You can sum up the gamer response to NVIDIA's DLSS 5 announcement with the ever-relevant Fallout 4 meme: "Everyone disliked that." Across social media and Reddit last night, I couldn't find anyone who's genuinely positive about the potential for DLSS 5, which uses AI to add "photorealistic" lighting and materials to in-game models and environments. Instead, it's mostly complaints about the feature being another avenue for AI slop. And you know what? I agree.

It's not unusual to see gamers being reflexively angry about new technology on the internet, especially when it's being pitched by NVIDIA as the “biggest breakthrough in computer graphics” since its RTX 20-series GPUs arrived in 2018 with real-time ray tracing. There was plenty of suspicion around DLSS's original AI upscaling model, as well as the "fake" frames generated by later iterations. But the few demos we've seen of DLSS 5 basically look like "yassified" AI filters for popular games.

Leon and Grace from Resident Evil: Requiem have more distinct facial and hair detail, but they look a bit too slick. There are more wrinkles on an old woman in Hogwarts Legacy. And the face, hair and clothing from a Starfield character gain an uncanny sheen.

None of the demos have the immediate impact of the Star Wars real-time ray tracing short ILMxLab produced with NVIDIA seven years ago. That demonstration showed us glorious reflections and lighting effects we'd never seen before in real-time. The DLSS 5 demos, on the other hand, don't look much different from the AI filters that make you look more presentable for Zoom calls. There's no genuine excitement for DLSS 5, just NVIDIA telling us that it's groundbreaking.

There's also plenty of concern about DLSS 5 straying from an artist's original intent, as well as a potential homogenization of game visuals if every developer starts using the feature. NVIDIA claims developers will have "detailed controls for intensity, color grading and masking," which will help DLSS 5 stay in line with a game's aesthetic. But we don't have any direct developer experience with the feature yet — some artists may want far more control than NVIDIA wants to give.

The difference between DLSS 5 and earlier versions NVIDIA's upscaling is like the difference between generative AI and more traditional machine learning models. NVIDIA relied on the latter to make low-resolution textures and models appear sharper, and later to insert generated frames to smooth out gameplay and raise your fps count. As Wirecutter and former Polygon editor Arthur Gies points out, you could argue those features were in service of delivering what developers originally intended. But DLSS 5's neural model applies its concept of "photorealism" on top of what games are rendering -- it's like watching a Pixar movie that let OpenAI's Sora do a final visual pass.

Part of the negative response towards DLSS 5 may stem from a widespread anti-gen AI sentiment, but that doesn’t devalue the criticisms either. Similar to AI generated text, images and video, there’s a dehumanizing aspect about DLSS 5. It can erase the work of human artists (despite how much control NVIDIA claims they have), and it also feels like a calculated attempt to appeal to gamers who just want shinier graphics. NVIDIA showed off how generative AI could be used to create dialog and voices for NPCs last year at CES, but that was also widely disliked (and I called it a genuine nightmare).

Of course, I can’t fully judge DLSS 5 until I see it in action beyond a short demo. But I think the visceral disgust is an important indicator that many gamers aren’t onboard with the AI-powered future NVIDIA is trying to sell us. And perhaps the idea of chasing “photorealism” may be a bit of a fool’s errand. It may be appropriate for some games, but as Nintendo and indie PC devs have shown, you can also make some of the best games of all time without striving for realism. Tears of the Kingdom could use a better framerate and higher resolution textures, but it certainly doesn’t need DLSS 5.

Just months after announcing DLSS 4.5 at CES, NVIDIA has unveiled its next major upscaling technology, DLSS 5. The company is doubling-down on AI for this next iteration, claiming DLSS 5 “infuses pixels with photoreal lighting and materials” using a real-time neural rendering model when it arrives this fall.

So what does this mean in practice? In an on-stage demo at NVIDIA’s GTC 2026 keynote, CEO Jensen Huang showed off the technology with Resident Evil: Requiem, Hogwarts Legacy and Starfield. DLSS 5 adds a noticeable amount of detail to character’s hair and skin tone, but it also appears it’s being compared to those games without any DLSS features turned on. It’s unclear how much of a difference it makes compared to DLSS 4.5 with path tracing and all of its features turned on.

“DLSS 5 takes a game’s color and motion vectors for each frame as input, and uses an AI model to infuse the scene with photoreal lighting and materials that are anchored to source 3D content and consistent from frame to frame,” NVIDIA said in a blog post. The company also notes that the technology runs in real time, and it works at up to 4K.

Huang showed off DLSS 5 while running a system with two RTX 5090 GPUs. Eventually, it will be able to run on a single video card (though I’d imagine it would have to be almost as powerful as two 5090s). Huang also paints DLSS 5 as a step towards offering Hollywood-like quality for real-time rendering, without the need for the GPU horsepower required by studios. It sounds a bit like a generative AI video model that can be directly controlled by developers, instead of just AI prompts.

NVIDIA, never shy from self-aggrandizing, claims DLSS 5 is also the “biggest breakthrough in computer graphics” since real-time ray tracing arrived in 2018. But given that ray tracing itself hasn’t been mainstream for many gamers, it’ll be interesting to see if there’s any interest in NVIDIA’s AI-produced pixels.