Try 4 years with VRR. For some reason it feels like Sony keeps shit close to the vest for the dumbest reasons.
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
The Verge: AMD's DLSS competitor, FSR 2.0, is coming to Xbox
- Thread starter digitalrelic
- Start date
What? Fsr 1.0 is just a lanzcoz filter. Who was saying that it couldn't be implemented. The question is why you would have implemented it when they were already using checkerboarding, which is much better.
FSR 2.0 is a much more viable replacement, with actual reconstruction
HDMI 2.1 and relying on 3rd party Soc like media tek chip to be upgraded enough to handle 4k 120hz HDR VRR, FSR 2.0 is entirely software base. Even PS4 pro could have handle it.
Why'd you quote me? And what you trying to convey? I'm very tech savvy but what your saying kinda makes no sense to what I said? I know what FSR is and in fact I was talking about VRR support and my comment as far as Sony being secretive is about many things.
But I agree and honestly it's bullshit that PS4/Pro never got VRR especially since even the Xbox One S has it.
Unless you're referring to FSR and PS4/pro then yeah they should too considering it is supposed to support even some pretty old GPUs too. Doubt Sony will even bother though.
Last edited:
Because it's possible FSR is better than TSR, not to mention FSR is a lot more marketable than TSR.
marketing and branding and potentially also works better. AMD did help Epic on implementing TSR, not uncommon for this sort of things to happen.
My understanding was that UE5's TSR was basically using FSR 1.0 on top of their original Temporal Resolution tech. So your question is valid.
Unless a dev is having trouble implementing TSR into their UE5 game, there's no reason at all to use FSR 1.0. But if FSR 2.0 by chance has better results than TSR, it could be maybe useful for a dev, but I don't really see how that would be cost effective or smart to do.
No, it does not mean anything close to being able to just assume that.
Thinking further on it, the absolute biggest difference this will probably have compared to DLSS, is temporal stability and handling moving small details. DLSS is great at dealing with aliasing and making things temporally stable, but suffers when there's wire-thin objects moving about on screen. I wanna see how FSR 2 handles both aspects.
That's definitely the make or break. Given how they're also using motion vectors, they're doing to have the same challenges, so we'll just have to wait and see how well it handles it
DLSS is actually better at wire thin objects, they appear more complete than native res at least on newer games. Transparencies and foliage are the biggest challenge for image reconstruction, DLSS is the only one that deal with those better than native res. Checkerboard was great for its time but it has become clear that it not quite up to task compare to newer techniques now even TAAU or temporal injection from insomniac. FSR 2.0 looks like it a good replacement.
It's good in stills, but in movement it starts to have problems where the underlying resolution shows itself.
Though playing with newer DLSS versions with DLSS swapper, the issue seems to have been made better recently.
It's good in stills, but in movement it starts to have problems where the underlying resolution shows itself.
Though playing with newer DLSS versions with DLSS swapper, the issue seems to have been made better recently.
Yeah, i remember the resolves of thin wire being a lot better in the newer version.
I hope Sony first party will jump on this asap. Particularly Horizon Forbidden West current shimmering issue in performance mode. This could be a game changing near term solution seeing how AMD it doesnt take long to implement. Second is ND, I hope they finally start looking into image reconstruction techniques, they were the last major sony first party that opted out of all types of image reconstruction. Hope they start by implementing it to the Uncharted collection with their R&D. Then we can have current 1440p 60fps mode > 4k 60fps FSR 2.0 quality mode scaling and 1080p 120fps mode > 4K 120fps FSR 2.0 performance mode. One can dream....
Last edited:
I think there may be some confusion regarding this as it appears that Forspoken is using FSR 1.0 according to the video released by AMD.
Thats with 6800xt and you can see the cost increases when going to 6700xt...From The Verge article, this is what AMD claims:
Quality Mode:
Performance mode:
We dont know the costs for upsampling on the consoles
2.0 is in development for Forspoken, that much was confirmed in a recent vid showcasing the game
Cool! Hope this gets well supported by devs. I wonder if knowing significantly improved reconstruction techniques were on the horizon made the Series S an even easier sell internally at Xbox. They already have their mid gen power refresh - in Series X. It just happened to launch alongside lol. Not really, obv, but that does kinda feel like how it's playing out.
I imagine that Sony's implementation of AMD features was less standard than Xbox's. It's why Xbox crowed about full RDNA2 when Sony could not. Doesn't necessarily mean the performance will be worse. But perhaps it means that the solutions will not be perfectly portable from one system to the other. And the PS5 version will take an extra day or two. Maybe. Who knows. But the more places that FSR2.0 can be used and makes a diff, the more quickly it will be adopted. So, I certainly hope we see it everywhere. It would be wild if the DLSS use of ML, which is a super cool innovation particularly for the high end, ended up limiting its wider adoption longer term.
I imagine that Sony's implementation of AMD features was less standard than Xbox's. It's why Xbox crowed about full RDNA2 when Sony could not. Doesn't necessarily mean the performance will be worse. But perhaps it means that the solutions will not be perfectly portable from one system to the other. And the PS5 version will take an extra day or two. Maybe. Who knows. But the more places that FSR2.0 can be used and makes a diff, the more quickly it will be adopted. So, I certainly hope we see it everywhere. It would be wild if the DLSS use of ML, which is a super cool innovation particularly for the high end, ended up limiting its wider adoption longer term.
OP
OP
Yes, I shared the photo. I can see that.
Series X GPU is just a bit less performant than a 6700xt, so we should have some idea of the cost here.
It could be, but this announced is that Microsoft have built it into the development kit
Horizon's awful IQ is not the fault of the console.
It looks worse than returnal despite rendering much higher lol.
I dont think fsr would help until they change the engine directly.
The easy solution is doing the Death Stranding widescreen 60 fps mode of the DC version. Perfect ( and with better Fov too ).
It's a little bit less aliased in quality mode - that's good. It doesn't feel like artificial sharpening this time.
Hopefully the Xbox version of DeathLoop will have it
I think it will considering Arkane is clearly working with the tech, possibly due to the fact that the Xbox port is being worked on at the moment.
OP
OP
This doesn't make anything clearer for me. Am I missing something?
That they used FSR 1.0 for the GDC demo but 2.0 is being integrated into Luminous Engine. FSR 2.0 is going to be in the game at launch according to them and all FidelityFx features in the game are being deployed on both PC and PS5.
you dont need to change the entire engine. Replace reconstruction tech that has less draw back and better anti aliasing will get them there albeit softer image quality but still tons better than having the same shimmer and flicker. Ultrawide is is not a simple solution to Horizon 2, DS is a last gen game eith barely any visual upgrade running on ps5 so there are tons of performance headroom. Horizon 2 is massively upgraded running running on ps5 not even a close comparison in anyway same engine or not. Reduction from 3840x2160 to 3840x1600 wont give you a 2x boost in performance in horizon 2 ps5.
This slide is interesting, i see AMD VRS getting mentioned and implemented on ps5. Guess VRS tier 1 software solution got is also part of AMD VRS. Whatever working behind the scene I hope AMD plug in makes it easy to implement and more dev will adopt it. Missing in a few games in ps5 version ie Doom Eternal.
I'm not versed in these technologies, so I'm not sure if this is any different or better than the reconstruction techniques that we've been seeing on consoles since they released the PS4 pro
DLSS was likely made as a counter to console upscaling and consoles still being marketed as 4K machines. It's an interesting turn around.
It all depends on how good FSR 2.0 is at filling in the details that are not there. There seems to be some confusion about DLSS magic hype and continuing that with FSR. DLSS is magic for two reasons, 1 being PC has been lacking in upscaling options and 2 being DLSS can actually produce good IQ even if the base resolution is 540p.
Any upscale solution is good at around 1440p/1520p mark because there is around 4 million pixels to work with, which is what generally console games aim for if they are achieving an upscaled 4K output. If FSR 2.0 can't pull off upscale magic at very low resolutions, there really won't be much more juice to be squeezed out of PS5/SX. PS5 is more tricky because FSR 2.0 is optimised for RDNA2 and AMD did just confirm this, these algorithms are not free and PS5 has hardware dedicated to checkerboard rendering already. At worst FSR 2.0 would offer a small quality bump over say checkerboard rendering with extra cost, at best it's like DLSS and can pull off a good fake 1080p output for a 120fps performance mode like you mentioned, as long as the CPU isn't bottlenecked.
OP
OP
It's way, way better if what AMD is claiming pans out. FSR 2.0 quality mode would be virtually indistinguishable from Native 4K, for example. Similar to what nVidia has done with DLSS 2.0-2.3.
But we'll have to see how things pan out.
Well that would be incredible and just what these consoles need. Always thought 4k native was a waste of resources anyway
OP
OP
Yes, completely agree. That's why I said it's a total game changer for consoles if it works as claimed by AMD.
Very few games are 4K native. The ones that are specifically chose to be 4K native, and I doubt that it was because they couldn't implement a temporal upscaling solution (or they couldn't find one adequate enough).
I think FSR 2.0 is great news, but more so for PC gamers who don't have cards that support DLSS. A developer who doesn't have an engine that supports TAA will still have to go through weeks of work to utilize FSR 2.0.
What I think would be a game changer is if FSR2.0 could/will be extended to encorporate ML in its pipeline in an industry open standard, while being backwards compatible (ie non-ML fallback).
Forcing an industry standard approach to not only upscaling/reconstruction, but it's future evolution would be awesome.
Edit: I guess I haven't exactly seen how this is a game changer for consoles. If Epic decides to wholesale replace TSR with FSR 2.0 in UE5, and commits to its further open source development, that'll be a game changer as well.
Yeah, FSR 2.0 is definitely a bigger deal for PC. Outside of DLSS, image reconstruction isn't really a thing on PC like it is on consoles. That and the fact that "AMD sponsored" will no longer equate to "AMD paid developers to keep objectively better tech out of their games so they wouldn't look as bad". Well... it'll probably still equate to that, but at least FSR will be a viable option for people who care about image quality.
That's really more of the situation but this topic assumes a game will have the tech involved.
We still need high native pixel density for reflections, particles and volumetric effects. 4k is a overrated standard that doesn't go far enough for me when I can push especially in offline rendering. In FH5 I still see pixelated smoke on smoke effects at 4k, completely smooth gradients by 5k.
Games and engines make it case by case too and where I can get to it or more why not when I can reflections, volumetric effects and particles all can be improved by resolution increase native or not, dlss makes them pop IIRC. If a dev can leverage it why not unless you employ a solution like DLSS and DRS at once there nothing better for quality than native with good post processing. GPUs are starting to chew up 4k in decent engines with some taking us to 6 to 9k.
I'd say it's more wasteful for things in the pipeline just to automatically want this type of res and what the computational needs are when we considering games in the future. Balance out performance hogs or watch your comeptitors dance on you with their sense of balance.
Since I've seen this paraded around a few times would people like yourself like to remind people like me how many developers are doing TAA vs overwhelming amount in existence who aren't? Cause that number is already very small if we are talking about their implementation vs what this is already showing. That's a lot of devs to leave out for at best maybe 2 to 3 dozen since I'm willing to play nice and concede some.
Native rendering is dead to me and not employing the tech in any fashion to me means you didn't want to swing for the fence either in performance or looks. It's great for any dev considering the end result is a better looking for performing game than what native would've afforded them especially if the engine is nothing like what is out there that scales well or looks good. This tech effectively gives back 1/4 to 1/2 of resolution performance cost.
Did you miss the Nvidia Streamline (IIRC) announcement I think that is as close as we get to that.
R&D is showing us someone would need an ultimate solution that takes what makes FSR good at times, TAA, DLSS, DLDRS, and FSR 2.0 and have it all in one tech. We are probably a couple years from anything that does that. Consoles have a lot to improve considering FSR 1.0 and DLDRS are showing you can really work the pipeline a certain way most devs until now have never considered or thought possible.
Last edited:
Wow, the first part of your post comes off super confrontational. Anyway,
1) I said that very few developers are actually releasing games rendered at native 4K, not how ubiquitous TAA is. I'm not even going to argue about quality between solutions when FSR 2.0 isn't even released. I do know that you can't just wholesale act like people aren't using TAA if you "don't like the image quality their games have"
2) Native rendering being dead to you is just your opinion. That has no bearing on what I said.
3) FSR 2.0 doesn't magically enable TAA in engines that don't support it. It'll take weeks of dev effort, as AMD themselves stated, if the engine isn't already utilizing motion vectors and have proper buffers set up. If a dev already has those things setup, they are likely already using TAA.
I'm personally hoping that AMD's solution will 100% provide consistently better results than bespoke solutions. I'm not against it being widely adopted. I'm just not clear on why it's inherently a game changer for consoles? The main reasoning I hear is that, "It'll free up so many resources for devs". I'm like, most devs are already rendering games at subnative resolution, when it comes to 4K. If a game is already rendering internally at 1440p/1080p/900p etc., FSR isn't going to gain the dev any extra performance. Games that are rendering at higher resolutions internally now likely do so due to them being cross platform in nature, with the majority of development happening even before the devs had current gen console dev kits.
I like when people spit real sorry for the rub it's all good though.
1. All good
2. It's dead to the industry that keeps using all this tech too.....
3. Doesn't magically enable in big engines that can't use TAA cause of design either. I think most of this tech discussion is relegated to when applicable due to sheer breath of 3d games alone that don't have it and never will. Future is forward looking and I think the amount of titles big and small using something is very nice to see. I said it in other topics forza horizon 5 or other recent titles are proof its not silver bullet but for future games it is especially if it's involved with one of the biggest engines in gaming and all 3 makers at sdk level. It's better than AA since it incorporates it and 3d literally took most of the 90s to achieve mass adoption and standarization despite being around since the early part. weeks and months isn't bad for the talents considering what kind of games cube and PS4 produced with little effort in weeks or months. This should be noted cause it's not years or unviable. Don't automatically assume I feel all games need it some games have little to no 3d rendering thus no need.
I'd said my main reasoning for any devs to be happy this exists. All devs are free in non native situations to leverage the performance given back by native situations lets not just say 4k plus but even 1440P plus.in any way they like. The more effective the engine at pumping out fps the more it can be converted in to more IQ of the tech running, more fx both variety and quality or other aspects of rendering. Harmony for all.
Epic and AMD working together is a good thing as both are borrowing from tech involved to make either better. I think the efforts seen on this tech are better than any api initiative some of these companies are involved with.
OP
OP
To your first point: it doesn't matter whether the resolution FSR 2.0 is replacing is Native 4K or not. I used Native 4K as an example but FSR will be a benefit at 1800p or 1440p or whatever resolution; it doesn't matter. Being able to internally render a game at 1080p w/FSR up to 1440p, for example, would be a massive benefit over rendering 1440p natively or using a lesser, blurry, upscaling alternative that consoles are stuck with now.
The point is no matter the resolution, if FSR 2.0 works as advertised it's going to be a huge benefit to that performance/image quality balance that we're always trying to find for these current gen consoles.
And again, any of our arguments on one side or the other all depends on AMD actually coming through with their claims, of course.
Last edited: