It’s also worth noting that game devs will be able to mask different objects in a game scene, changing the structural and tone intensities for each independently. Devs might, for instance, want to give some background objects the full DLSS 5 makeover but keep a lighter touch with the characters.
This is about the only vaguely useful part of it. Just disable it for character faces. That said, why even use this garbage in the first place I don’t know.
Is there anyone that actually wants this garbage?
I know some people that do. They are also some of the biggest idiots I’ve ever met.
“New and expensive” is enough for most gamers. The community is spearheaded by consumerists.
Godot has more sliders to make a light switch.
Maybe there is eventually a way of making these filters look good by adding (many) additional parameters, but they’re still ignoring the much more important question: how are they going to get this to run on anything other than the compute server they’re currently using.
how are they going to get this to run on anything other than the compute server they’re currently using.
You’ll stream your game from an AI datacenter, of course. For only $49.99 per month. And now you can play all the games on trash hardware! While the corp analyzes and records your every move! But only if you have a good internet connection, of course!
By forcing people to pay to stream games from it?
But like, they can already do that with exclusivity deals if its just for games, nonetheless bad looking ones. The only way this would benefit them is if they genuinely think this actually looks good, or they’re trying to further inflate the AI bubble, and one of those two options seems far more likely than the other.
I think that’s the idea. “Experience the biggest game of the year, only on [insert exclusive rent-to-play cloud based platform]”.
Capitalism: You will own nothing and be happy about it, or else…
But like, they can already do that with exclusivity deals. This tech just adds extra overhead to that.
I’d assume it’s for 5000 series GPUs. And maybe someone will hack it onto 4000 cards?
I mean, given that they previously said it was only running on multi-gpu systems, I can’t see one generation of advancement being enough to make this feasible on consumer hardware. At this point, I’m thinking that this is just a paper product meant to help inflate valuations without any intention to launch it.
But they also previously said that when it releases it will be running on the same GPU your games are.
I’d bet its an unoptimized prototype.
This is quite doable on 5000 cards, theoretically. I was gonna say they’ll make it exclusive to 6000 cards and market it for them, but it’s not even clear a 6000 series is coming at this point…
I seriously doubt the river actually going to release this. No one wants it, and it also only runs on high-end cards, which also don’t need it.
Speech to text took “doubt they’re ever” and gave you "doubt the river*?
If there isn’t a Nic Cage mode, why even bother?
This is embarrassingly poor journalism. It’s clear the author has a slider with Facts on the left, and Pandering on the right, and they’ve cranked it all the way to pandering.
They define “tone” as
lighting and, well, overall tone.
Clearly not familiar with the concept of high/low frequency details.
And I blame gamers, including the people in this thread. It reminds me of how conservatives label objective journalism as biased, so right wing news outlets just describe whatever reality the audience already lives in.
They’re strength sliders. It’s not like the entire frame is defined by DLSS. The artists make thousands of decisions when rendering the frame, and use DLSS’ sliders to determine how much interference is happening on high/low detail features. In most titles, this tuning is going to be dynamically adjusted at runtime based on what lighting phenomena are in the scene, so it’s not even like an artist is going to be hand tuning these. That’s not what they’re for.
The more I think about the comparison to right wing news the more apt it feels. The topic is very technical, and their audience is…well, a bunch of gamers. It’s not like they’re going to be able to convey useful info, but they still need to say something, because their audience has heard the term “DLSS 5” and expects something. So all they do is regurgitate useless pandering.
“Guys this AI stuff isn’t actually bad. And I’m just like you, look at me try to criticize right wing news. We’re so much alike”
More like, as a rendering engineer, we’ve been dealing with vendor black boxes forever. It always sucks having control over our frame taken away, and then having to advocate for hardware level access to be able to do it ourselves so that actual innovation can happen. But DLSS isn’t new, and it isn’t defining our frame. Only end users who literally know nothing about what they’re talking about think this.
I don’t even know why I try. Sure, AI bad, art dead, all I play is AAA.
I know we don’t like AI, but, I think this is a fine use of it.
It does justice to the source material, and just just saves the time and energy that go into fine tuning it enough to satisfy the viewer.
Not a peep from them about lag or input delay. That technology, the more heavily it is used, the worse the game feels. Imagine playing a game that requires precise analog input, FPS in the hundreds but with a 150+ ms latency. It ruins gameplay. Paradoxically, those games where input lag is not an issue, like grand strategy or visual novels, don’t require such high frames. And those games that benefit from more frames, are actually hurt by this tech by becoming sluggish and disgusting to play. Such an unnecessary invention. Optimizing the game is hard work, but it makes for a better experience that’s far more impactful than AI generating will ever be.
I think it looks like shit
It’s dog shit. We’ve gone from running games, to approximating what the game would look like if it was running at a higher resolution, to inserting frames that could have been rendered, to this. Fuck all of it. It’s all a blurry, sloppy, fucking mess.
Nvidia called non-dlss and upscaled rendering “brute force rendering” in the announcement of DLSS 5. That’s just called rendering…
More like DLSS 5 is “vibe-rendering”.
And frame gen has a huge perf hit as well. It’s like a 30% perf hour to double frames. Meaning actual frames go up fast less than double and now they look worse and latency is far worse too
Meanwhile mobile games keep getting better and better looking, all whilst using MSAA since it’s super cheap.
PC/Console devs need to completely throw away the idea of deferred rendering.
It does justice to the source material
Funny joke.
That’s nice. You’re wrong, but that’s nice.
I found the bot guys!
Oh yeah, thankfully now we can automate that phase of game development where all character faces are made “beautiful” and totally unrecognizable from the initial concept and where graphic artists spawn new hair and nostrils out of nowhere.
Yeah. That’s what I expected.
