people with a vision, not by an algorithm with statistics.
But here's the problem: indie developers are not the ones who decide the future of the industry. Todd Howard from Bethesda, Jun Takeuchi from Capcom — these are the people who create games that millions of people play. These are the ones who have influence. And they decided that DLSS 5 is the future. Of course, Howard said it's a "very early preview", but that's exactly the kind of cautious language we hear every time a big corporation wants to get us used to something we know we won't like.
In Poland, the situation is somewhat different — the Polish gaming market is dominated by indies and mid-sized studios that have traditionally cared more about artistic quality than California giants. Polish games like Cyberpunk or Witcher have always been known for characters that looked like real people, not AI-generated mush. If DLSS 5 becomes the standard, it would mean the end of such an approach to character design.
Motion smoothing for games — but worse, because it changes reality
Andrew Webster from The Verge aptly compared DLSS 5 to motion smoothing on televisions — that feature practically nobody wants, but TV manufacturers keep adding because they can. Motion smoothing interpolates frames to make a film look "smoother", but in reality it changes the original artistic intent. The director wanted the film to look a certain way, and the technology changes that.
But DLSS 5 is even worse. Motion smoothing only changes the speed at which things move. DLSS 5 changes how things look. It changes faces. It changes character characteristics. This is not a subtle change — it's a complete visual transformation that's hard to turn off or ignore, especially if the game uses it by default.
Nvidia claims DLSS 5 will be optional, but we know how that usually works in practice. When a feature is enabled by default and makes a game "faster" or "better" (at least theoretically), most players won't turn it off. It will be the default experience. It will be what everyone sees.
Why this is happening now — and who profits
Nvidia makes billions on graphics cards. AI is a trend. The combination of these two things means Nvidia has enormous motivation to include AI in every possible aspect of gaming. DLSS 5 is not technology created because games needed it. It's technology created because Nvidia can sell it.
And why do big studios support it? Well, partly because Nvidia has enormous influence in the industry. If you want your game to look good on GeForce cards, you have to work with Nvidia. That's the economics of the industry. Partly also because big studios are always interested in automating expensive processes — and character design is expensive.
Interestingly, Jensen Huang, Nvidia's CEO, when asked about criticism of DLSS 5, said: "Well, first of all they're wrong". That's the classic response of someone who knows they're doing something unpopular but has enough power to ignore it. He doesn't try to explain the technology. He doesn't try to show why it's good for games. He just says everyone is wrong. That's the arrogance of power.
The future of games — will they look like Instagram?
If DLSS 5 catches on, if it's used in major games, if it becomes the standard — what does that mean for the future of games? It means all characters will look similar. It means character will be blended into algorithmic mush. It means games will look like everything else — like Instagram, like YouTube, like every other platform that's already been taken over by AI-generated content.
Games have always been a medium where artists had relatively significant control. You could create a character that looked exactly how you wanted. You could give them distinctive features, quirks, things that make them memorable. DLSS 5 changes that. It changes games toward greater standardization, greater homogeneity, greater boredom.
And that's a problem that will deepen. If DLSS 5 catches on, Nvidia will have motivation to make it better — but better in the sense of "more efficient" and "more widespread", not "more artistic" and "more distinctive". The algorithm won't learn how to create more interesting faces. It will learn how to create more average faces, faster.
What it means for Polish players and creators
Polish video games have their own identity. Cyberpunk 2077 — despite all the problems at launch — had characters that looked like real people. The Witcher had distinctive, memorable characters. The Polish gaming industry has always focused on authenticity and artistic vision. DLSS 5 would threaten that approach.
If Polish studios want to compete on global markets, they'll feel pressure to use the same technology as Bethesda or Capcom. But that would mean Polish games would look like every other game — like everything else in the world that AI has taken over. That would be a loss.
Fortunately, the Polish gaming industry is independent enough and strong enough to say "no". They can decide that the faces of characters in their games will be designed by artists, not generated by algorithms. They can even make it a selling point — "a game created by people, for people, without AI". That would be authentic. That would be memorable.
But it will require a conscious decision. It will require saying "no" to Nvidia, "no" to market pressure, "no" to promises of faster renders and lower costs. It will require faith that artistic vision is more important than efficiency. And that's something they'll have to fight for.
DLSS 5 is testing the boundaries. Nvidia is watching how far it can go in replacing human work with algorithms before people rebel. Indie developers have already rebelled. But will enough people pay attention? Will enough players care that faces look like AI-generated mush? Will they just play because the game is faster? These are the questions whose answers will shape the future of gaming for the next ten years.









