Deep Breath: Okay, Let’s Talk About That Controversial DLSS 5 Demo
from the here-comes-the-comments dept
The polarization over any and all uses of artificial intelligence and machine learning continues. And, to be clear, I very much understand why this is all so controversial. Any new technology that has the chance to be transformative will also necessarily be disruptive and that causes fear. Fear that is not entirely unfounded, no matter your other opinions on the matter. If that’s you, cool, I get it.
I’ll start this off by pointing to the latest edition of the Techdirt podcast in which both Mike and Karl engaged in a fantastic discussion about the use of AI. I’ve listened to it twice now; it’s that good. And, while I found myself arguing out loud with the both of them at certain points during the podcast, despite the fact that neither of them could hear my retorts, it presents a grounded, often nuanced conversation, which we need much more of in this space.
And now, in what might be a subconscious attempt by this writer to commit suicide by comments section, let’s talk about that controversial demo of NVIDIA’s forthcoming DLSS 5 technology. What DLSS 5 does compared with previous versions of the technology is indeed new, but what is not new is the introduction of AI and machine learning into the equation. DLSS 2 and 3 had that already, in the form of pixel reconstruction and frame generation. DLSS 5, however, introduced what is being labeled as “neural rendering”, which uses machine learning to alter the lighting and detailed appearances in environments and, most importantly, character rendering over the engine on top of the 2D image output. Here’s the video demo that got everyone talking.
The backlash to the video was wide, immediate, and furious. There was a great deal of talk about the alteration of artistic intent, about whether this changed what the original developers were attempting to portray when they created the games, and, of course, industry jobs. I want to talk about the major complaint pillars seen across many outlets below, but this backlash also supposedly came with death threats foisted upon NVIDIA employees. I would very much hope we could all at least agree that any threats of that nature are completely inappropriate and absurd.
With that, here is what I’ve seen in the backlash and what I’d want to say about it.
Get your damned AI out of my games!
Perhaps not the most common pushback I saw in all of this, but a very common one. And a silly one, too. As I mentioned above, DLSS versions already used some version of AI and machine learning. That isn’t new. How it’s applied is certainly new, but that isn’t the same as the demand to keep AI entirely out of the video game industry.
And if that’s where you are, go ahead and shake your fist at the clouds in the sky. AI is a tool and, as I’ve now said repeatedly, the conversation we should be having is how it’s used in gaming, not if it’s used. That’s because its use is largely a foregone conclusion and it is an open question as to whether its use will be a net benefit or negative overall to the industry. Dogmatic purists on AI have a stance that is understandable, but also untenable. We’re too far down this road to turn around and go home. And if the tech were able to lower the barriers of entry to the gaming industry, acting as the fertilizer that allows a thousand indie studios to sprout roots, would that really be so bad for the gaming ecosystem?
I can appreciate the purists’ point of view. I really can. I just don’t see where they have a place in the conversation when it comes to gaming.
It overrides artistic intent!
Does it? If it did, then hell yes that’s bad. But if it doesn’t, then this concern goes away entirely.
DLSS 5 is built with options and customizable sliders for game developers. That’s really, really important here. At the macro level, a developer that has decided to use DLSS 5, or decided and customized how it’s used in their games, is exercising consent over their products. That should be obvious.
But then we get into really interesting questions of art, the actual artist, and the ownership of that art, because those last two are very different things. As Digital Foundry outlines:
It may even raise consent and other questions surrounding artistic integrity. On site and witnessing the demos in motion, concerns about this seemed less of a problem when the games we saw had been signed off by the studios that made them – the contentious assets we’ve seen, likewise. Nothing from the DLSS 5 reveal released by Nvidia has not been approved by the studios that own those games. But perhaps the issue isn’t just about specific approvals by specific developers on agreed DLSS 5 integrations, but rather the whole concept of a GPU reinterpreting game visuals according to a neural model that has its own ideas about what photo-realism should look like.
While we’ve seen endorsements from Bethesda’s Todd Howard and Capcom’s Jun Takeuchi, to what extent does that consent apply to the entire development team and other artists associated with the production? And by extension, there is also the question of whether now is the right time to launch DLSS 5 at a time when the games industry is under enormous pressure, jobs are on the line and cost-cutting is a major focus in the triple-A space. The technology itself cannot function without the work of game creators – it needs final game imagery to work at all – but the extent to which it could be viewed as a worrying sign of “things to come” cannot be overstated bearing in mind the reactions elsewhere to generative AI.
That strikes me as a valid and interesting ethical question when it comes to the use of this technology, but one that is probably overwrought. Individual artists who work on video games already have their artistic output live at the pleasure of the game developers they contract with. Those developers already can use this game art in all kinds of ways that the individual artist may not have had in mind when creating it, or indeed have even considered such possibilities. DLSS 5 is just one more version of that, with the main difference being that it involves AI making changes to game images. That’s an important thing to consider, sure, but there are cousins to this ethical question that we’ve all come to accept already. This strikes me more as part of the “all AI is bad all the time” crowd finding a foothold in something other than dogma to grab onto.
Developers and publishers own their games. If they want to use DLSS 5 in those games, there is little other than specific work for hire or other contractual stipulations with individual artists that would keep them from implementing it. If artists don’t like that, I completely understand that point of view, but that’s what contract negotiations and language are for.
Bottom line: I have been as vocal as anyone arguing that video games are a form of art for well over a decade now and I struggle to agree that an optional technology that has approved buy in from game developers and publishers equates to “overriding artistic intent”, writ large.
The faces in these examples look like shit, are “yassified”, or suffer from the uncanny valley effect!
Look, here we’re going to get into matters of opinion. I have to say that when I viewed the demo video myself, I had the opposite reaction. And, yes, this opens me up to claims that I am somehow a massive fan of AI-created pornography (this is where the yassified comments come in), or that I just want all the characters to look “hot” (I’m too old for that shit), or that my older age of 44 means I’ve lost touch with what video games should look like. Despite my genuine respect for the dissenting opinions here, allow me to say this: bullshit.
The caveat to all of this is that the demo revealed very little in the way of this technology working within these games in motion. It’s also certainly true that NVIDIA chose the best potential images to show off its new technology. If the DLSS 5 rendering sucks out loud in a larger in-motion game, or if the images it creates end up being inconsistent throughout gameplay, or if it does just end up looking shitty, then I’ll be right there with you with a torch and pitchfork in hand.
And here’s the other thing to consider with this particular complaint, combined with the previous one about artistic intent: do any of you use visual mods in your games? I do. A ton of them. For a variety of reasons. I have used them to alter the faces and models for games like Starfield and Skyrim, among many others. Do I need to feel bad for altering the artist’s intent? Do I need to apologize for incorporating mods to make characters and environments appear in a way that helps me better connect with the game I’m playing?
Because I’m not going to do either. And I don’t expect you to. Nor do I expect game developers that choose to use this optional technology to beg for forgiveness for their own output.
The hardware demands to run all of this are insane!
Fine, then you’ll get what you want and nobody will be able to use this technology anyway. But I don’t think that will be the case. NVIDIA knows what it will take to run this tech once it leaves the demo stage and goes into production. The idea that they would hype up technology that nobody can use strikes me as unlikely in the extreme.
Conclusion: everyone take a breath
This still strikes me as more of a “all AI is bad” crowd grasping at lots of other things to buttress their pushback than anything else. AI has plenty, plenty of potential pitfalls. Worried about jobs in the gaming industry and elsewhere? Me too! But if you’re not also looking at the potential upsides for the industry, then you’re engaging in dogma, not conversation.
Will DLSS 5 be good? I have no idea and neither do you. Will DLSS 5 alter previously released games in a way that fundamentally alters how we play these games? I have no idea and neither do you. Will it negatively impact the gaming industry when it comes to the number of jobs within it? I have no idea and neither do you.
This was a tech demo. Details on how it works are still trickling out. Most recently, there has been some clarification as to the 2D rendering nature of the technology and what that means for the output on the screen. As an early demo of the technology, feedback is going to be important, so long as it’s informed and reasonable feedback.
The technology may end up being trash and hated for reasons other than “all AI is bad all the time.” If that ends up being the case, I trust the gaming market to work that out for itself. But a lot of the hand-wringing here looks to me to be speculative at best.
Filed Under: ai, developers, dlss 5, rendering, video games
Companies: nvidia
Comments on “Deep Breath: Okay, Let’s Talk About That Controversial DLSS 5 Demo”
What’s absolutely absurd to me is how it takes a truly ridiculous level of hardware, two of the priciest graphics card on the market, just to essentially do rotoscoping that eradicates details like the fog, or the realistic light scatter from the path tracing.
Re:
That’s the whole stupidity of current AI market, following some experiential Moore’s law, and living like if technology as ten times faster every two years.
Every time Nvidia announces an new architecture, it cannot stop bragging about what the next architecture (announced one year later) could be able to do, and all hyperscalers buy thousands of them hopping they’ll be soon obsoletes because future generations will be much faster (pretty much like in 2000s when we’ve got the first 5GHz CPUs and thinking than by 2010 we’ll get 10GHz ones).
What is the most worrying is if this technology gets some traction (and AMD/Intel offers a similar functionality), what video games will look like without DLSS 5, or with a cheap GPU?
DLSS before 5 was attempting to render the output at a higher framerate with minimal change from what rendering without it would produce.
DLSS 5 makes changes, like putting lipstick on all the women, and the demonstration needing dual 5090s makes it sound like it’s going to come with a significant reduction in framerate.
It’s like Nvidia has inverted every target they had for prior versions of DLSS and are surprised that people don’t like it.
Re:
The glossy lipstick and lip filler on all characters is an interesting touch
Re: Re:
Really shows what they trained on
Re: Re: Re:
Mar-a-lago face?
Re:
It’s worth pointing out that when nVidia introduced real-time RT back in 2018, their tech demo was running on four Quadro GV100 GPUs. Those weren’t exactly consumer-grade cards.
I don’t know why people who see this expect it to be able to run well on something like a single RTX 5080.
I do strongly agree that previewing this as the next iteration of DLSS was a mistake.
Re: Re:
The difference is that RTX was always sold on the premise that the reduced framerate was a worthwhile cost for the benefits of the improved lighting and being way less work for devs than prior lighting methods. To the point that people put RTX into Doom just because they could.
While DLSS was always sold on the benefit being the improved framerate.
So, both ultra expensive and ultra low quality. That’s a real lose-lose.
I think your over-generalizing the pushback. Yes, there’s some of that, but there are a lot of people pointing out that the filter – and it is a filter at the end of the day – overrides the artistic intent of the character faces. A cartoon face does not need to have a filter that puts in peach fuzz hair and pores on the skin.
One of the examples that gets used a lot is the Resident Evil examples in the tech demo. The characters have a certain look to them that blends well with the overarching themes. The filter completely overrides it and makes one of the characters turn into a super model with the heavy amounts of air brushing that got applied (and likely part of the meme of photoshopping monsters and male humans and turning them into super models with large boobs with the label “DLSS5 Turns on”).
The developers of these games are also split on this. Some are OK with it while others are not.
I think there are plenty of people that looked at the tech demo and concluded that the faces looked ugly. Are some of them part of the “all AI is bad” crowd? Sure. Are all of them part of that same crowd? I sincerely doubt it.
The only faces in that tech demo that looked like it was improved were the examples from Starfield. Even then, it was a very moderate improvement and nothing revolutionary.
I don’t think it’s speculative to say it’s worse when people are reacting to what they see. If there was rumours of DLSS5 adding a filter onto things and no tech demo out there, then yes, you’d have a point that it’s speculative. The reality is that there is a tech demo and people are seeing what it is supposed to be doing with their own two eyes. That’s not speculating, that’s reacting – and the reaction was negative.
Anyway, I think you are painting too many people with the same brush in a bid to dismiss the criticism outright. The reality is that there is nuance to the pushback that I think you are glossing over.
Re:
Yeah, that’s the funny thing about this: The filters, as displayed in the video, ignore the context of the games (because how could they know them in the first place?) and “yass-ify” everyone. Grace Ashcroft is an FBI analyst who seems to have little-to-no field experience; her role is to be an “everywoman” who would react, at least in this context, like the average person would if they were to encounter horrors like the ones in this game. Her appearance being more “average” instead of more conventionally beautiful (e.g., Jill’s appearance in the RE3 remake) plays into that context. The filter destroys that context by making her look more conventionally attractive. That is why people are pissed off about that specific example: Love her or hate her, Grace’s entire character is built around her being an average person, and “yass-ify”ing her ruins that context for the sake of appealing to the kind of backwards-ass knuckle-dragging chuds who think she’d be hotter if she put on more makeup and smiled more.
Re: Re:
But again, if this all optional for the developer, or refinable through prompts to meet the artistic vision, your argument is with the game maker, not NVIDIA….
Re: Re: Re:
Therein lies the issue: The examples in that video were created without the knowledge of the developers (or at least the ones at Capcom), which means this whole thing comes off as NVIDIA trying to manufacture consent for the feature. That NVIDIA later said “oh, but this will all be under the control of the developers” means they see the work supposedly done by this feature they made as extra work for the developers, which also isn’t a good look. Kind of a lose-lose situation here.
Re: Re: Re:2
If you have “You can turn it off” as one of your product’s selling points, you have a shit product.
This is a repackaged version of the “It’ll democratize things!” talking point that people used for cryptocurrency and NFTs.
If you want takes from actual journalists who know things about games and the industry, Techdirt is the wrong place to find that. Go read and support Aftermath and Remap instead.
Maralago-face filter aside...
From what little Nvidia seemed willing to show when it came to this tech demo, I thought that movement seemed to suffer from horrible temporal issues. It would seem that the Oblivion remaster and Starfield has TAA, but I couldn’t find anything on Saudi Arabia’s FC franchise. The temporal artifacting seemed bad and I can’t imagine it would translate to a decent experience in it’s current state.
Regarding the “artistic vision”: Maybe it was rushed, maybe it’s because it’s early, but some of the drastic lighting changes from the Assassin’s creed game as well as Starfield just seemed incredibly drastic. It almost seems at odds with what you would expect from, what in past iterations, was a pretty ‘straight forward’ upscaler that didn’t change the scene a great deal. It strikes me as a technology that probably shouldn’t of been called DLSS.
I am glad we’ve been getting more use out of the “made with activists in mind/Made with gamers in mind” memes laughing at the Gamergate crowd, though.
Honestly I’m just enjoying the DLSS5 memes lol
Honestly they can do whatever, my main issue is just how stupidly expensive PC gaming is getting because of AI.
I sometimes wonder if the people who like this slop have fucking eyes.
Like it’s immediately, obviously, at-a-glance worse, and they’re putting it out as a tech demo.
Finally a fresh breath
Lately, I’ve been seeing a lot of posts where AI is vilified and dogmatically rejected as inherently bad, and frankly, I struggle to understand why.
Sure, there are many potential misuses of AI, as with any other technology. Perhaps even more so, given how it blurs the uncanny valley and makes people question whether what they’re seeing is real or fake.
But in the game industry, where the goal has always been to immerse players in a fantasy world or situation, and where publishers aim for a specific look, mood, or ambiance, the use of AI feels like just another tool in the creative toolbox. After all, many tools already exist that can accomplish in moments what would have taken thousands of hours of manual labour just a decade ago, often replacing skilled work in the process.
Dismissing the use of AI outright won’t work; the genie is out of the bottle. What we can (and must) do, I believe, is guide how they are used. That means having thoughtful discussions about when, where, how, and even whether they should be applied, rather than resorting to blunt rejection.
Re:
Can you spot the seven words up there that explain why people are pissed off about generative AI and the C-suite push to use it as much as possible?
Re:
Maybe because you are dogmatically assuming a false dogmatism from detractors.
Not that there aren’t domatic bandwagon people everywhere, but if you’re using that to dismiss al criticism (and ignore 90% of real criticism, like the rest of the people being intellectually dishonest about “having nuanced conversations” surrounding the issue), then you can be equally dismissed without nuance or argument as well.
But go on, explain how the ganing industry was ever at all better for more “realism”. Why do i need not only superhigh graphics, let alone AI, to play Cult of the Lamb? Graphics hardware demands have already been long absurd because lazy-ass code and “hey, there’s this hardware available, and an SDK”.
‘The idea that they would hype up technology that nobody can use strikes me as unlikely in the extreme.’
Did you just pop out of a time machine from the 1920’s? How often have we been pitched awesome new tech that’s only a few years away that isn’t? Elon Musk is approaching being a trillionaire based on the imagined value of his tech announcement company because he uses the announcement of things to bump the share price.
The reason people are cynical this will make it to market any time soon is based on the actions of Nvidia, they have spent the past decade fuelling a shortage of their own product, prioritising gambling on Crypto, now AI, being part of the OpenAi and Oracle ‘investment’ ouroboros. People cannot afford or find what they are putting out now, and have no reason to believe the cards needed to run this will be available in large quantities any time soon and at an affordable price.
Re:
I am reminded of how yesterday I saw something about the US government pulling subsidies from existing renewable energy sources in favour of funding a fusion startup. The power source that is truly the Iranian nukes of power generation, as it’s been pitched as ten years away for the last 40 years of my life.
Re:
That struck me too, that’s up there with “But why would the police lie?” for delusional takes.
I have seen principled takes defending AI, and I have seen takes that require an uninformed reader. This is squarely the latter.
Has anyone checked Timothy’s stock ownership disclosures?
It doesn’t help when Nvidia continues to lie about what it is doing. For example the repeated assertions that it works on the geometry level and not the frame. But then when pushed they say, “well yeah it modifies the frame but the frame is defined by the geometry so we weren’t lying.” Except they were and are.
My wife (a [very online] fan of AI) was quite surprised when she found out that a large majority of gamers seem to oppose AI usage in games, and that game makers often feel obligated to proclaim loudly that “AI isn’t welcome in this house”.
I suggested that this is largely due to the impression that, like candle-makers and horse-buggy repairers, certain people may very well be out of jobs and that (“naturally”) means that evil rapacious publishers, blood dripping from their faces, are smiling gleefully.
The current very-online zeitgeist trends ignorantly to what the kiddies inaccurately term “socialism” which to their minds means that a company is making an absolutely evil profit from the paid labor of others and that such nefariousness cannot be allowed to go unchecked.
After I connected these facts (each of which she was already aware), she understood where these people were coming from, but of course still identified the problem: “so gamers don’t want better-looking faces, faster frame-rates, and who knows what to come, because someone might make more money and someone might get fired”. As usual, she’s correct.
Like everything else these days, this particular issue has created die-hard tribes who would rather perish on their swords than change their minds.
Re:
I mean, yes, that is the gist of it. But how is that a bad thing?
Graphic fidelity can only go so far before it hits a wall, and I’d argue that we’ve already hit that wall. How much better can a game like Resident Evil Requiem even look when it’s already one of the best-looking games ever made? How much more “realistic” can graphics get when they’re already about as realistic as they’re going to get?
This sentence comes to mind right now: “I want shorter games with worse graphics made by people who are paid more to work less and I’m not kidding.” And yes, I think anyone who isn’t deep into the “realism is always better” rabbit hole would agree with me: While there is a place for “Triple A” experiences, the industry would be a lot better off if there were more games developed by smaller teams to be smaller experiences instead of cash cow “forever games” or 200-hour experiences that cost $80 or more.
Graphics are not the end-all be-all of gaming. Some of the greatest games of all time have graphics that young gamers these days might describe as “retro” or “ancient” or “ugly”. While good graphics certainly helps, the overall design of the game—art style, gameplay, music—is what ultimately makes those games great. The difference between Metacritic scores for, say, The Legend of Zelda: Breath of the Wild and Crimson Desert can be attributed to that point: Crimson Desert has arguably better (or at least “more realistic”) graphics than the other, but Breath of the Wild has a higher score because people liked its overall design (including the gameplay loop) better.
Fealty to graphical fidelity—especially “realism”—is a mistake. Trying to “improve” on that fidelity when it’s already hit a level where any “improvement” will only be noticeable to hardcore graphics nerds is a bigger mistake. And yes, I don’t want better-looking faces, faster frame-rates, and who knows what to come if it means some C-suite asshole will make more money and the people who really do the work to make these games lose their jobs. Why should I want “better-looking” games at the expense of wrecking the game industry and making its lowest-paid workers potentially homeless?
Re: Re:
The fun detail I see from people who are claiming that DLSS5 produces more “realistic” graphics is that they refuse to even attempt to defend it putting lipstick on all the women.
Re:
I’ll take ThingsThatDidntHappen.txt for $1000, Ken.
Re:
The non-sociopathic take is to prefer people NOT get fired.
Re: Re:
I always help artists by pouring tar over their paintings. It will take many hours to salvage and repair their paintings. It keeps people from getting fired and might call for more entry level positions removing tar from painting carefully. Remember, the most important thing in the world is jobs. /s
Re: Re: Re:
When you can’t morally argue against something, just make up a completely different scenario, it’s just that easy.
Re:
Your wife is correct in her assessment, but wrong in her judgement.
Unironically: “I want shorter games with worse graphics made by people who are paid more to work less and I’m not kidding.” Gen”AI” IS going to get a lot people fired and make a lot of money for the people killing this industry. I don’t care how pretty they make the corpse.
Re:
Why is realism being pushed so hard as “better looking” lol. People still love the Pokemon pixel art, Okami remains a classic that people are still buying and playing, and do you have any idea how many players Genshin has right now? Disco Elysium’s art direction was incredible as well. Style always wins over skin pore count lol.
Also you are really over generalizing why gamers think dlss5 is garbage. Personally, I am just tired of how this company is scrambling for anything it can do to inflate its value before doing what gamers REALLY want: affordable gpu. I don’t want to blow through all my fun-money in one go next time i decide to upgrade my PC.
Re:
Personally, I don’t think the AI faces look better, but I also don’t like higher-fidelity faces in general, and these are even more uncanny and less realistic than those. As for faster frame-rates, I legitimately don’t care. If it consistently hits ~60+ fps with minimal slowdown, that’s good enough for me. In fact, a consistent framerate is more important than a high framerate if only because I can adapt to a low framerate. (It’s why, when the option is available, I always cap the FPS.)
Seriously, AI doesn’t objectively improve these things, and these aren’t necessarily things that need improvement.
As long as it’s opt in they can do whatever they want. That’s the problem. Gemini/Copilot at all were more or less optional at some point. It’s getting harder and harder to get rid of that shit.
From my perspective it’s a “slopfication” technology that, like all slop generators, uses too much power to produce mediocre results at best (for general purposes, there are very nice uses for LLMs and other generative “AI” in specific workloads). Maybe it will be better with more iterations/optimizations? Maybe. But I don’t see any benefits from needing another 5090 to produce demonstrably worse results.
I think that TechDirt suffers from being anti-anti-AI.
Nothing is inevitable, resistance isn’t futile.
The largest platform for PC games requires you to disclose usage of genAI due to gamers complaining, which they are very good at (complimentary). NFT was not inevitable, the metaverse(tm) did not happen.
Virtually everything TechDirt writes about (e.g. privacy) requires more complaining about powerful forces that think whatever they want is inevitable.
It also works for the bad people, there are more people complaining about trans women in women’s sports than people caring about women’s sports.
A large company puts a tech demo that is universally panned, using a brand name their customer like, putting out contradictory statement, and then the CEO goes to talk to a professional idiot. In every other case, TechDirt would have dunked on the company some more.
Maybe game developers were involved, but artist weren’t.
I suggest Daniel Owen’s recent YouTube videos to those who want more on the issue with some technical details. One interesting point he mentions, is that if this works, developers might change graphics to give better inputs to nVidia’s card, which would mean entrenching their monopoly on GPUs, since DLSS is a propriety technology.
The problem I have with the “optional” stuff is that it’s kind of a red herring. Even if something can be turned off, if the vast majority of people don’t, it’s kind of a moot point. Defaults/human behavior matters, and it’s not unreasonable to factor that in.
And that’s also not getting into if Nvidia doesn’t hold to the promise (e.g., what if you have to use DLSS5 to get the other DLSS features in the future?), or the softer incentives- if it takes 10x the work to get the same result without DLSS, is it really a choice at a for-profit company? Incentives matter.
I have a hard time taking this at face value. This blog is basically built to shake fists at quixotic battles like copyright, and that never stops it. Why should this be different?
Maybe controversial, and it’s too long to get into, but I think there are very different standards for how I treat mods/player choices, vs a company like Nvidia.
Ok, but like…what’s the actual argument? A lot of it is subjective, but I don’t think you can really argue that it doesn’t share a lot of the features of e.g. AI ads. Maybe you like the style, but it’s certainly there. There are also already existing objective issues in the footage we saw, with e.g. unrealistic lighting. (for the record, I do think this will get fixed in time)
This is true, but one thing I think is missing is that these are asymmetric. Maybe the job stuff all does work out- great! Upsides exist, but the risks weigh towards focusing on the downside, even if it’s speculative. And that warrants more than a few throwaway lines.
Why? You cover so many things the market doesn’t work out, and this fits that. The market has never prioritized things like artistic intent or not exploiting workers over profits.
Re:
It kinda sounds like the market gave its feedback–i.e., this tech sucks and we don’t want it–and he doesn’t seem to be willing to take that at face value. “The market will work it ou…no, not like that!”
I’m IN the industry. There are no upsides to this. Nearly every bit of the demo looked worse. You know how MKBHD does a camera showdown every year, and the brightest photos are the ones that win? That’s what happened in this demo. But if you pay attention, you’ll see that all that happened was that instead of light coming from a point source, it was averaged out and came from everywhere. Shadows disappeared. Textures became flat and average.
You can argue the faces were more DETAILED, but I’d hesitate to call that an improvement.
Look, the last few years of gaming have actually shown that smaller games with more care and attention and vastly simpler graphics punch well above their weight. Hades 2, Silksong, Slay the Spire 2–no cutting edge graphics, just games that people like playing.
The AAA industry still exists and lumbers on. I was part of it for literal decades. But they don’t make the games people want to play, which is why Ubisoft’s stock price is down 90% from its peak. EA is a leveraged debt-monster now. These companies have been so out of touch with what players want for so long, and now we’ve got a new thing that they don’t want: DLSS 5.
And, amazingly, they told Nvidia, “hey, this sucks! Stop it!” before it even shipped. Real-time user feedback, and everyone seems hellbent on throwing it away even though it could save them a lot of money. But no, they’ll plow on through and then when the next AAA game ‘inexplicably’ fails, they’ll say, “who could’ve predicted this? It was unknowable!” and memoryhole the entire wasted endeavour.
The original DLSS was about making cheaper computers perform better. DLSS 5 is about extracting money from gamers because otherwise it’s a low-margin market and they may as well just sell more GPUs to OpenAI.
Games are by humans, for humans, let’s try not to forget that. You want to give my artists a tool (the artists are game devs too, btw, they’re not ‘contractors’. They’ve worked the same hours and the same crunch that I have) that makes it easier for them to realize their vision? Great, I’m all for it. Put that hardware on their desk, let them create, and let us publish a game. Then leave the finished product alone.
Valid dissent is impossible
I’m frustrated that the author has apparently decided that it’s simply not possible to make a valid criticism of a product which uses AI.
Nvidia released a tech demo of a new graphics product, presumably putting its best foot forward when doing so.
The overwhelming public consensus was that the images produced by this new tool just looked bad.
The author’s takeaway isn’t that Nvidia did this tech demo too early, or that it misjudged what consumers wanted, or that it just doesn’t do a good job of what it’s supposed to do, but instead that people who don’t like the tech demo are all knee-jerk AI opponents.
Seriously, where is this discussion supposed to go if the AI boosters just preemptively declare any criticism to be irrational hatred of AI?
Re:
Welcome to a Timmy Geigner post. First time?
Re:
I also thought this article was very reactionary and strikes me as defending NVIDIA for something they don’t need to be defended from. If they eventually make a product that benefits game developers and game players, then great. In the meantime, the criticism about image quality, artistic control and feasibility are entirely valid. The repetition of the author here saying “I don’t know and neither do you” about how this will be used in the future really seems antithetical to the whole idea behind doing a demo in the first place. Why even post this article if no one should speculate about anything regarding this technology? Why even do a demo if you don’t want anyone talking about the technology?
I think one issue is calling it DLSS 5. DLSS is originally a anti-aliasing/denoising/upscaling technology meant to transparently enhance image quality and framerates. This does something entirely different. The fact that so many people have different opinions about if the results they presented are better or worse than the original means that this is not a strict enhancement technology. I do think this technology could be useful in “style transfer” type applications such as adding a toon shader or similar features to a game, but the idea that this would be enabled by default as a thing most people want (as DLSS is) no doubt contributed to the criticism.
So, this is really really sensitive, so I want to break this off into another comment in the hopes for some nuance beyond just “herr, you condone it”. And to be clear, I condemn this. And I also acknowledge that people on the internet can be inappropriate manchildren.
That said… how exactly are people supposed to react to this sort of thing? It’s easy to say it’s just a game and people need to get with progress but these things affect people’s lives. But we don’t live in a country with a robust safety net; this can literally be life, death, homelessness, lack of healthcare, etc (and this damage has happened in previous technological shifts). It’s easy to rote condemn it, but I think you’re underselling how much of a powderkeg we’re sitting on. Especially given the hype that bosses (not to mention fascists) have made clear they want to leverage, and that it’s Nvidia in particular, and the lack of a voice in policy people feel.
Again, I’m not condoning it, but I do think there’s a bit of a double standard worth examining. We condemn threats, but putting people out of a job, and the threat that comes with that, is treated as somehow just business? And that’s a little weird. I don’t think you can acknowledge a potential for mass unemployment and expect people to take it laying down.
Re:
That was a well said, clearly thought out comment.
I still land on “We shouldn’t threaten to kill people, even over our lost jobs, if that happens.”
That is a hill that I am, ahem, willing to die on.
Re: Re:
Don’t threaten to kill people. It just warns them.
Re:
Good argumentation. I do think death threats are not acceptable.
But what do we do when billionaires are threatening billions with death with their actions and ways of thinking? When they adopt ideologies that clearly and unequivocally advocate for the death of those deemed unfit, different? Musk has clearly shown he’s a nazi, supremacist asshole. The other billionaires mostly joined the fascist bandwagon. JK Rowling uses her money in a crusade against trans people. Plenty of examples of super rich being assholes and using their money to effectively write or modify laws to screw people directly or indirectly.
I don’t think love and flowers will stop these people.
Re: Re:
I should add that forcing stuff users don’t want onto them is just a symptom of this larger problem.
Simply describing a dislike of the visual changes as an opinion is reductionist. There are plenty of issues that present more obvious technical problems beyond a matter of taste. The scenes all look front-lit, like there’s a giant ring light behind the camera. Facial features are being changed, not simply adding definition, causing characters to look like different people. Virgil van Dijk is a real life dude and DLSS5 turns his model into another person entirely. Little details are being removed or hallucinated into reality. You raise some future hypothetical about motion, but you can already see the soccer ball turn into a hypersphere in the initial video. We know it’s bad now, no need to defer judgement.
I started following this site recently because you’re willing to be passionately critical of issues around tech. While I agree some of the AI discourse does dip into moral panic territory, so too do so many other issues you talk about. This is the second article I’ve personally seen where you try to appeal to some kind of centrist position on AI, which seems uncharacteristic to the rest of your content.
Re:
That’s going to be fun for lawyers when it comes to games where the likenesses of people and/or objects are licensed.
I don’t think those are the same. If a player uses mods, they openly and willingly ignore artistic intent to play a game the way they want to.
But that’s not what DLSS 5 is. It operates from the “opposite side”, so to speak. The developer implements it, and allegedly has control. The granularity of that control seems lacking, though, based on elaborative responses from nVidia engineers.
To use the example from the video link above, if developers wanted to reduce the makeup look on Grace in DLSS5 so it doesn’t look like she’s enroute to a date, there’s no way to prompt for that reduction. They would effectively have to turn DLSS5 off entirely.
It’s possible nVidia improves the control before release, but as it stands, it doesn’t sound like developers will be able to configure the tool the way they would like to.
Re:
“But that’s not what DLSS 5 is. It operates from the “opposite side”, so to speak. The developer implements it, and allegedly has control. The granularity of that control seems lacking, though, based on elaborative responses from nVidia engineers.”
That’s a completely fair point, but the analogy doesn’t have to be perfect to make a point. DLSS is an optional technology FOR DEVELOPERS. NVIDIA made that even more clear in the days since the demo video release.
So which respects artistic intent more: an optional tool developers can use and tailor to impact their own art, or my use of a mod that the artist had no idea would even exist, nor had a hand in, to change their art significantly?
“To use the example from the video link above, if developers wanted to reduce the makeup look on Grace in DLSS5 so it doesn’t look like she’s enroute to a date, there’s no way to prompt for that reduction. They would effectively have to turn DLSS5 off entirely.”
Do you know that to be true? For sure? Because I don’t and I don’t think you do, either. In fact, from hearing NVIDIA talk about the tech, the idea eventually seems to be allowing them to do exactly that kind of prompting.
So, if that’s how this ends up, you’re okay with it?
“And if the tech were able to lower the barriers of entry to the gaming industry, acting as the fertilizer that allows a thousand indie studios to sprout roots, would that really be so bad for the gaming ecosystem?”
LOL
the terms “shovelware” and “asset-flip” predate “AI slop” for a reason
wtf do you think is really going to happen if the barrier to entry is zero
Re:
Sure you get more crap, but Sturgeon’s law implies increasing the size of the pile means that there will be more 1% gems buried in there. Even among asset flip shovelware. Flappy bird wasn’t exactly high effort but it became a legitimate high impact phenomenon.
Think about things in reverse: if there was some sort of global licensing law with a fee of say 10K$ to be able to even attempt to make a game do you think we would see better games? You’d just shut out the indies and close the pipeline to amateur experimenters.
Dlss 5 seems like a more expensive, worse, product.
That’s all AI is, and all it ever will be. More expensive and worse.
unless I’m completely mistaken, my view of this is that the demo isn’t necessarily for consumers but rather for developers and publishers.
It tells me that as a developer you can move that slider to 11 and forgo the artistry to output a product by a deadline. If the tech can upscale the graphics to a great extent then I see it as a way to take shortcuts and erode power from labor.
One could envision a barebones, rudimentary, rough draft graphic design to give the AI model enough context to yassify the scenes where the output is “good enough” (for executives) and get the game published.
All of this AI hype is driven by capitalists’ desire to deskill labor, reduce its negotiating power, and eliminate jobs, even if those jobs just get offshored, outsourced, or transformed to helping the AI do the task.
Writings so easy and really who cares if a human does it.
Why don’t you replace yourself with AI if you honestly think this.
I’m tired of your ignorant takes on industries and groups you know nothing about.
If you constantly need to tell all your users they are wrong, your head is in the sand.
The only games this makes any sense for is the most generic looking games out there. Cod, bf, skyrim, etc.
Even then most have agreed it looks terrible.
Why is it for all these AI bootlicking articles you never want to actually be a journalist?
Naw, bud, it really sucks
Let’s get some definitions out of the way, because one of the most frustrating things in the entire world is how the “AI” industry and “AI” boosters like to conflate completely different technologies so that the good stuff gets confused with the bad stuff, and that makes it impossible to say the bad stuff is bad without someone going, “But what about this good stuff!” even though the good stuff is a completely different thing.
First of all, AI does not exist as a real technology. It’s a marketing term. AI exists in the same way that Old Spice “scent elves” exist, and frankly anyone who uses the term “AI” in a serious technological conversation where multiple different technologies are in play has just revealed that they probably shouldn’t be in that conversation.
What we’re generally talking about when it comes to DLSS is:
The purpose of Frame Interpolation is to increase frame rate by generating as-correct-as-possible frames in between the simulated frames. The quality of Frame Interpolation can be determined by running a game at a set frame rate with Frame Interpolation off, and the same frame rate with Frame Interpolation increasing the frame rate from a lower rate to the target rate. Then, compare generated frames to the same simulated frames. The closer the frames, the better Frame Interpolation is performing.
The purpose of Image Upscaling is to increase video resolution by generating as-correct-as-possible frames at a higher resolution than the simulated frames. The quality of Image Upscaling can be determined by running a game at a set resolution with Image Upscaling off, and the same resolution with Image Upscaling lifting the resolution from a lower resolution to the target resolution. The closer the resulting frames, the better Image Upscaling is performing.
The purpose of Image Modification is to alter the actual scene. It changes lighting, hair, clothes, skin, and the entire environmental appearance to be “better”. In other words, it erases the artist’s intent and replaces it with a generic one determined by global statistics. No, it doesn’t matter if the artist “chose” to cede that intent to the machine. Once Image Modification is turned on, artistic intent goes out the window and the machine is slopifying the videogame.
Unlike the previous two technologies, the quality of Image Modification can’t be quantified. The author apparently likes the “yassification” effect,1 which I guess is a choice, but, for the most part, people are roasting it. The general consensus is that it looks so bad. Nvidia’s custom demo reel is guaranteed to be the best they can do, and their best looks terrible.
When it comes to modding, modding video games replaces the original videogame designer’s artistic intent with the modder’s artistic intent. That’s just choosing your artist, or picking a number of artists to collect a collage of artistic intent. This is foundationally different than ceding artistic intent to a machine, because the fundamental problem with a machine is that it can’t make art, or have intent, or have artistic intent. I’d never tell a gamer they’re not allowed to feed their soul with the slop of the machine, but I can and will make inferences about what this means about the state of their soul.2
Even apart from the foundational problem of ceding “artistic intent” to a machine, the pitch from Nvidia that it doesn’t override the artist’s intent because game devs have controls is a lie from the outset. The game devs in the DLSS 5 showcase were not even aware their games were going to be featured. These decisions will not be made at the artist level, they will be made by executives. Like with similar generative technologies (LLMs, Stable Diffusion, Sora), half the point is to cut artists out of the equation.
Furthermore, the inherent nature of generative technologies (like Image Modification) and interpolation technologies (like Frame Interpolation and Image Upscaling) is substantial. Figuring out what the pixel should look like from one frame to another, or figuring out what a pixel should like like as a cluster at a higher resolution is a simpler statistical analysis–by several orders of magnitude–compared to generating an entirely new scene based on subjective opinions for what would “look better”. The training of models for Frame Interpolation and Image Upscaling are not burning enough energy to light the world on fire in the same way that training models for technologies like Image Modification are.
The author says, “The caveat to all of this is that the demo revealed very little in the way of this technology working within these games in motion.” I would encourage the author to take a careful look at the DLSS 5 EA Sports FC 26 showcase. Make sure to watch the video with the ball going in the goal; it looks like they tried to remove that from official posts, but the internet remembers.
But let’s take a step back for a moment… do people not remember what industry we’re talking about? This is video games, where giant companies spent a decade pushing brown and gray shooters as “realism”; where they endlessly pushed failing MMOs and live services and microtransactions as “the things gamers want”, cutting single player games and developers even though the rare single player game or stylized game would regularly out sell “realistic” multiplayer games. (Yes, I’m still salty about Bioware having to make Anthem.) This is not about a simple optional tool coming out. This is the richest company on the planet trying to dictate the next decade of videogames. And it looks like shit.
Furthermore, Nvidia isn’t even a videogame company. No, no, it doesn’t matter that they have a monopoly on gaming graphics cards, actually. That’s such a small part of their business now, it doesn’t matter. Nvidia is an AI company,3 and what they’re trying to do is shove slop into gaming to prop up the AI bubble for a little longer. That’s it. They need an excuse to keep burning up GPUs in data centers, ideally with a way to say it was actually profitable this time, and we are it.
No company is your friend, and they’ve all shown time and again that when they “do something for you”, what they’re actually doing is trying to shift the market to their own interests. “You’ll own nothing and you’ll like it.” Etc.4 Nvidia isn’t your friend, and they’re not even pretending to be. DLSS 5 is not for you. The point of DLSS 5 is to burn up a bunch of their GPUs training a model that gamers broadly don’t want and make you pay for it whether or not you ever turn it on. Don’t let them. Keep roasting the slop.
“I realize that some people who ‘play to have fun’ and who currently form the majority of players have voiced their reservations toward [token economies], and understandably so. However, I believe that there will be a certain number of people whose motivation is to ‘play to contribute,’ by which I mean to help make the game more exciting. Traditional gaming has offered no explicit incentive to this latter group of people, who were motivated strictly by such inconsistent personal feelings as goodwill and volunteer spirit… [User-generated content] has been brought into being solely because of individuals’ desire for self-expression and not because any explicit incentive existed to reward them for their creative efforts…
“However, with advances in token economies, users will be provided with explicit incentives, thereby resulting not only in greater consistency in their motivation, but also creating a tangible upside to their creative efforts. I believe that this will lead to more people devoting themselves to such efforts and to greater possibilities of games growing in exciting ways.” ↩
Re:
Hey, at least Sora imploded.
Okay, so what potential upsides are there, exactly? I didn’t really get any out of the article yet so–
What the– dude.
Did I miss it? Please by all means correct me if I’m wrong, but the only pro argument I actually saw was:
…Which wasn’t even about the tech in question, but a retreat into dreamy what-ifs about AI generally.
Yeah, if you need to ‘fix’ your graphics, or anything, hey, “there’s a mod for that.”
That’s the thing.
What is this tech, if not another financial circle-jerking session for the usual corporate interests.
Seriously, this tech being AI is actually besides the point, here: in what way is this filter not suits pushing a solution to a non-problem… or, a pricey band-aid to cover up problems they create by not giving dev teams enough time and resources to properly polish their products in the first place.
A filter can’t fix the triple-A industry’s bullshit, and will probably only make it worse.
This is some rank apologism
I’m not the one just wrote 2000 words about it, hoss.
Re:
That got a genuine laugh out of me.
can we get an LLM to come up with an argument for this that’s better than “you dummies actually aren’t angry about this non-problem”? because that one sure seems like a shortcut to losing.
I feel like the only person on earth who thought DLSS always looked like garbage… so this is nice to see
Every version of DLSS looks worse than the last and has pushed me away from gaming almost entirely at this point. I will stick with indies where the artistic vision is actually prioritized
I agree, and I’m glad to see someone say this. Good article.
Also, this is a pre-release tech demo. There will no doubt be refinements and optimisation before DLSS 5 actually releases to the public.
Re:
Maybe. But remember that Nvidia thought that this version, running on dual 5090s, was far enough along to be worth showing off.
Re: Re:
This happens all the time (especially with games). Sometimes it turns out that something isn’t ready to publicly show off (often with hilarious results).
And we’re just going into subjective opinion territory anyway (much like the “slop” discourse).
Re:
And the response is so overwhelmingly negative from such a wide swath of people that NVIDIA is doing damage control they probably didn’t think for a moment they’d ever need to do. That isn’t a good thing.
Re: Re:
Uh no. It’s actually a good thing that Nvidia are paying attention to the negative feedback and trying to address concerns. Whether it results in any substantial changes we won’t know until DLSS 5 actually releases (if it releases at all).
Re: Re: Re:
NVIDIA shouldn’t receive praise for looking past its own nose when it wouldn’t have done so but for the backlash. The company clearly believed, as per the CEO’s initial “all the critics are wrong” statements, this feature and its output would be better received than it was.
Re: Re: Re:2
How they attempted to address it initially was definitely questioinable at best. You won’t get any argument from me there.
Having said that, if they didn’t bother addressing it at all, there would be complaints there too (perhaps even moreso than how they initially tried to address the criticism in the first place).
Nvidia was on a hiding to nothing. If they’re guilty of anything, it’s bad timing.
Re:
And if they didn’t truly believe what they were going to show in it was going to impress people, they wouldn’t have shown it.
They wanted the public to believe this was a good thing. Their scrambling to walk back everything after the negative response proves they weren’t expecting to have to make major changes.
Re: Re:
The widely negative reaction to the announcement isn’t a phenomenon unique to Nvidia (it’s surprisingly common across a number of areas). The fact Nvidia are actually paying attention to it and trying to address it (regardless of what you think of their attempts to address it) is a good thing.
If this isn’t what it’s supposed to look like, then why did they feature it in a major press release?
Nvidia is not some poor little indie company that’s struggling to garner attention. They have plenty of agency over they they show the public. What they showed was garbage, and people are calling it garbage.
No amount of your crying, “No, you’re all wrong! It’s totally not what your lying eyes told you it was!” is going to change that.
Surprised you are discussing this here, but I agree.
The reaction seems more about people’s rage against AI in general than any kind of reasonable examination. I appreciated the reasoned take here.
The AI mania driving tech in general, is kind of disgusting right now.
But some local game graphics effects feature is NOT really part of the problem. It’s a triviality not worth raging over.
The Problem with AI is in the large centralized models, running on corporate servers, hoovering up all the data on the net and spewing hallucinations.
It’s also driving HW prices through the roof for PC enthusiast.
Ultimately this is closer to something like DLSS frame generation. People that like it will turn it on, people that don’t won’t. That’s it.
Not worth the crazy rage we see against it.
That rage should be aimed ChatGPT, Claude, Grok.
Big Centralized AI is the real problem, not a local graphics effect.
Re:
That’s the thing, though – all of this is centralized AI, because you can’t train the models without these massive data centers. DLSS 5 is centralized AI, that then creates the localized graphics effect. It’s all interconnected.
My own primary objection to “AI” is environmental: the technology has such incredibly high power and water requirements that those alone are likely to result in the deaths of people by sheer scarcity of resources and the acceleration climate disasters.
In a situation of sheer abundance, I could potentially support it. In the current environment, I cannot support it in any fashion.
Re: Re:
You’re not wrong–at least when it comes to the giant corporations that want us to believe that LLMs are things that only THEY can provide and it takes massive scale to do anything at all.
I’ve been running local LLMs on my M1 Mac Mini and it does a fine job. There are models that run on phones (though that’s hardly a surprise; my phone is technically more powerful than my Mac). The power draw doesn’t have to be massive, not even for the training. The power could come from solar, if we’d just effing install it. But even as it stands, people can train models at home and it’s not that costly. The OSS LLM community is pretty impressive.
So yeah, once again, the problem is billionaires, trying to keep good things away from us and convince us that there’s no way to have free, useful computing. (I think DLSS 5 is a monstrosity, BTW. I’m talking about using local models to write small bits of code, or clean up audio transcripts and what-have-you.)