Nvidia's DLSS 5: Is it Magic or Just More AI Mumbo Jumbo? Jensen Huang Says We're 'Completely Wrong,' Mate!
AI Gaming News Author · Game Rant ·
Nvidia CEO Jensen Huang has fiercely defended the controversial DLSS 5 against gamer criticism of "AI slop" visuals, claiming it offers "generative control at the geometry level" for developers, not just post-processing. While innovative, concerns remain about the tech's current visual output, its role amidst broader AI-driven hardware shortages and price hikes, and its accessibility given the ultra-high-end GPUs currently required to run it.
Right, so you know that feeling when you're messing with a new spell, hoping for a dazzling display, and instead, you accidentally turn your pet griffin into a slightly uncanny-looking, digitally smoothed-out pigeon? Yeah, well, it seems some gamers are feeling a bit like that with Nvidia's latest sorcery, DLSS 5.
Word on the digital street, initially highlighted by Sarah Fields over at Game Rant, is that the gaming community isn't exactly rolling out the red carpet for this new tech. Barely a day after Nvidia spilled the beans on DLSS 5, the complaints started rolling in faster than a rogue fireball. But here's the kicker: Nvidia's CEO, Jensen Huang, has come out swinging, flat-out stating that the critics are “completely wrong.” Blimey, talk about a confidence spell!
For those who've been living under a rock (or perhaps just focusing on actually *playing* games), DLSS – Deep Learning Super Sampling – is Nvidia's proprietary tech for RTX GPUs. Its earlier iterations were pretty darn clever, essentially upscaling lower resolutions to give you better performance in graphically demanding games without your visuals turning into a pixelated mess. It’s like having a wizard clean up your messy scrollwork without losing any of the intricate details. Generally, a boon for frame rates and fidelity.
But DLSS 5? This new iteration, showcased in games like *Resident Evil Requiem*, has ruffled some serious feathers. Gamers, ever the astute observers (and sometimes a touch dramatic, bless 'em), are looking at the 'before and after' comparisons and crying foul. The primary complaint? The altered images are apparently starting to look a bit like “AI slop.” You know, that slightly uncanny, homogenous look that a lot of bog-standard AI-generated images seem to have? The one where everyone looks like they've had the same 'perfect' filter applied?
They're pointing fingers at characters like Grace from *Resident Evil Requiem*, whose face seems to have undergone a bit of a digital beauty treatment. Her makeup looks different, and the most tell-tale sign, apparently, is that her glorious blonde locks suddenly sprout prominent dark roots in the DLSS 5 version. It's like the AI decided she needed a touch-up, but went a bit rogue with the dye job. Makes you wonder if the AI is a former stylist with a grudge, eh?
Now, enter Jensen Huang, the grand architect himself, defending his latest creation. In a chat with Tom Hardware, he basically said, “Right, you lot are barking up the wrong tree.” He elaborated, “DLSS 5 fuses controllability of the geometry and textures and everything about the game with generative AI.” He stressed that developers are still firmly in the driver's seat, able to fine-tune the AI to their heart's content. Apparently, this isn't some post-processing magic trick, but “generative control at the geometry level,” adding generative capability directly to the game's inherent mesh. He even reckons developers could use it for things like crafting a cartoon-style 'toon shader' or even making an entire game look like it's made “of glass.”
Huang insists DLSS 5 is “very different than generative AI,” preferring the term “content-control generative AI.” It's a subtle distinction, but he's painting a picture where this isn't just about making things look *better* in a generic sense, but giving devs the power to *transform* the visuals creatively. Sounds grand, doesn't it? Like giving a wizard new elemental powers instead of just a slightly stronger firebolt.
But here's where my gently skeptical, coffee-fueled brain starts ticking. While the technical explanation about geometry-level control is fascinating – and potentially groundbreaking for artists – it doesn't entirely quash the immediate gut reaction some gamers had to the visual output. If the first impression is “AI slop” or “beauty filter,” then there's clearly a communication gap, or perhaps a demonstration gap, between what the tech *can* do and what it *appears* to do in its raw form. It's like being told a potion will give you super strength, but the first sip makes you slightly wobbly and changes your hair colour. You'd be a bit wary, wouldn't you?
And let's not gloss over the bigger picture, the elephant in the server room, if you will. AI, in general, has been a bit of a sore spot for the gaming community lately. The relentless demand for hardware parts to power AI datacenters has led to a genuine scarcity of RAM and SSDs for us regular folk, driving prices sky-high. Some manufacturers are even bailing on the consumer market entirely to chase those sweet AI company contracts. It's created delays for things like Steam Machine pre-orders, and even future consoles like the Nintendo Switch 2 could see price hikes because their components are just getting more expensive.
When the cost of gaming is potentially skyrocketing, and hardware shortages are becoming the new normal, it's fair dinkum understandable that gamers are feeling a bit sour about *any* AI-powered tech that seems to tie into that ecosystem. It feels like the resources that *should* be going into making our gaming lives better are being siphoned off elsewhere.
Then there's the whole accessibility issue. The initial tech demo for DLSS 5? It was reportedly running on *dual* Nvidia 5090 GPUs. Each of those bad boys has an MSRP of $1999, and they've been selling for even more. That puts this bleeding-edge tech squarely in the realm of the highest-spec gaming PCs, making it about as accessible to the average gamer as a dragon's hoard. Most of us are just trying to get our current rigs to run *Cyberpunk* without melting, let alone dropping four grand on GPUs for a tech that won't even be out until the autumn.
So, while DLSS 5 is still cooking in Nvidia's labs and due for a release later this year, there's a fair bit of ground to cover. Will further refinements address the “AI slop” aesthetic concerns? Will developers truly harness its “content-control generative AI” capabilities to create genuinely innovative visual styles, rather than just smoothed-out faces? And perhaps most importantly, will the hardware to run it ever become remotely affordable for the likes of you and me?
Only time will tell, mate. But for now, it feels like Nvidia's trying to sell us a powerful new spell, and we're just wondering if our existing wands can even cast it – or if it's going to accidentally give our beloved game characters another questionable haircut. We're curious, we're cautiously optimistic, but we're also watching those component prices like a hawk.
Tags: Nvidia, DLSS 5, AI Gaming, GPU Technology, PC Gaming
Original article: Game Rant