• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

DLSS 5 preview

Devs will have to be very careful about how they implement this
But that's the problem - Devs have currently, as per Nvidia explanations, miniscule amount of control. Unless one believes CEO of Nvidia, who claim it has gigantic amount of control - but that's directly contradicting materials posted by Nvidia. As if CEO has seen completely different thing than they shown the public, or he's talking total marketing rubbish as damage control.
 
What I would be more interested in seeing is if DLSS5 can fix the issues with facial animation, get rid of the mannequin effect and manage to make lips and all the incidental little muscle movements feel more lifelike, because no one has really cracked that (Naughty Dog getting closest).
How could it do any of that in current form, being just a glorified ai filter on final frames? If they change it quite a bit, maybe. In the form they shown? None of that is possible.
 
Interesting the Death Stranding comparison. Not a lot of artistry involved with the characters, they’re not the product of an artist (primarily at least) they’re actual scans of actual people (well known ones at that). Not denying they look good and are very effective mind.

What I would be more interested in seeing is if DLSS5 can fix the issues with facial animation, get rid of the mannequin effect and manage to make lips and all the incidental little muscle movements feel more lifelike, because no one has really cracked that (Naughty Dog getting closest).

Some missing the point though.Death Stranding sold 19 million copies and Hideo Kojima is far more experienced with game development than the pseudo experts in this thread.Mike York is also an animator who worked on Death Stranding 2 and multiple other games.

The fact his team and himself have found issues puts the bed the conspiracy theory that anyone criticising any issues with the current state of this tech is ignorant. Even DF in their recent video said the same thing. It definitely needs much more work on it!
 
Last edited:
But that's the problem - Devs have currently, as per Nvidia explanations, miniscule amount of control. Unless one believes CEO of Nvidia, who claim it has gigantic amount of control - but that's directly contradicting materials posted by Nvidia. As if CEO has seen completely different thing than they shown the public, or he's talking total marketing rubbish as damage control.

Things can change. Just like DLSS. If we was still on DLSS 1.0 hardly anyone here would be using it.
 
Some missing the point though.Death Stranding sold 19 million copies and Hideo Kojima is far more experienced with game development than the pseudo experts in this thread.Mike York is also an animator who worked on Death Stranding 2 and multiple other games.

The fact his team and himself have found issues puts the bed the conspiracy theory that anyone criticising any issues with the current state of this tech is ignorant. Even DF in their recent video said the same thing. It definitely needs much more work on it!
There are plenty of issues with it right now, is anyone saying there aren't?

I know in the internet age we are supposed to plant ourselves on one side or other of a fence and just throw poop at each other, but we're better than that on OCUK (most of the time) :)
 
  • Like
Reactions: TNA
Things can change. Just like DLSS. If we was still on DLSS 1.0 hardly anyone here would be using it.
Sure but that's not what Nvidia is saying now, people ask them all the time for clarification and they say the same thing all the time - it's just a filter, with just colour, blending and masking as control, nothing else is planned. But again, CEO claims it's a completely different system with full control of style and that it works on geometry level? But currently it doesn't even know what geometry is and definitely doesn't have style control :D Talking about confusion. Maybe he's thinking about completely different system than is planned for dlss5 or they planned different thing initially but current hardware couldn't do it so scaled back for release - who knows.

Point being, currently that corporation can't even agree with itself what it actually is and how it's supposed to work, so all we can do is just guess and imagine. However, I'll stick to what they shown and not what they rubbish in marketing words, as that's all we know for sure.
 
Last edited:
There are plenty of issues with it right now, is anyone saying there aren't?

I know in the internet age we are supposed to plant ourselves on one side or other of a fence and just throw poop at each other, but we're better than that on OCUK (most of the time) :)

Yes, unfortunately some on the Internet are pretty much trying to deflect from any criticism of this tech when its quite clear there are issues.

Any company will pretend it's new fangled widget is perfect - it's standard PR.

It reminds me of DLSS1 which had its set of issues and there was a vocal minority who got really riled up about it when issues were highlighted which was rather silly.

But clearly that feedback lead to DLSS2,DLSS3 and newer versions which have progressively worked on solving issues and become more usable.

I can see the potential of this technology,but it is early days. Nvidia also hadn't helped itself by being a bit opaque on what it is actually going on. It's only meant to be affecting lighting but even DF now has two of it's members saying more is happening. They had insider access.

So maybe Nvidia needs to perhaps make a clearer clarification on this. I also think this was previewed way too early.

They should have demoed this on a close to release version IMHO.
 
Last edited:
So will you need a 5090 for this
For a simple AI filter that it is shaping to be currently - likely not. Mobile phones these days can do face swaps live with sensible FPS, streaming software as well (Asians can for a while now change whole face age and gender whilst streaming live with superb quality). But if it's supposed to be actually good - then likely a beefy hardware with proper amount of vram will be needed, unless one wants to have all games with it looking very similarly.
However I understand the concerns that this may be taking away the artistic rights from the developers?
That's how it works as shown to the public currently - it replaces lighting and faces with what it learned from social media and stock photos, it seems, changing details about characters (plastic surgery, makeup, adding hair that didn't exist before etc.) - plenty of examples shown online from Nvidia own presentation. You might as well go to civitai or Instagram or FB, look at all the AI generated profiles that look completely fake, but so professional - you will get exactly the same effect in games with this. Unless they start training it on actual people and not AI fakery scrubbed from the internet. :)
If that's how it works. As for us if you don't like it just don't turn it on.
Same thing people said about upscaling and then FG and yet in many cases you don't really have a choice anymore, because they look and work really bad without it on mid range GPUs. Dlss 5 in current state lets publisher cut cost on development, using bad textures and faces etc. and count AI filter will fix it - that's why Nvidia already has big publishers on board with it, whilst developers working for those publishers already posted plenty online how much they got blindsided and shocked seeing what AI filter is doing to their games and how much it destroys their vision, plus that they didn't even know about dlss5 before it was announced. So no, developers aren't on board with this, publishers are. Very very different thing.
Be interesting though to see how much it hurts performance
A lot, judging by Nvidia admitting frame generation is always enabled with it to achieve playable FPS. It also requires as good lighting done by game as possible - they said PT is preferred for good effect. So, it's not replacing PT, it layers on top of it, killing performance even more.
 
Last edited:
But clearly that feedback lead to DLSS2,DLSS3 and newer versions which have progressively worked on solving issues and become more usable.
Dlss 2+ is a completely different thing than dlss1 was though, the whole dlss1 was scrapped and they redone it from scratch in a very different way - and that took a long time. This is possibly why the big discrepancy between what CEO and the engineers of Nvidia say about dlss5 currently, they describe 2 very different systems. We can only wonder which one will be the final one. The one they shown is horrible, the one CEO describes sounds much more interesting.
 
Last edited:
The fact his team and himself have found issues puts the bed the conspiracy theory that anyone criticising any issues with the current state of this tech is ignorant. Even DF in their recent video said the same thing.


It really doesn't. Everyone is largely ignorant of the tech other than nVidia and the handful of game developers & artists who've actually been able to implement it and work with it in person so far.

Everyone else, other game developers included, are all speculating based upon a small handful of clips about exactly what's going on under the hood, or how it looks and feels in person. What we do know is that the developers who have been able to use it and work with it, have either all come out strongly in support of it, or have been utterly trashing these ridiculous, hivemind responses; and in some cases very publicly.

Pointing to a single admittedly legendary developer, who has historically spent a small fortune hiring actors and implementing elaborate scanning techniques in order to create facial models for a small handful of characters, being not particularly excited over a technology that has the potential to make his techniques and edge largely obsolete overnight, is not really the kind of strong argument you think it is.

The fact that we have the fledgling shoots of a technology here that has the potential to be able to take the kind of facial graphics that Hideo has been able to with his budgets and experience (and far better, especially considering the environmental and lighting improvements), and make them immediately accessible to far smaller studios and independent developers, at a scale that can support expansive, densely populated game worlds across all NPC interactions, as opposed to a handful of expensively scanned ones, is quite a breakthrough.

Also...

That's a complete strawman. Absolutely no one has said there aren't issues or improvements to be made.

This is a technology like any other.

The pushback is regarding the ridiculousness of the brainded levels of immediate dismissal of the tech.
 
Last edited:
Yes, unfortunately some on the Internet are pretty much trying to deflect from any criticism of this tech when its quite clear there are issues.

Any company will pretend it's new fangled widget is perfect - it's standard PR.

It reminds me of DLSS1 which had its set of issues and there was a vocal minority who got really riled up about it when issues were highlighted which was rather silly.

But clearly that feedback lead to DLSS2,DLSS3 and newer versions which have progressively worked on solving issues and become more usable.

I can see the potential of this technology,but it is early days. Nvidia also hadn't helped itself by being a bit opaque on what it is actually going on. It's only meant to be affecting lighting but even DF now has two of it's members saying more is happening. They had insider access.

So maybe Nvidia needs to perhaps make a clearer clarification on this.
Well, the whole "it's only affecting lighting" is pretty vague. It's mean, it's affecting the final pixel output....everything that is visible is 'lighting'.

I'd really like to see a toned down version running against an already good realistic style source scene, and see if it can push it that much further forward, because I think the fundamental improvements to skin, hair, fabric rendering *are* an improvement.

Looking forward to giving it a whirl on some Metahuman faces when the SDK is out anyway....will be able to make up my own mind.
 
Sure but that's not what Nvidia is saying now, people ask them all the time for clarification and they say the same thing all the time - it's just a filter, with just colour, blending and masking as control, nothing else is planned. But again, CEO claims it's a completely different system with full control of style and that it works on geometry level? But currently it doesn't even know what geometry is and definitely doesn't have style control :D Talking about confusion. Maybe he's thinking about completely different system than is planned for dlss5 or they planned different thing initially but current hardware couldn't do it so scaled back for release - who knows.

Point being, currently that corporation can't even agree with itself what it actually is and how it's supposed to work, so all we can do is just guess and imagine. However, I'll stick to what they shown and not what they rubbish in marketing words, as that's all we know for sure.

Once it is out it will.be easier to point and laugh. Right now it is at 6-8 months away. It might even get delayed due to the reception it got.
 
Once it is out it will.be easier to point and laugh. Right now it is at 6-8 months away. It might even get delayed due to the reception it got.
It better be delayed if the current state is what they planned to release. :) And I'll be happy to test it at home for sure! But I'm a person who never uses mods in games that swap characters for "beautiful" versions, unlike what some do - that's likely why we have 2 camps about it, with the minority belonging to the char swappers.
 
Last edited:
Dlss 2+ is a completely different thing than dlss1 was though, the whole dlss1 was scrapped and they redone it from scratch in a very different way. This is possibly why the big discrepancy between what CEO and the engineers of Nvidia say about dlss5 currently, they describe 2 very different systems. We can only wonder which one will be the final one. The one they shown is horrible, the one CEO describes sounds much more interesting.

But that is a point in hand. They had to scrap it because it had big issues and the approach used by DLSS2 onwards worked far better. That only happened because of feedback on its functionality.

Maybe,what JHH is describing is his ultimate vision of where this is headed. It wouldn't surprise me if the engineers were given a GTC target to make the demo and had to push out whatever they could.

We see this with games all the time,when they are released half broken,because PR has come up with a shipping date which doesn't align with what the engineers need.


Well, the whole "it's only affecting lighting" is pretty vague. It's mean, it's affecting the final pixel output....everything that is visible is 'lighting'.

I'd really like to see a toned down version running against an already good realistic style source scene, and see if it can push it that much further forward, because I think the fundamental improvements to skin, hair, fabric rendering *are* an improvement.

Looking forward to giving it a whirl on some Metahuman faces when the SDK is out anyway....will be able to make up my own mind.

I increasing think an aribitary GTC deadline was given for Nvidia to preview it and their engineers had to push to release "something" to demo. There is unfortunately increased pressure from investors for companies to do this.

It better be delayed if the current state is what they planned to release. :) And I'll be happy to test it at home for sure! But I'm a person who never uses mods in games that swap characters for "beautiful" versions, unlike what some do - that's likely why we have 2 camps about it, with the minority belonging to the char swappers.

Some of the most downloaded mods for Skyrim and Fallout are the character mods. But even I use them because the original models are terrible - visible neck seems,square feet and other weirdness. Also,very few actual variations in NPC designs(its a world of clones apparently). Also some of the animations are also fixed.
 
Last edited:
Everyone else, other game developers included, are all speculating based upon a small handful of clips about exactly what's going on under the hood, or how it looks and feels in person. What we do know is that the developers who have been able to use it and work with it, have either all come out strongly in support of it, or have been utterly trashing these ridiculous, hivemind responses; and in some cases very publicly.
Handpicked developers by Nvidia themselves support the tech, that they were given exclusive access to.
Really? You don't say. wow... I wonder why?
 
Last edited:
But that is a point in hand. They had to scrap it because it had big issues and the approach used by DLSS2 onwards worked far better. That only happened because of feedback on its functionality.

Maybe,what JHH is describing is his ultimate vision of where this is headed. It wouldn't surprise me if the engineers were given a GTC target to make the demo and had to push out whatever they could.

We see this with games all the time,when they are released half broken,because PR has come up with a shipping date which doesn't align with what the engineers need.




I increasing think an aribitary GTC deadline was given for Nvidia to preview it and their engineers had to push to release "something" to demo. There is unfortunately increased pressure from investors for companies to do this.



Some of the most downloaded mods for Skyrim and Fallout are the character mods. But even I use them because the original models are terrible - visible neck seems,square feet and other weirdness. Also,very few actual variations in NPC designs(its a world of clones apparently). Also some of the animations are also fixed.

You are wrong CAT. Tinek doesn't like or use them, therefore they should not exist!
 
Handpicked developers by Nvidia themselves support the tech, they were given exclusive access to.
Really? You don't say. wow... I wonder why?

:rolleyes:

That's just mental gymnastics being employed in order to handwave away the opinions of the developers and artists who've actually worked with it.



If you have some evidence to support the idea that the only developers and artists to have actually seen and used it in person, are tweeting about it positively because they're being paid to, then show us.

Otherwise, it just sounds a bit silly:

"Wahhh the artists and developers who've actually used something disagree with me about it, they must be shills...."
 
Last edited:
:rolleyes:

That's just mental gymnastics being employed in order to handwave away the opinions of the developers and artists who've actually worked with it.



If you have some evidence to support the idea that the only developers and artists to have actually seen and used it in person, are tweeting about it positively because they're being paid to, then show us.

Otherwise, it just sounds a bit silly:

"Wahhh the artists and developers who've actually used something disagree with me about it, they must be shills...."
Do you honestly think that devs handpicked by Nvidia, to test their unreleased product that they are going to use to justify why you need to buy their next GPU, would say anything bad about the product publicly? Are you really that Naive?
 
Do you honestly think that devs handpicked by Nvidia, to test their unreleased product that they are going to use to justify why you need to buy their next GPU, would say anything bad about the product publicly?

The individual developers and artists aren't handpicked for crying out loud. These are are individuals who work for companies who make the games it's been tested in so far. They're not executives making company PR statements, they're artists and experts discussing it on Twitter. :cry:

Do you really think that multiple individual artists & developers, who've actually seen and used the technology, would publicly go against this egregious onslaught of mob stupidity and put themselves in the firing line, especially after it's become so extreme that it has resulted in Digital Foundry receiving death threats; just to engage in twitter debates in an attempt to explain that they were in fact extremely excited about what they've been able to see and use, if they didn't genuinely believe it?

Good grief man...

Are you really that Naive?

Are you champ?

You're inventing a conspiracy based upon no evidence whatsoever, in order to avoid confronting the possibility that some artists and experts who've actually used something, disagree with you.
 
Last edited:
Back
Top Bottom