• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

DLSS 5 - Yes or No?

Do you think DLSS 5 is the future?

  • Yes and I like it

  • Yes but I don't like it

  • No, it's ugly and we'll forget about it

  • No opinion/other

  • No, we need less AI not more


Results are only viewable after voting.
I dont like it yet. At first I thought at least the environments will look better, but it just destroys shadows.

However, future games will be made with DLSS5 in mind, so they will look better with it turned on due shortcuts and also the tech isn't finished.. It's not so much a tech for enhancement of current games and instead one that should speed up development and deliver better graphics..

It does appear to destroy some shadows but I don't imagine that being too hard to fix. Shadows could be blended back in again or just have the AI get better at avoiding from doing that in the first place.
 
Just don't care. A few looked worse with the filter. And the screenshots don't seem representative of typical gameplay. RArely paying attention enough in the normal course of a game to notice either. And the brain adapts. EVen (if) I notice better detail/graphics it wears off pretty quickly.
 
Last edited:
This seems to basically be generative AI, though. Just like you can't expect to get the same results from ChatGPT and Grok with the same prompt, this kind of tech will not generate the same results across Nvidia/AMD/proprietary platform implementations, making it very difficult for devs to support multiple technologies.

It definitely wont look the same across models and platforms at least not 100%. But the goal is gonna be really similar.
 
The lighting is fantastic and transformative while not perfect. It removed shadows, etc, since it did not know where the sun/direct source was, as well as certain artistic environmental noir and smoke effects were missing, but surely that would be taken into account during development of a game... the snapchat filter faces, very hit or miss. Very few good, a lot bad.

In the early stages, it essentially looked like you put a video or image into an AI generator. Something we all made fun of, yeah? I wonder what changed for some?

Dual 5090s to produce it is hefty (on people's wallet) as well.
 
Last edited:
the fact that the most number of votes went into "i like it", shows how downhill we are going as a society, it looks horrendous. The only time it looked good was in AC Shadows when it was a small improvement to the environment.

Shame that many of you like it
 
Last edited:
It's an intersting tech.
Then again people having a rabid sperg-out over it on the basis of a demonstrative implementation as if it was its ultimate form (mostly because "AI is bad, 'k?") is exactly what I expect from social media these days.
 
Last edited:
It's an intersting tech.
Then again people having a rabid sperg-out over it on the basis of a demonstrative implementation as if it was its ultimate form (mostly because "AI is bad, 'k?") is exactly what I expect from social media these days.
I think it's great and I love it.

I also hate having to take part in performative fear based AI hate by even acknowledging it.
 
Last edited:
My undertanding is that DLSS5 is an AI process that improves how lighting behaves. It does not changes the geometry of the models or the levels. And if it does changes some aesthetic it through the devs.

Some might look good some might look bad. It depends on the game and how its implemented. I can see racing games, and games that pursuing realism being the most beneficied.

And it is a optional feature that right now is running on two RTX5090. The tech is awesome, if you don't like it don't use it. The discussion around if it should be implemented, and how impact develompmet, artistic vision and so forth is another complete topic for itself.
 
Last edited:
The actual tech behind it and the engineering to get it working at all is impressive. Problem is it looks like shit. I think it has a place when used to target specific areas, like enhancing certain lighting or environmental effects, but it looks like pure ai slop when it's replacing characters faces etc.
 
Last edited:
How do you vote yes if you see it just as an optional toggle? It's the optional future?
I mean it is the future just as many other AI based features.

Also, I think there is no reason to complain as anyone not wanting to use it doesn't have to.

I feel like we've had many discussions like this over the years with stuff like RT, RR, DLSS, MFG, etc... None of that has disappeared and whomever thinks it doesn't enhance games or is not interested in it doesn't have to use it.

I find it hard to find downsides to having more options.
 
I don't get why people say DLSS5 is replacing character's faces with "AI slop" when clearly the slide over thing demonstrates it doesn't. It improves the lighting and the behaviour of the textures in a way that's mindblowing.

It's a fair point that the impact is too aggressive in the Harry Potter example, but that's training and tuning. So is the amount of shininess on Claire. That improves quickly like it did for Midjourney. It seems insane anyone would prefer the Starfield original to the DLSS5 version.
 
Last edited:
I mean it is the future just as many other AI based features.

Also, I think there is no reason to complain as anyone not wanting to use it doesn't have to.

I feel like we've had many discussions like this over the years with stuff like RT, RR, DLSS, MFG, etc... None of that has disappeared and whomever thinks it doesn't enhance games or is not interested in it doesn't have to use it.

I find it hard to find downsides to having more options.
It's not the same though. Every other AI technique you mentioned was chasing after the ground truth as reference, to mirror the original intent at the fraction of the cost. Every AI experiment that has actually been adopted in development workflows are ones that efficiently get to that ground truth without wasting hardware or developer time. This, by it's very nature isn't doing that. The underlying tech is certainly revolutionary and will be used extensively in neural rendering workflows. But DLSS 5 specifically is, like you said, just an option for those curious or looking for a different flavor. The reasons people didn't like DLSS 1 were because they weren't good enough to replicate ground truth. Some teething issues exists here too, as there is always room to improve and get better, but the bigger issue is that it isn't even trying to match ground truth outside of lining up every pixel. No context is being fed for inference. It's running off of a single frame with color and motion vectors to guess what the scene even is. As a result, it's defining a whole new look, designed and trained by Nvidia, a tech company. Until devs can actually feed the ground truth into a system like this for its training from scratch and own it as their signature look, this is a whole different animal. Having intensity sliders and masks isn't even trying. It has be a whole IDE integrated into the game engine.

May be that will happen in future iterations (let's say DLSS 6) when Nvidia learns to toe the line better, but 5 isn't it as it is defined now. And the poll is "Is DLSS 5 the future?". Not "Is Neural rendering the future?". Unless your counter is "5 is obviously a step to 6". Then sure, that's trivially true. All future looking development is a step towards the future...
 
Last edited:
JBEaOLwZXZ2g7tsQ.jpeg
 
You are going to see it with AMD on the next gen consoles, especially with the next Xbox, when it comes to backwards compatibility. They will use this style of technology to enhance games like Halo, Gears 2, etc
 
KpK1WTbSb9QYARNm.jpeg
57VYh7Im4eueVbpr.jpeg
IBVZYDwRRJvnNDGN.jpeg


This all reminds me of the famous Dave Hill or Jill Greenberg photography.

Everything in the presentation has that studio lighting and way over processed look which I don't care for.
 
Ignore the comment in the Tweet and just watch the GIF. It's the same person


Exactly.

It seems the backclash is because people think DLSS5 somehow raids a bank of human images to hallusinate a face to replace the original one with whereas that very clearly not the case. The geometry is the same. The lighting is radically better. In some cases, too aggressive for the intent of the game - but that should be adjustable.
 
Bro. It's optional
What if it's not optional? Or what if games end up looking significantly worse without it? We used to hear that RT would always be just an option, but today it's become the foundation for some titles. DOOM: The Dark Ages requires RT just to launch: there's no non-RT mode anymore.
 
What if it's not optional? Or what if games end up looking significantly worse without it? We used to hear that RT would always be just an option, but today it's become the foundation for some titles. DOOM: The Dark Ages requires RT just to launch: there's no non-RT mode anymore.
Why would developers even spend the time on art direction anymore if they can just use this to do the work for them?
 
You guys think all AAA games look the same now just wait until AI puts the same face on every character.


images

0358cfe8-b9a3-4a1b-8223-40c5616109ba.jpg

images
We as men love diversity in women, those 3 dont look similar enough to be sisters even, and hell if they were, so fucking be it , its still way above what most western devs give us anyways :P
tom-brady-lets-go.gif

U know of any straight male who said no to have genuine 8/10 and up in their videogames/movies/videoclips etc? We love that shit. :messenger_smiling_hearts: Hell in sequel they can even make them triplets we fuck and no1 will complain, as long as they look hot enough :messenger_ok:

I think instead of playerbase complaining(activists and woke journos would 4sure but who tf cares about those cucks :P ) such a triplets sexscene would become instant classic, kinda like GoW3 sexscene is :messenger_sunglasses:
 
Is it THE future? It's part of it, but it's going to be a very useful feature. This is a great use case for AI and its good to see. I think all the "influencers" shitting on it fail to recognize it's just a toggle on/off feature. This is also the worst it'll be, it'll only improve. This is EXACTLY how we're going to reach photorealistic graphics.

But in cases of games not looking for realism I wonder how the game's looks will change? Doesnt seem like a useful feature in those cases.

I also wonder how well this is going to work initially. Apparently the effect was running on a different card entirely while another card rendered the game?
 
Last edited:
So you can basically run a filter now and make a PS4 game look like a PS8 title?

This is why Bluepoint was closed isn't it?

Publishing execs have basically seen this and believe that the future is essentially zero-cost remakes/remasters.
 
I have, and will always be, a gameplay first gamer. Graphics are nice and all, but I rate it pretty far down the list of features I look for in a game. One of my favourite games I played last year was Wizardry Gaiden III for the GB for crying out loud, so I'm probably not the target audience. 😋

For the topic at hand - as long as a game has a good foundation (gameplay/mechanics/systems) I'm not fully against them using some AI to pretty things up or punch above their weight class. But if the "bones" of your game are soulless trash, then throwing all the graphics on it means absolutly fuck all.

I mean I'd much rather punch the charming goblins in the game TintoConCasera TintoConCasera is making than experience some cinematic and cringe hollywood-wannabe game from western AAA devs.
 
It's not the same though. Every other AI technique you mentioned was chasing after the ground truth as reference, to mirror the original intent at the fraction of the cost. Every AI experiment that has actually been adopted in development workflows are ones that efficiently get to that ground truth without wasting hardware or developer time. This, by it's very nature isn't doing that. The underlying tech is certainly revolutionary and will be used extensively in neural rendering workflows. But DLSS 5 specifically is, like you said, just an option for those curious or looking for a different flavor. The reasons people didn't like DLSS 1 were because they weren't good enough to replicate ground truth. Some teething issues exists here too, as there is always room to improve and get better, but the bigger issue is that it isn't even trying to match ground truth outside of lining up every pixel. No context is being fed for inference. It's running off of a single frame with color and motion vectors to guess what the scene even is. As a result, it's defining a whole new look, designed and trained by Nvidia, a tech company. Until devs can actually feed the ground truth into a system like this for its training from scratch and own it as their signature look, this is a whole different animal. Having intensity sliders and masks isn't even trying. It has be a whole IDE integrated into the game engine.

May be that will happen in future iterations (let's say DLSS 6) when Nvidia learns to toe the line better, but 5 isn't it as it is defined now. And the poll is "Is DLSS 5 the future?". Not "Is Neural rendering the future?". Unless your counter is "5 is obviously a step to 6". Then sure, that's trivially true. All future looking development is a step towards the future...
Well, from what I've read people reject AI for all sorts of reasons. This is the first time I see the line drawn at this point although it sounds similar to what I've heard some people saying about MFG. It's fair but my opinion still stands. To me it's just one more feature set that may or may not be used by devs and may or may not be activated by players.

If context is being fed or not I think is secondary to me. If it doesn't match the creator's intent but the public prefers the result with AI is it really a problem? Would you say the same about a mod or would you consider that differently because there is another creator behind a mod?

And yes, of course DLSS 5 is just an intermediate point towards something better as always. I don't think anyone thinks that this is the only tech we will see in years to come but future technologies might share the same approach. So if you are looking for precision then I'd say: "yes, I think AI is the future" (this includes DLSS and any other similar tech, curreent and future versions).
 
We as men love diversity in women, those 3 dont look similar enough to be sisters even, and hell if they were, so fucking be it , its still way above what most western devs give us anyways :P
tom-brady-lets-go.gif

U know of any straight male who said no to have genuine 8/10 and up in their videogames/movies/videoclips etc? We love that shit. :messenger_smiling_hearts: Hell in sequel they can even make them triplets we fuck and no1 will complain, as long as they look hot enough :messenger_ok:

I think instead of playerbase complaining(activists and woke journos would 4sure but who tf cares about those cucks :P ) such a triplets sexscene would become instant classic, kinda like GoW3 sexscene is :messenger_sunglasses:
Please jerk off before posting.
 
7zKHwS21m1KMwqkc.jpg


Thats impressive how much closer it is to the actual model. So my vote goes to yes. The tech is incredible and iam sure we see some devs making great use of it.
 
Last edited:
Yes, and realistically it's the best way to push visual quality at this point. Graphics have been stagnant for about a decade, and studios aren't suddenly going to start producing Naughty Dog–level projects. This looks very promising, the videos already show great results, and the tech is only going to get better with time.
I'm satisfied with current-gen visuals: Resident Evil looks fantastic! The real issue is that physics and environmental interactivity have completely stalled. Too many game worlds feel like static dioramas where nothing moves or reacts.
Developers just don't prioritize it because flashy graphics sell games, while physics don't pop in screenshots. It's much easier to market "eye candy" than complex systems you actually have to play to appreciate.

Also, I don't get why so many players here obsess over photorealism. To me, realistic games are boring! When I want realism, I just go outside. :messenger_grinning_smiling:
 
Hell naw, another technology I won't be using even if one day my future PC (or console) will be fully capable of supporting it.

I'm still hoping that the overall negative reception will make Nvidia rethink their approach and what DLSS should be moving forward.
 
Last edited:
So you can basically run a filter now and make a PS4 game look like a PS8 title?

This is why Bluepoint was closed isn't it?

Publishing execs have basically seen this and believe that the future is essentially zero-cost remakes/remasters.
We'll probably never know, but I think you're onto something. I don't recall how long industry insiders have known about or how long ago they'd first seen DLSS 5 in action, but if it's been months, the timing would line up.
 
By the time that 40% of ppl who said yes in the poll gonna actually be able to test it in their games it will likely be ps6pr0 and nvidia 7000 series time so 2032 earliest.
 
This is really bad for video games artistically. Developers create something, only for some AI algorithm to "override" their artistic vision.

Hey, how about we do that with EVERY art piece we visually consume? TV shows, pictures, paintings - everything. What's the fucking point of creating art then?
 
Top Bottom