• 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.

KyoZz

Tag, you're it.
The title is self-explanatory, what do you think?

STARFIELD.gif
FIFA.gif
RE9.gif

Thanks Represent. Represent. for the GIFs





HDj6TzZaoAADlzr


55-scaled.jpg
5-scaled.jpg

22-scaled.jpg
2-scaled.jpg

1-scaled.jpg
11-scaled.jpg
 
Last edited:
Its current form looks bad, but I have no doubt that once devs start using it with full customization, it could be good. It'll all be down to how it's implemented and improved upon.

But either way, it's been absolutely hilarious to see all of the memes and backlash today lol.
 
Yes, so far, i like it, i was impressed with the games they showed, whether it will turn out as good at the time i don't know, we have to wait until then.
 
Last edited:
I like it but i dont think its the future. I think it's a cool tool that will help with the modding scene but its not going to replace standard graphics.
 
Very hard to say all honestly, it could go either way.

Is it impressive? Yes

Did Nvidia pick the best shots for the press release? Most Certainly

Is Nvidia overselling this? Probably

Will it disappoint at release? Maybe

Is this the uncanny valley? Yes
 
Last edited:
Cool tech. Ugly results. Probably has a ton of issues since what they've shown are either still shots or the camera moving around slowly.

Basically catnip for investors, think and wish this will get nowhere.
 
Last edited:
Yes mainly because we've all seen it and are using it now. This is why we can pass judgement based on our own hard data!

DLSS5 releases this fall
 
Look terrible. Changes too much to the point we get into uncanny valley territory. Changing lips, eyes, backgrounds, etc to the point that it's ruining the game it was used on.
 
Yes, but will probably never use it since I do not have or will have an RTX 5090+ anytime ever.
 
Last edited:
I've only seen those few clips from the DF video.
Turning the flat lifeless Starfield into something with actual depth with shadow and contrast is impressive.
I like the few clips of Hogwarts Legacy that were shown. The game's Raytracing was broken and never fixed, and DLSS 5 was able to help that post launch? That's kind of crazy.

I can imagine that developers are only going to need to add certain 'cues' to future games to 'help' DLSS.

I need to play around with it in an actual game in my actual home to really judge. Right now it looks like a quick demo of what might be possible in the future.
 
I guess I'd need to know more about how much control developers have over it. If it's doing all this on its own with no substantial developer input, then no, not really a fan of that.
 
I'll use it if I can run it, but I'm curious to see what the requirements will be. My money is on this will end up being very intensive performance-wise and, therefore, will be used as a feature that draws people into streaming.
 
Last edited:
I'm all in… balls deep with all 4 inches.

This is worth a read btw, it was posted in the main thread:



Also it seems that everything we saw today was by the devs/artists themselves, meaning everything we saw Bethesda, Capcom, Ubi, etc., was their own doing. It wasn't just a thing Nvidia did by themselves.
 
Do I think DLSS 5 is the future? Not necessarily, but I do think similar tech will be. IMO AI will be the key to producing truly realistic looking games; the industry will go bankrupt before we get there with traditional methods.
 
I'll post what I said in the graphics thread:

They really picked the worst way to showcase this. You do not take Requiem, a game with perhaps the best character models out there, and turn it into AI sloppa. I do not blame anyone for thinking that looks abysmal, as I do too.

But Nvidia also clarified that developers have a large amount of control over the artistic side of the presentation. They can use masking on any part of the visuals which they don't want DLSS5 to touch. The ideal would be if developers offer the ability to disable it on faces via the graphics options, but leave everything else to DLSS5. Look beyond the faces in those Starfield comparisons and it's working some real magic in the environment. When used conservatively, this might be able to clean up some long-standing issues in rendering.

It will also be interesting to see how this impacts development. Likely not much since it's proprietary tech. This could be a massive equalizer though for shit-tier devs, like pitting a granny vs a 6ft 6 bodybuilder except now you've given granny a stub nose revolver in her purse. I imagine it could theoretically speed development up a lot, but again, it's proprietary tech so has limited use on consoles. No developer is going to replace their whole rendering pipeline with this any time soon.

Digital Foundry also said in their shill video that developers get better results with better inputs. We will need to see what level of shit DLSS5 can be fed and still churn out a good result. Perhaps DLSS5 may even make Nvidia's own Remix mod tools irrelevant if it works well enough. Perhaps no more terrible looking path-tracing mods for ancient games, with pitch black hallways and plastic looking textures which everyone pretends looks "better".

I also wonder how Switch 3 will fare. Presumably Nintendo will be continuing their relationship with Nvidia for backwards compatibility reasons. With the other console manufacturers sticking to AMD's garbage, Nintendo may be able to low-key step over them completely with a DLSS5-lite. Nvidia did say they can run this on a single card with some cutbacks and a lot of VRAM.
 
This iteration is a bit wonky but can still look amazing. Assuming that the persistency of how things look is maintained throughout a game or if not this can be fixed with further iterations - future games will be able to look look photorealistic or like CGI with a fraction of the power used in the current best movie CGI.
 
Last edited:
Nah man. I'll just play retro/HD-2D games if this is where we're going with the tech. I'll save a lot of money.

DLSS 5 is for gooners who want the characters in every video game to look like the ads on porn sites.
 
Where's the "Wait and See" option?

It's too early to judge how this will really look when it's released later this year. Hopefully Nvidia takes some of the criticism they have received into considering and works on the training of the AI model they are using so it doesn't look like a SeeDance video
 
It absolutely won't be the future, because it's proprietary to Nvidia and likely exclusive to the 5000 series of cards.

Something like it (FSR, PSSR) may become widely adopted, but it's extremely unlikely that this tech will.

Shit, man, I'm not even sure Nvidia is going to release another generation of GPUs at the rate they are raking in AI cash.
 
Last edited:
It absolutely won't be the future, because it's proprietary to Nvidia.

Something like it (FSR, PSSR) may become widely adopted, but it's extremely unlikely that this tech will.

Shit, man, I'm not even sure Nvidia is going to release another generation of GPUs at the rate they are raking in AI cash.
Keep in mind that it has been largely nvidia continuing to innovate. I hate the pricing stuff but they have not fallen asleep at the wheel like intel did with CPU's. That's why they continue to have equity with me.

The engineers at nvidia GFX are second to none.
 
But Nvidia also clarified that developers have a large amount of control over the artistic side of the presentation. They can use masking on any part of the visuals which they don't want DLSS5 to touch. The ideal would be if developers offer the ability to disable it on faces via the graphics options, but leave everything else to DLSS5. Look beyond the faces in those Starfield comparisons and it's working some real magic in the environment. When used conservatively, this might be able to clean up some long-standing issues in rendering.
Here's the exact quote from https://nvidianews.nvidia.com/news/...red-breakthrough-in-visual-fidelity-for-games btw
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.

This list of their research labs is a bit freaky. They're really into this ai shit, huh... https://research.nvidia.com/research-labs#

Anyways, I don't like it because I think devs should be doing this stuff themselves, but at least this part of dlss is optional, so it's hard to dislike it that much.
 
Last edited:
It's the future and I don't like it.

Staying with Pathtracing DLAA Frame Gen on or Pathtracing DLSS Quality Frame Gen off.

A.I stuff would be way too demanding and expensive for me anyway.
 
Keep in mind that it has been largely nvidia continuing to innovate. I hate the pricing stuff but they have not fallen asleep at the wheel like intel did with CPU's. That's why they continue to have equity with me.

The engineers at nvidia GFX are second to none.

Agreed that it's Nvidia who typically innovates, but I think this is very different from DLSS/FSR. This heavily affects the final image and how the game's art is presented, and clearly is going to need to be integrated in the production pipeline early to ensure desired results are achieved. I'm just not sure most studios will be able to justify it for the small amount of GPUs in the market that support the tech. Clearly some will, especially those that Nvidia pays and/or offers engineering resources to support, but I can't imagine it gets widespread adoption unless an accessible platform agnostic version comes out, but it seems less likely that something will emerge unlike more traditional upscaling techniques.
 
Last edited:
Top Bottom