Monday, March 16th 2026

NVIDIA Unveils DLSS 5 with Real-Time Neural Rendering

NVIDIA today unveiled NVIDIA DLSS 5, the company's most significant breakthrough in computer graphics since the debut of real-time ray tracing in 2018. DLSS 5 introduces a real-time neural rendering model that infuses pixels with photoreal lighting and materials. Bridging the divide between rendering and reality, DLSS 5 empowers game developers to deliver a new level of photoreal computer graphics previously only achieved in Hollywood visual effects.

"Twenty-five years after NVIDIA invented the programmable shader, we are reinventing computer graphics once again," said Jensen Huang, founder and CEO of NVIDIA. "DLSS 5 is the GPT moment for graphics—blending hand-crafted rendering with generative AI to deliver a dramatic leap in visual realism while preserving the control artists need for creative expression."
Bridging The Cinematic Gap
Since the dawn of GeForce, NVIDIA has strived to deliver the graphics horsepower required for game developers to create incredible, realistic worlds—where lighting, reflections and shadows obey the laws of nature.

From programmable shaders with GeForce 3 in 2001, to CUDA with GeForce 8800 GTX in 2006, to real-time ray tracing with GeForce RTX 2080 Ti in 2018, to path tracing and neural shaders with GeForce RTX 5090 in 2025, NVIDIA has delivered major architectural innovations and a massive 375,000x increase in compute to meet this challenge.

However, the rendering horsepower available to a 16-millisecond game frame remains a tiny fraction of that available to a photoreal Hollywood VFX frame, which can take minutes to hours to render. Real-time rendering cannot bridge the gap to photorealism through brute force alone.

NVIDIA DLSS was released in 2018 as an AI technology to boost performance, first by upscaling resolution, and then by generating entirely new frames. It has been integrated in over 750 games, becoming a gold standard for the industry. Launched at CES this year, DLSS 4.5 uses AI to draw 23 out of every 24 pixels seen on the screen. Today, DLSS is evolving beyond performance to transform visual fidelity in games.

Video AI models have rapidly learned to generate photoreal pixels, but they run offline, are difficult to precisely control and often lack predictability, with every new prompt generating bespoke content. For games, pixels must be deterministic, delivered in real time and tightly grounded in the game developer's 3D world and artistic intent.

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. DLSS 5 runs in real time at up to 4K resolution for smooth, interactive gameplay.

The AI model is trained end to end to understand complex scene semantics such as characters, hair, fabric and translucent skin, along with environmental lighting conditions like front-lit, back-lit or overcast—all by analyzing a single frame. DLSS 5 then uses its deep understanding to generate visually precise images that handle complex elements such as subsurface scattering on skin, the delicate sheen of fabric and light-material interactions on hair, all while retaining the structure and semantics of the original scene.

DLSS 5 provides game developers with detailed controls for intensity, color grading and masking, so artists can determine where and how enhancements are applied to maintain each game's unique aesthetic. Integration is seamless, using the same NVIDIA Streamline framework used by existing DLSS and NVIDIA Reflex technologies.

Availability & Game Developer Support
DLSS 5 will be supported by the industry's biggest publishers and game developers, including Bethesda, CAPCOM, Hotta Studio, NetEase, NCSOFT, S-GAME, Tencent, Ubisoft and Warner Bros. Games.

"Bethesda has such a rich history pushing graphics with NVIDIA, going all the way back to Morrowind, with that incredible water," said Todd Howard, studio head and executive producer at Bethesda Game Studios. "When NVIDIA showed us DLSS 5 and we got it running in Starfield, it was amazing how it brought it to life. We've played it. We can't wait for all of you to do so as well."

"At CAPCOM, we strive to create experiences that feel cinematic, compelling and deeply believable—where every shadow, texture and ray of light is crafted with intention to enhance atmosphere and emotional impact," said Jun Takeuchi, executive producer and executive corporate officer at CAPCOM. "DLSS 5 represents another important step in pushing visual fidelity forward, helping players become even more immersed in the world of Resident Evil."

"Immersion is about making the world feel real. DLSS 5 is a real step towards that goal," said Charlie Guillemot, co-CEO of Vantage Studios. "The way it renders lighting, materials and characters changes what we can promise to players. On Assassin's Creed Shadows, it's letting us build the kind of worlds we've always wanted to."

DLSS 5 will come to games including AION 2, Assassin's Creed Shadows, Black State, CINDER CITY, Delta Force, Hogwarts Legacy, Justice, NARAKA: BLADEPOINT, NTE: Neverness to Everness, Phantom Blade Zero, Resident Evil Requiem, Sea of Remnants, Starfield, The Elder Scrolls IV: Oblivion Remastered, Where Winds Meet, and more.

DLSS 5 will arrive this Fall. See the first preview this week at GTC, and below, you can check out several examples of NVIDIA DLSS 5 in Resident Evil Requiem, EA SPORTS FC, Starfield, Hogwarts Legacy, and the NVIDIA Zorah tech demo.

Resident Evil Requiem
EA SPORTS FC
Starfield
Hogwarts Legacy
NVIDIA Zorah Tech Demo
Source: NVIDIA
Add your own comment

305 Comments on NVIDIA Unveils DLSS 5 with Real-Time Neural Rendering

#176
Zazigalka
I like the results, but it's going to be very taxing for GPUs. I suppose next gen Vera xx70S/Ti will be entry level to use it. From this gen, 5080 and 5090 only is my prediction.
Posted on Reply
#177
Visible Noise


Ryan Shrout:

I Just Saw DLSS 5 Running Across Multiple Games. It's Not a Face Filter.

NVIDIA just dropped DLSS 5 at GTC 2026, and the internet already has opinions.

I was in the room and I went hands-on. Not watching a sizzle reel, not scrubbing through a carefully curated 30-second trailer, but sitting in front of multiple games with DLSS 5 toggling on and off in real time. Hogwarts Legacy. Starfield. Assassin's Creed Shadows. Oblivion Remastered. The Zorah tech demo. The visual improvements are significant. Not incremental. Significant.

But if you've been scrolling social media, you'd think NVIDIA just shipped an Instagram beauty filter for video games. And I get why that's the first reaction. But it misses the true picture by a wide margin.

Why Faces Get All the Attention

We've had photorealistic environments in games for a while now. Water reflections, volumetric lighting, incredibly detailed cityscapes and forests. The hardware and the rendering techniques have gotten us to a place where environments can look stunning under the right conditions.

But faces have been the holdout. Getting a human face to look truly photorealistic in real time has been one of the most expensive problems in computer graphics from a compute standpoint. Subsurface scattering on skin, the way light interacts with individual strands of hair, the micro-expressions that make a character feel alive rather than like a wax figure. All of that requires an enormous amount of rendering horsepower..

I've probably seen ten different "floating head" tech demos over the course of my career. That's not an exaggeration. They're always a single head with no hair, no body, no environment, because rendering a photorealistic face at that level of quality is so expensive that it can only be done in isolation. You never see it inside an actual game, because the performance budget won't allow it.


DLSS 5 closes that gap in a pretty dramatic way. And because that's the area where the delta between "before" and "after" is most visible, that's what everyone is reacting to. The NVIDIA team put it well during my demo. It's a psychological effect. You've seen environments rendered really well before. When you suddenly see a character rendered at that same photorealistic level, your brain flags it immediately. It stands out.


Fair enough. But focusing only on the faces is wrong.

It's Happening Everywhere, Not Just on Character Models

What I saw in the demos was a comprehensive improvement across the entire scene. And the moment that really drove this home wasn't a face. It was a coffee maker.

In Starfield, there's a countertop scene with a coffee machine, some paper towels, a cup, napkin holders. Standard environmental clutter. With DLSS 5 off, everything looks flat. The coffee maker fades into the background. Toggle it on, and suddenly the objects have shape. The lighting wraps around them naturally. The spatial relationships between the items and the surfaces they're sitting on become clear. It goes from "assets placed in a scene" to "objects that actually belong in a room."


The same thing played out across every title. In Oblivion Remastered, the water went from good video game water to something that could pass for real, with the kind of light interaction and shimmer you'd expect from an offline render. In Assassin's Creed Shadows, the trees and distant foliage gained dramatically better depth and separation in how light moved from the canopy down through the branches. In the Zorah tech demo, which is a 300 GB courtyard scene built by 20 full-time artists, the subsurface scattering on foliage was just as impressive as anything happening on character faces. Leaves picked up that translucent glow from backlighting that is incredibly difficult and expensive to model and render through traditional means.


The AI model powering DLSS 5 is a single unified model. Same model for every game. It's not trained per-title, per-face, or per-object type. It takes the raw color buffer and motion vectors as input, analyzes the scene semantics from that single frame, and enhances the lighting and material response while staying anchored to the original 3D content. It recognizes the difference between skin and metal and water and stone and foliage, and it processes each of those materials differently based on how light should interact with them.

That's not a filter. That's a fundamentally different approach to how the final image gets assembled. And it's deterministic and consistent from frame to frame, which is a hard requirement for games.

The Developer Angle Matters More Than People Realize

One of the things I came away most encouraged by is the developer control story. This is critical. If DLSS 5 were a black box that slapped a one-size-fits-all enhancement over every game, the artistic intent concerns would be completely valid. But that's not what this is.


During the demo, the DLSS research talked through the level of granularity available. Developers don't just get an on/off switch. They get intensity controls that can be dialed anywhere, not just full strength. They get spatial masking, so they can set the water enhancement to 100%, wood to 30%, characters to 120%, all independently within the same scene. They get color grading controls for blending, contrast, saturation, and gamma. All of this runs through the existing SDK, which means studios already using DLSS and Reflex have a familiar pipeline to work with.


The developer support list tells you something. Bethesda, CAPCOM, Ubisoft, Tencent, Warner Bros. Games, and others have already signed on. But what struck me more than the names was what the NVIDIA team shared about the reactions inside those studios. When developers previewed the technology, their technical artists were apparently co-advocating for it internally, because it gets them closer to what they actually intended their characters and environments to look like when they were designing them in their authoring tools. Then those assets get dropped into a real-time game engine with a finite performance budget, and compromises happen. DLSS 5 lets them claw back some of what gets lost in that translation.

I think that's the right framing. DLSS 5 isn't NVIDIA applying its stylistic choices on top of someone else's game. It's providing a tool that helps developers close the gap between what they can render in 16 milliseconds and what they actually want the player to see. That's a meaningful distinction, and it's a big reason why the developer response has been positive.

The Hardware Story Is Interesting Too

The demos I saw were running on a pair of RTX 5090 GPUs. One was handling the game rendering, the other was dedicated entirely to running the DLSS 5 AI model. NVIDIA was upfront that there's still significant optimization work to do, and the plan is to ship DLSS 5 running on a single GPU when it launches later this year.


But I think the dual-GPU setup itself is worth mentioning. For years, multi-GPU gaming has been effectively dead. SLI is gone. CrossFire is gone. The idea that you'd run two graphics cards for a better gaming experience felt like a relic of the mid-2000s. And yet here we are, with a legitimate use case where a second GPU running an AI workload alongside a primary rendering GPU produces a dramatically better visual result.


Is that where this ends up for enthusiasts? Probably not at launch. But the concept of dedicating GPU compute specifically to AI-driven visual enhancement, separate from the rendering pipeline, is an interesting architectural idea. It wouldn't surprise me if that becomes a real conversation again as neural rendering matures.

Where This Goes From Here

DLSS 5 is targeting a fall 2026 launch, which means we've got several months of optimization and refinement ahead. Developers are just getting their hands on it now, and they'll need time to work with the controls and dial in the right settings for their specific titles. First-wave games include Starfield, Assassin's Creed Shadows, Resident Evil Requiem, Hogwarts Legacy, Phantom Blade Zero, The Elder Scrolls IV: Oblivion Remastered, Delta Force, and more.

It's also worth noting that this works across rendering approaches. Rasterized games, ray-traced titles, and path-traced experiences all benefit. And the higher the fidelity of the input, the better the output. DLSS 5 isn't replacing good rendering. It's amplifying it.

The early social media reaction is predictable. New technology that changes how games look will always generate strong opinions, especially when AI is involved. But the knee-jerk "it's just a face filter" take doesn't hold up once you've actually seen the full scope of what DLSS 5 is doing across an entire scene, across multiple games, in real time. Go look at a coffee maker. Go look at stone textures. Go look at the way light passes through a leaf. That's where the real story is.

What do you think, is neural rendering the next big unlock for game visuals? I'd love to hear from people who have spent time with these games.
Posted on Reply
#178
Frick
Fishfaced Nincompoop
Visible Noise

Ryan Shrout:
"In Oblivion Remastered, the water went from good video game water to something that could pass for real, with the kind of light interaction and shimmer you'd expect from an offline render."

And that's what is puzzling to me. Why is this considered a universally good thing? Why does "good graphics" = "photorealism"?

We'll see though, I'll definitely play around with it in Starfield.
Posted on Reply
#179
TumbleGeorge
Frick"In Oblivion Remastered, the water went from good video game water to something that could pass for real, with the kind of light interaction and shimmer you'd expect from an offline render."

And that's what is puzzling to me. Why is this considered a universally good thing? Why does "good graphics" = "photorealism"?

We'll see though, I'll definitely play around with it in Starfield.
The only games I want photorealism in are those for adults. :D
Of course, I think that the rest of the games are realistic enough with the old processing methods that do not need hundreds of teraflops of performance and tens of gigabytes of VRAM.
Posted on Reply
#180
Vya Domus
TomorrowMale characters too unless i missed something but that doesn't seem to bother people. People like beautiful people. If that makes them gooners, then we're all gooners regardless of our gender.
You can simply make the characters attractive to begin with, I don’t want to have to pay NVIDIA a bajillion dollars to unbustify ugly character models.

These idiots are trying to ruin games in ways I haven’t even thought possible.
Posted on Reply
#181
Onasi
FrickAnd that's what is puzzling to me. Why is this considered a universally good thing? Why does "good graphics" = "photorealism"?
The main problem I see, though this isn’t really what NV is trying to solve here, is that the more “realistic” the graphics, the more all the things that developers have been spinning their wheels on for years will be noticeable. Shit physics, low interactivity, poor AI, overall “feel” and reactivity of the virtual world. They are chasing advancement in one area without considering, seemingly, how it would affect everything else.
Posted on Reply
#182
Visible Noise
FrickAnd that's what is puzzling to me. Why is this considered a universally good thing? Why does "good graphics" = "photorealism"?
In certain instances it does. In others it doesn’t.

Which is why this is an optional tool for developers to use as they wish.
Vya DomusYou can simply make the characters attractive to begin with, I don’t want to have to pay NVIDIA a bajillion dollars to unbustify ugly character models.

These idiots are trying to ruin games in ways I haven’t even thought possible.
I know right? We should all just be playing games made of 2D sprites. Anything else is just ugly. Quake should have never happened.
Posted on Reply
#183
CornyOC
Bobaganoosh
This was captured straight from a GeForce video. I checked lol. I was also noticing 6 fingers at one point in the video on a player. I'm surprised they put this out there.
Nvidia thought the AI slop mask mode is ready, but clearly it isn't lol
Posted on Reply
#184
Legacy-ZA
Visible NoiseImagine that, changing lighting - such as self shadowing - could make a face look different!
Exactly, it makes a massive difference. If people would use the slider comparison, they would see the face remains largely the same, but the hair shadow falling on Leons brow and beard as an example, can help hide an expression, it doesn't remove it. The muscle crease lines, nose shadows, the ears, the beard etc. They have to look way more closely this is truly amazing in my opinion.

What we can glean from this whole thing is; there are those that don't see it, and will rightfully criticize it because they have been "conditioned" so to say over for many years have we played games now? Who is a 40 year+ gamer? Mmmmm, yes, remember, our virtual worlds, were never real and we have gotten so used to virtual lighting to simulate realism, that how something should look. I think artistically, developers will have to be careful or face extreme backlash like we see here and there. They will have to use different lighting to capture the scene/face they wanted. Or nGreedia will have to find a way to toggle certain assets to use DLSS5, and others, not.

For example, use DLSS 5.0 for faces, not the rest of the gameworld or vice versa. Think of something like Alien Isolation (YA sequel coming, ehehehe, can't wait) you want to see real faces and a big nasty Xeno, but the environments should be anything but a sunny forest, so they can use it, but will have to be careful not to overlight the world or build their world around DLSS5.

Anyways, I see this whole thing as a net positive, what scares me, is lazy development/artistic direction.
Posted on Reply
#185
remekra
Visible Noise

Ryan Shrout:
"Those faces look uncanny to you? Well, guess you just haven't seen enough real people mate".
Posted on Reply
#187
cinemaware
Holy Cow, Jensen with a real mic drop moment here. And, predictably, the internet loses its collective mind. I guess the chance to shout "slop !!1!" is still irresistible for many.

I suppose that there you could say there's an iota of reason for all this crying , because at a quick glance there is an element of the uncanny AI valley at some instances re: faces. But:

-these are only few, overall I'd say the look is 80/20% in favour of the new tech

-it's just an early demo. Remember DLSS 1 or 2018 RTX? And what we have now?

-"you need two 5090s to run it": see above

-the materials, objects, and structures look OUTSTANDING, and that's what matters most to me. Can't wait to see it applied in Oblivion Remastered.

Ultimately what will make or break it is how it looks in motion. Until then, I'm quite sold, I say bring it on. And yeah, it will be an optional so all the people now talking about the end of the world can simply skip it.
NinWeird. I don't think these pictures are doing Nvidia a favour at all. The characters' faces look completely different from their actual designs. The ones in Resident Evil and Hogwarts Legacy especially. All these pictures have that GenAI feeling to them--they look beautiful and impressive, but they don't look right. Maybe in some top-down games I'd turn this on, but in games where character designs are important, I'd give this a wide berth.
I couldn't disagree more. Clearly you haven't compared them by flipping from one to another, if you did you’d see the models are exactly the same and all the stuff is just enhanced. Of course, in some cases the effect is quite dramatic but that's because the originals are so basic.
NinOn the first EA Sports picture, the cloth of Van Dijk's shirt even folds differently. Not the way the game natively renders it. It looks like DLSS is drawing on top of it with AI. ([Edit] Oh, it is. I should have read the text better.) Just like on the characters' faces, in fact. also note how Grace gets fuller lips in the screenshot where she's standing in the street, but not in the screenshot with the close-up of her face. Awful.
The cloth is supposed to be dynamic, just like the crowds behind him :D As for Grace, comparing two different scenes with different positioning and light is pointless. All these characters get enhanced features, like I said above.
Posted on Reply
#188
Vya Domus
cinemawareAnd what we have now?
What we have now is the same crippled performance we had back then + a million band aid solutions to fix a problem that did not exist before. That’s what we have now.
Posted on Reply
#189
TumbleGeorge
It's obvious that Nvidia fans can't wait to buy a new collection of leather jackets for Mr. Huang.
Posted on Reply
#190
BoggledBeagle
This is a mess.

There are these points.
1) a game creator intention.
2) a game creator/engine/development limitations, which caused him to deviate from his intention
3) gpu limitations, which make things look worse than they can for end users of the game.

If there are some tools available to help the game creators not to deviate from their intentions too much, they should be used, and it does not matter if they are AI related or not.

But this help needs to be done in the game develompent stage.

Any on the fly improving/reworking of the game WITH MASSIVE COMPUTATIONAL AND ENERGY COST happening while people play the games is TOTALLY UNACCEPTABLE.

GPUs are there to RUN THE GAMES ONLY. Not to remake them. Nvidia should concentrate on making the GPUs more powerful to run the games as they are, and not to develop new games on the fly.
Posted on Reply
#191
Vya Domus
This whole discussion of photorealistic faces is dumb, we already know how to do that without it looking uncanny, it’s called photogrammetry, we’ve been doing it since the early 2000s, look at Death Stranding.

Studios have been making characters ugly, busted looking and uncanny in the last few years for reasons unrelated to technical limitations that I should not go into but everyone knows why.

Yet another dumb solution to a problem that was artificially introduced.
Posted on Reply
#192
TumbleGeorge
All games that are fully and qualitatively completed, not bloated and well optimized, can be played at 8K resolution and do not require 32 GB or more of VRAM. Performance starting from the RTX 3090 is sufficient.
Posted on Reply
#193
Visible Noise
CornyOCNvidia thought the AI slop mask mode is ready, but clearly it isn't lol
A tech demo at a technology conference isn’t production ready yet?

Wow, I’m shocked!

(Plonk)
Posted on Reply
#194
cinemaware
Vya DomusWhat we have now is the same crippled performance we had back then + a million band aid solutions to fix a problem that did not exist before. That’s what we have now.
I'm sorry that in the parallel universe you inhabit people still have to run CP2077 et al looking flat and naked. Here it’s all quite different though.
Posted on Reply
#195
Visible Noise
BoggledBeagleAny on the fly improving/reworking of the game WITH MASSIVE COMPUTATIONAL AND ENERGY COST happening while people play the games is TOTALLY UNACCEPTABLE.
Tell us you don’t know what this is without telling us.
Posted on Reply
#196
MufasaV
Testsubject01Changing the creators visual intent is usually a big NONO in the professional industry anyway.

Especially since gen Ai will show a different result between each instance a model is shown in the game. (most likely why longer cutscenes/ gameplay was omitted in this showpiece as well).
Then showing FIFA, which even has expensive licences on the likeness of the featured players and advertises that a large part… good job slopping faces with Ai here.
Gaming has never cared about artistic intent. High quality mods or just mods in general is way too large an industry otherwise.

Heck, everyone shits on the artistic direction of other things like film grain, motion blur, etc.

Considering only the artists have any control over how DLSS 5 looks, that emotional appeal makes no sense
Posted on Reply
#197
BoggledBeagle
Visible NoiseTell us you don’t know what this is
What? So far we have seen just ray tracing - which is improvement of ligting the scene, generating interpolated frames to improve motion smoothness and upscaling. That is nothing like changing the game itself, as completely reworking faces of human characters.
Posted on Reply
#198
CornyOC
Visible NoiseA tech demo at a technology conference isn’t production ready yet?

Wow, I’m shocked!

(Plonk)
They have been showcasing missing arms, six fingers, and all sorts of AI weirdness. They are hilariously bad, and they though this is ready for public showcase. You didn't look at stuff all that much I presume.
Posted on Reply
#199
Visible Noise
BoggledBeagleWhat? So far we have seen just ray tracing - which is improvement of ligting the scene, generating interpolated frames to improve motion smoothness and upscaling. That is nothing like changing the game itself, as completely reworking faces of human characters.
JFC. Did you read before posting?

It doesn’t change the game. It’s something game developers implement.

Again:

www.techpowerup.com/forums/attachments/img_0875-webp.437254/
Posted on Reply
#200
remekra


Nothing to see here. Just some AI blinking. Sure it only affects lighting.
Posted on Reply
Add your own comment
Nov 20th, 2026 15:45 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts