Moving from one console generation to the next, we used to see huge graphical leaps. Remember those days? We thought that kind of epic bump was a thing of the past.
Now, we’re not so sure. Nvidia’s newly announced DLSS 5 may be the kind of single-generation graphics jump that gamers haven’t seen in some time. To be sure, it’s already raising hackles and courting controversy—so much so that CEO Jensen Huang had to give it a quick defense out of the gate in the court of public opinion. It goes well beyond the image-upscaling efforts of earlier versions of DLSS, and beyond the frame-generation efforts of later ones.
At Nvidia’s GTC event last week, I was given the chance to try a short series of demos of DLSS 5. Yes, it’s early. No, it’s not realistically optimized for existing hardware (yet). Even so, the technology wowed me several times, making me wonder: If Nvidia can make it work, is this the future of gaming graphics?
DLSS 5: The Promise, and the Controversy
DLSS is best known for using AI models to increase a PC game’s frame rate for smoother gameplay, whether via upscaling (rendering a game at a lower resolution, then upticking it to a higher one) or frame generation (using AI to splice in additional frames between classically rendered ones). But last week, Nvidia introduced DLSS 5, which uses a “neural rendering” model to add photorealistic effects. It’s a whole new layer to the DLSS game.
The company has been quietly developing the technology for over three years. The result can make game characters feel startlingly alive by injecting even more shadows, textures, and definition over faces, clothes, and environments, creating a new sense of depth.
DLSS 5 on and off for the game The Elder Scrolls IV: Oblivion Remastered. (Credit: PCMag)
But despite the improvements, Nvidia’s DLSS 5 announcement has already drawn some backlash over concerns that the GPU maker is merely adding an Instagram-like image filter to game characters’ faces. Another criticism is that DLSS 5 is acting like an AI slop generator and allegedly forcing AI imagery on top of carefully crafted characters created by game developers.
Those worries were on my mind as Nvidia gave me a closer look at DLSS 5 at GTC. But as I saw the technology in action, it also became clear to me that DLSS 5 could take computer graphics to a whole new level.
Eyes On With the Next DLSS
One thing is clear: DLSS 5 is no simple face filter. Video-game rocks and stones suddenly looked like rocks from real-life. The same was true of trees, water, a medieval castle, the interior of Hogwarts School, and even an espresso machine: DLSS 5 added a new level of photorealism that traditional game rendering had struggled to achieve.
Another “wow” moment came when DLSS 5 was activated during a demo of The Elder Scrolls IV: Oblivion Remastered. Characters originally modeled two decades ago instantly began to look more like real people; their odd “potato” faces had vanished, replaced by fully fleshed-out faces with photorealistic hair, skin, eyes, and clothes. Sure, I was not seeing the ugly, but charming, facial models from before. But in return, DLSS 5 unlocked a new level of immersion.
Get Our Best Stories!
Your Daily Dose of Our Top Tech News
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
In Assassin’s Creed Shadows, DLSS 5 turned the game’s lush forests into something that seemed organic. The effects added even more variety to the light and texture of the dense foliage and rocks, making the depicted landscape indistinguishable from real-world photography.
Crucially, the experience didn’t feel fake or forced. Nor did it act like a conventional AI image filter, which can alter a person’s face, but in a clumsy, heavy-handed way that masks over the original.
Nvidia points out that DLSS 5’s neural rendering is designed to understand 3D characters and objects, including colors, hair, fabric, skin, and movement, as well as the surrounding environment. In other words, the technology is supposed to preserve game models before enhancing them.
The Power Question: What Will It Take to Run DLSS 5?
That all said, DLSS 5 remains a work-in-progress. In fact, Nvidia demoed the technology using not one but two GeForce RTX 5090 graphics cards—each starting at $1,999, though actual pricing has since risen far higher due to the ongoing memory shortage.
Recommended by Our Editors
The goal is to optimize DLSS 5 so it can run on a single GPU. But in the demo I saw at GTC, one RTX 5090 card was used to render the game, while the other added neural rendering effects. That suggests DLSS 5 will need major tweaks to make it practical for the company’s graphics cards. Nvidia will need to move quickly on its optimizations, since it plans to launch the technology this fall. We wouldn’t be surprised if the initial launch is limited in scope.
Other big unknowns include whether DLSS 5 will introduce a major performance hit—a potential irony, considering that DLSS was developed to boost frame rates on sometimes underpowered hardware. How DLSS performs over an entire game is another major question. I was only able to briefly try out the feature in both Hogwarts Legacy and Oblivion, and my hands-on session was limited to walking around a single scene, rather than battling enemies or throwing magic spells.
Understandably, some gamers may be skeptical or even alarmed, given the ethical issues and legal battles surrounding generative AI. At the same time, the PC market is reeling from an AI-driven memory shortage that risks undercutting DLSS 5 by inflating the cost of admission to buy Nvidia GPUs.
But putting all that aside, I have to say DLSS 5 displayed the most realistic gaming graphics I’ve ever seen—and I’m looking forward to experiencing more. Once you see DLSS 5 in action, it’s hard to deny the potential it holds.
In the meantime, Nvidia is already responding to some of the backlash, explaining that game developers will have full artistic control over DLSS 5 and can fine-tune the model to their liking. Some major developers, including Bethesda, Capcom, and Ubisoft, are already on board and preparing to support the technology in their own games.
About Our Expert
Michael Kan
Senior Reporter
Experience
I’ve been a journalist for over 15 years. I got my start as a schools and cities reporter in Kansas City and joined PCMag in 2017, where I cover satellite internet services, cybersecurity, PC hardware, and more. I’m currently based in San Francisco, but previously spent over five years in China, covering the country’s technology sector.
Since 2020, I’ve covered the launch and explosive growth of SpaceX’s Starlink satellite internet service, writing 600+ stories on availability and feature launches, but also the regulatory battles over the expansion of satellite constellations, fights with rival providers like AST SpaceMobile and Amazon, and the effort to expand into satellite-based mobile service. I’ve combed through FCC filings for the latest news and driven to remote corners of California to test Starlink’s cellular service.
I also cover cyber threats, from ransomware gangs to the emergence of AI-based malware. Earlier this year, the FTC forced Avast to pay consumers $16.5 million for secretly harvesting and selling their personal information to third-party clients, as revealed in my joint investigation with Motherboard.
I also cover the PC graphics card market. Pandemic-era shortages led me to camp out in front of a Best Buy to get an RTX 3000. I’m now following how President Trump’s tariffs will affect the industry. I’m always eager to learn more, so please jump in the comments with feedback and send me tips.