transcript

Speaker 1:
[00:00] Amazon presents Jeff vs. Taco Truck Salsa. Whether it's Verde, Roja or the orange one. For Jeff, trying any salsa is like playing Russian roulette with a flamethrower. Luckily, Jeff saved with Amazon and stocked up on antacids, ginger tea and milk. Habanero? More like habaner, yes. Save the everyday with Amazon.

Speaker 2:
[00:32] Well, hello there and welcome back to this, the DF Direct Q&A show. Some great questions to get through this week, but discussing it all with me, first of all, Oliver Mackenzie, hello.

Speaker 3:
[00:42] Yes, Rich, always a pleasure to be with you here this Monday morning to record this Q&A show for you guys.

Speaker 2:
[00:51] And Alex Batalia, hello. You're sitting down.

Speaker 4:
[00:55] Yes, I'm equipped with a new chair, a sciatic nerve pillow, and I guess an attitude. I'm ready.

Speaker 2:
[01:04] What, like knuckles?

Speaker 4:
[01:06] Yes. No, I'm more of a shadow kind of guy.

Speaker 2:
[01:10] Okay. Fair enough. Well, it's great to have you here. And although the treadmill, it kind of worked for you.

Speaker 4:
[01:18] I'll be switching up on and off, believe me. I can't just do one or the other, so you'll see me on both. Okay.

Speaker 2:
[01:23] Trampoline?

Speaker 4:
[01:25] One day, yeah. I don't know if the neighbors below will like me with that one, but.

Speaker 2:
[01:31] Fair enough. Okay. Well, let's crack on with our first question. But first of all, shout out to the DF Supporter Program. Basically, if you want to get involved in DF Direct, if you want early access to DF Direct, if you want to contribute to the Q&A show, please do consider supporting the team. Right. First question. This one's coming from Dajarco in Vakistan. I've noticed with a few recent and upcoming games, developers boasting about quote unquote increased internal resolutions on PS5 Pro. I immediately think to myself, surely we want the opposite. Now it's packing a decent upscaler. Why does this keep happening? Could somebody make it stop? As our Capcom and Remedy, the only ones using Pssr correctly to drop the internal resolution. Alex, interesting question here, because it is essentially all about quantity of pixels versus quality of pixels, and the new PSSR straddles the line between delivering both reduction in pixels, which means you can put more GPU power into each pixel, and the upscaler is doing a pretty good job.

Speaker 4:
[02:32] Yeah. What Dan is referencing here is the fact that with a scalar like Pssr2, or Fssr4, or DLSS4, you don't need a lot of internal resolution to get good anti-aliasing and upscale on most rasterized green elements. And as a result, you don't need like previous upscalers like Fssr, or sometimes TSR, depending upon the game. Maybe you'd be looking for like 1440p and above, or like old TAAUs, you'd want to be higher than that F4K, maybe even to get a 4K approximation that looks great. So you don't need such great internal resolution. Games that are boosting something like 1440p internal and then using PSSR on top of it, it's arguably wasting a bit of resolution there and performance that could be used for other things like increasing ray tracing samples, turning on one extra ray tracing effect, etc. So I'd agree with Dan there. To a limit, I would argue that when I play with DLSS on PC, for the most part, I actually try and impose a performance limit there. I say, I really want to just be using what is performance mode at 4K, for the most part. I could of course go under and I'd still get great edge and quality and detail in most surfaces. But as soon as you get less than one fourth, the quality of internal resolution there, since other effects are scaling with resolution as well, you start getting knock-ons that I'd say are less than positive, and they become more and more obvious over time. So if you have stuff like post-processing that doesn't scale with the output resolution, or some other internal effect, or ray tracing which doesn't, which it shouldn't probably, then you can start seeing stuff like certain edges that are affected only by ray tracing, or certain in-surface detail that is affected by ray tracing, starts looking really low res, and maybe sparkly and other things. So I think Remedy gets off with it pretty well here in their game, in the performance mode, which is targeting 60 FPS, I'd say Capcom actually has a little bit more issues at its lower internal resolution there, mainly due to what happens to the quality of ray trace reflections in a game like Progbana. So I'd say there are limits, and it's cool that it works at all, unlike it wouldn't work with that Pssr, but I'd still think you'd maybe want to balance that a little differently personally, yeah.

Speaker 2:
[05:21] Yeah, yeah, I'm inclined to agree. Oliver, you've basically tested anything and everything on the PlayStation 5 professional, Pssr related. What do you make of this?

Speaker 3:
[05:32] Well, I don't think there's one correct or incorrect way to do this, because if you've got nothing else to do with the extra GPU power in PS5 Pro, outside of implementing Pssr, maybe you tweak one or two settings, but you're going into placebo level settings adjustments when you're going into things like adjusting shadow resolution slightly or adjusting the SSR resolve slightly. I feel like just bumping up internal resolution is one obvious area you could go, perhaps the obvious area where you could go. I think this person is referencing the fact that I believe for Saros, Sony is advertising that it has an increased internal resolution on PS5 Pro. Of course, they aren't specifying what that is. We'll have to find out ourselves when we take a look at that game. But yeah, I just feel like this is just one of those areas where you can easily deploy GPU power, so it makes sense that you'd put GPU power there. But at the same time, I am in complete agreement that these more aggressive upscale factors, like 864p to 4K, which is what we're seeing in Pragmata and Alnoic 2, that can look very, very good, especially at normal viewing distance with PSSR. It's awfully good, I think, and I don't really think in a lot of titles, especially rasterized titles, primarily rasterized titles that don't have super noisy ray tracing or ray tracing that, again, is dependent on resolution to resolve and look correct. I think that it's fine to just do 864p in many cases to 4K. I mean, it's going to depend on the game. But like what we see in Allenoid 2 is remarkably clean, clear, good texture detail, great stability, no ghosting really, no, like all the visual gremlins we complain about usually aren't really there. So why not just go with the more aggressive upscale? To me, though, I think with Pssr, you don't want to just like dial in a low resolution arbitrarily, or even to high frame rates or whatever. I think you want to target like a fixed refresh, like let's say 60 FPS, and then dial in additional meaningful effects while dialing down resolution. Like adding additional ray tracing into the mix. I think that's what kind of squares the circle and makes it make sense. But if you don't have anything else to do, I don't really begrudge people for just bumping up internal res. Because what else are you going to do if you don't have like meaningful effects to stack on top of that, right?

Speaker 2:
[07:40] So my general thoughts on this fairly straightforward. I'm all in on like lower resolutions if it frees up more GPU resources for better quality visuals, and especially ray tracing. I do think that there is an issue with obviously, we don't have ray regeneration on the PlayStation 5 Pro. We don't have ray reconstruction. And I think that's sort of like a key component. That's the next thing I'd like to see added. But at the same time, I'm a bit concerned that ray regeneration on the AMD side doesn't seem to be like a holistic solution that includes upscaling. Meanwhile, on the PC side of things, we don't seem to have had much movement recently with ray reconstruction in terms of new models for that. And we've also got the bizarre situation of games only allowing ray reconstruction on path tracing games. That's not what it's supposed to be for exclusively. I think you'll agree there, Alex. I mean, if a game supports ray reconstruction, why would you restrict it just to path tracing and it can do an excellent job on ray tracing as well? It doesn't make sense to me.

Speaker 4:
[08:49] Yeah, for that, I agree with you. And I think one of the few reasons why it is the case is just the extra QA time and the, you know, like, you need to have like a certain setup for it to work well. And then you'd have to essentially add an extra mode for it and all the work that would mean. So I think it is about saving time at work, which is money.

Speaker 2:
[09:10] Yeah. But, you know, I'd say get used to those 864 P's even moving into next gen.

Speaker 4:
[09:16] Oh, definitely. Especially if there is some sort of like rate regeneration on the next, you know, PlayStation console.

Speaker 2:
[09:22] It's almost certainly going to happen. Okay, let's move on to the next question.

Speaker 5:
[09:26] Eczema is unpredictable, but you can flare less with ebgliss. A once monthly treatment for moderate to severe eczema. After an initial four month or longer dosing phase, about four in 10 people taking ebgliss achieved itch relief and clear or almost clear skin at 16 weeks. And most of those people maintain skin that's still more clear at one year with monthly dosing.

Speaker 6:
[09:45] Ebgliss, Lebrichizumab LBKZ, a 250 mg per 2 ml injection is a prescription medicine used to treat adults and children 12 years of age and older who weigh at least 88 pounds or 40 kg with moderate to severe eczema. Also called atopic dermatitis that is not well controlled with prescription therapies used on the skin or topicals or who cannot use topical therapies. Ebgliss can be used with or without topical corticosteroids. Don't use if you're allergic to ebgliss. Allergic reactions can occur that can be severe. Eye problems can occur. Tell your doctor if you have new or worsening eye problems. You should not receive a live vaccine when treated with ebgliss. Before starting ebgliss, tell your doctor if you have a parasitic infection.

Speaker 5:
[10:19] Ask your doctor about ebgliss and visit ebgliss.lilly.com or call 1-800-LILLY-RX or 1-800-545-5979.

Speaker 2:
[10:26] This one comes from some guy person. After Nintendo was able to add in the quote unquote handheld boost feature, it got me thinking about other potential improvements Nintendo could make for Switch 2. Do you think Nintendo could potentially force older games into a 120Hz container and potentially cut latency by 8ms? Would there potentially be a way to add VRR to games that implement Nintendo's API for Vsync? Or are those APIs compiled into the game code as opposed to system functions provided by the Switch operating system itself? So, Oliver, I think the handheld boost mode, I think we can sort of universally agree that it was something that we actually asked for, the concept of being able to run last gen games importable in their docs configurations. I think that's a really good move, but it's also something that I kind of never thought Nintendo would do. So I guess there is sort of potential for using Switch 2 features on Switch 1 games, perhaps more aggressively. What do you think?

Speaker 3:
[11:25] Yeah, I mean, definitely the handheld boost mode thing is something that, like you mentioned, it's very uncharacteristic of Nintendo to offer those kind of specialty enhancement features. We're used to seeing that from Sony a little bit more, maybe from other vendors, not so much from Nintendo. Certainly, Nintendo's backwards compatibility has generally been hardware based in the past for them to go down this software based solution and actually offer meaningful performance upgrades. And then on top of that offer, the boost mode functionality is just a great thing. I'd love to see 120 hertz output in older Switch games. I think that would be really cool. I think it could cut latency by quite a lot, actually. I think that would be pretty neat. There are some issues potentially though. I mean, it is 1440p output and obviously a lot of Switch 1 games are upgraded to 4K. That would not be an ideal situation necessarily to just force those games into 120 hertz containers. In general, I think pairing the update of the game and the refresh rate of the panel is probably a good idea for many sets for a variety of reasons, but I think that would be cool. For VRR, I think that would be a bit more complex. Maybe Alex can speak to that. But in general, I think that going down this route of adding additional enhancements to older Switch software is pretty cool. I do wonder if 120 hertz in particular might be going a little bit into the weeds, considering that Nintendo does not even support 120 hertz very well in the console to begin with though.

Speaker 2:
[12:52] Yeah, fair enough. Alex, Switch 2, your favorite piece of gaming hardware ever. What do you make of this question?

Speaker 4:
[13:00] Well, just to clarify a handful of things, it would reduce latency by a lot more than 8 milliseconds. You have to think like with Vsync, triple-buffered Vsync, there's going to be front buffer, back buffer, back buffer, and you're talking about reducing all of those by half. You could be looking at the game's input latency being one-half of what it currently is, which would be a huge amount, more than 8 milliseconds for certain. But there's some complications to that. One, Nintendo's own games very rarely are actually triple-buffered. They are pretty much always double-buffered for some reason, probably for input latency concerns. Also, because maybe Nintendo doesn't actually have games that run on Delta time. Maybe they actually are forced-tick games where they run at 60 or they run at 30, and there's different profiles for those. But in-between framerates need to run the game at half speed in order to actually have the game run at full speed, if you know what I mean. There's no in-betweens there. I think a lot of Nintendo games are like that, actually. But for third-party games, that is definitely not the case. So, I think that there is some use in the third-party realm for those kind of things, for sure. Regarding VRR, that is more of a panel thing and a driver-side thing. The game itself is going to be putting out stuff assuming Vsync is going to be on, in which case, I think it arguably could be done on the software side for Nintendo. It's just a matter of whether they would want to do it. In that case, yeah. I mean, because when you run VRR on PC, you can also run it with Vsync. That's just one of the aspects of VRR. I think it's fully doable. But Oliver's point about the fact that the games are 4K versus 1440p, which is the only way to get VRR really on that Nintendo docked profile. And that is 120Hz, yeah, 120Hz without the handheld. In the docked profile, that complicates things a lot more. If it was just, we supported HMI 2.1, I think it would be a slightly different story, and this would be much less of a complex thing because the games are outputting at certain resolutions, and then the console has to do like two-way scaling maybe, or you know, it's like, this is getting a bit too complex for a niche thing that Nintendo would have to support.

Speaker 2:
[15:40] My only thing I'd like to add is, well, when you've got a system where you actually are in control of the panel, that opens up some intriguing possibilities because everybody will have the same panel, certainly in the current production run, we understand. There is some sort of latitude there for doing some interesting things. The thing is that they shipped with a VRR solution that doesn't support low-fiber rate compensation, which is like baffling.

Speaker 4:
[16:07] Yeah, it is.

Speaker 2:
[16:08] So it kind of suggests that it's kind of not a priority for them, but maybe it will be at some point. In the meantime, you've actually had Ubisoft step up and say, hey, you can do a software version of low-fiber rate compensation, here's how. Which is kind of like great on the one hand, it's ingenious, but you kind of want the platform holder to be like using its hardware a bit more effectively. Anyway, let's move on. Got a question here from DarkRod99. Hi, guys, seeing how good Pico performs in Death Stranding 2 and Horizon Forbidden West that even rivals DLSS solutions, is it fair to say that maybe we don't really need that much DLSS or FSR or XCSS in order to get a good upscaler? Not to dismiss all of what DLSS and others have done, but being too dependent on vendor solutions has its downsides, like the time DLSS version was broken for a while, giving some regressions in ghosting and image quality. I think devs should take some of what Gorilla is doing and maybe have a small team of engineers trying to develop in-house solutions where they can control the pipeline and the output and not wait for Vendor X to fix an issue that's causing their game to look bad. I know that in the current industry, economics is hard, but is an investment worth to do? I hope to see much more in-depth video of Pico in the future from you guys. Cheers. Alex, well, this is like Gorilla know what they're doing, and they have their own bespoke completely custom engine, and that combination has produced some really impressive stuff. That's like a unicorn style scenario.

Speaker 4:
[17:45] Yeah. One, we haven't actually done a Pico image quality review in the aspect of, we have to see how it scales at 4x res and 9x res, like the super hardcore scaling factors to actually give a sense of those things. Scaling from 1440p up to 4K is, even when we went to those FSR3 scenarios. What you're looking there at then, that case is stuff like how soft the output is, and where do image quality regressions usually in the face, in terms of ghosting, show up the most. And there you're looking at like flavors differences, a little bit more so than how quality is the upscaler. Because different games are going to have different regressions, depending upon where they actually have motion vectors. So we haven't done an image quality review of Pico, to say the least. That would be an interesting thing to do with infinite time and infinite money. But looking at...

Speaker 2:
[18:46] The infinite money sounds good.

Speaker 4:
[18:48] Yeah, why not? Infinite time... I don't know about that, actually. Would I say that? But looking at stuff like, should a developer make their own image quality upscaler? And I would say, hell no. Dude, by the time PS is like... They made Pico in the time period when PSSR1 was kind of bad. I think under different pressures and scenarios, they would have just said, actually, wait, let's use PSSR2, right? You know, maybe they would have made that decision. Because someone taking all that engineering time and infrastructure and implementation time and doing it for you and then you drag and dropping it in as much as possible, that is way easier than doing what Gorilla probably had to do to make Pico and that it was probably a big engineering investment time. And it should definitely not be expected for everyone else. But I do agree with the fact that it is... We definitely need the scalers to be less vendor specific, especially now. We have stuff like... It's now changed its name. It's no longer called Cooperative Vectors. I forget the name, what Microsoft called it. But you can basically do machine learning within DirectX at the moment on all the vendor's cards that support it. And at this point in time, I humbly submit that DLSS, XESS and FSR should just run on that instead of whatever proprietary junk they're doing right now. And that would be great because then people will just choose what they like the best. I would vastly prefer that situation. And then you wouldn't have to wait for Vendorex to fix an issue. You would just run a different upscaler like DLSS on an AMD card. I think that's the best future. And I would like to push for that now more than developers make your own super expensive upscaling solution.

Speaker 2:
[20:48] Interesting. Yeah. I mean, I'd love to see it happen. It's kind of like almost a utopian style vision for the future of upscaling, in a way, it has become sort of segmented. But if there is a solution to unify it, then maybe it is just a broad direction of travel. We'll eventually get there. Oliver, what do you make of all of this?

Speaker 3:
[21:08] Well, I would just say that Sony is in a position as a platform holder to dictate this. And they think that the universal solution is better than everyone else, everyone going and rolling their own Picos. We've seen this from studios like Insomniac, rolling their own temporal injection solution. Different kinds of players, obviously, rolling their own TAAs. But I think over time, what we've seen in game development, just in general and also within Sony, is that we are moving towards more homogenous solutions. You see titles like Gosev Yotay using Fsr3, using these commodity solutions. And I think PSSR is just a very bespoke commodity solution. So to speak, that everyone can use. And they've decided that's a good investment for their platform. So I think, you know, it's hard to argue with results, and it's hard to argue with their logic there, that devoting some time within AMD, within Sony, to creating a really high-quality app that can be used for a wide range of titles is just simply much more efficient than trying to roll a bunch of bespoke, customized to the engine solutions in a bunch of different games like Pico, even though Pico does seem to be quite good. I'd be interested to see if Pico actually holds up that well like 1080p to 4K or like 860p to 4K, though because all we've seen so far in our own testing and in our own testing of PS5 Pro titles and in our recent testing on PC has been basically 1440p to 4K, which not much of an upscale. It's not that impressive to achieve a really good looking image. I mean, it is impressive. It is impressive. It doesn't look really, really good, but I mean, don't get me wrong there. But it's not an enormous accomplishment relative to kinds of upscaling scenarios that we're kind of seeing more and more often on PS5 Pro where you really do want that kind of two by two upscale or even 2.5 by 2.5 upscaling each access. That is much more important. So yeah, I think it's hard to generalize from Pico, which does seem to be a very competent, very good looking anti-aliasing and upscaling solution created by some very talented engineers. It's hard to extrapolate that to studios that are gorilla in my opinion.

Speaker 2:
[23:14] Yeah, I mean, I tried it on 1440p DRS to 60 and it looked pretty good. I'm not going to say I've done an in-depth comparison to DLSS, but I think the other thing to bearing mind when you're using DRS is that clearly there is a significant decrease in computational cost in using Pico versus DLSS, which would feed into a higher base resolution if you're using dynamic resolution scaling. So that's kind of like swings and roundabouts or worst. But it's certainly a very, very interesting technology and to have an engine-specific GPU agnostic scalar is actually very, very good. It's just as you guys say, it's extremely difficult to do. To illustrate how extremely difficult it is to do, well, look at Epic's TSR. Epic have basically got the equivalent of orders of magnitude, more engineering resources than the typical developer, and TSR is okay, right? But it's not the finished article, so to speak. I don't think that is the future though, and I think you're right, just having a plug-in component that does a pretty good job. Hopefully, a plug-in component that has sufficient versatility for developers to tweak it and improve it so it's not quite the best fit for their art style or for their pipeline. That's sort of like the way forward. But PSSR2 on PS5 Pro just seems to, with one or two exceptions like Starfield, as a plug-in replacement, it seems to be doing a really, really good job there. Interesting nonetheless, and it's quite interesting to see so many people perk up at the findings there, but let's move on. I've got a couple of questions here. I've just lumped them together. It's basically the same thing. We'll start with this one from Todd C. Hello gents! With regards to Capcom, Switch 2 ports and the want for a frame rate cap, would one solution not be to limit the frame rate on the display itself? If possible, surely a PC monitor could be capped at 30 frames per second. And many TVs still have an HDMI port that only supports 30 FPS. Would this actually make the gameplay experience better? I hope you have a great week. Thanks. Before we begin the discussion, no, it won't. Display panels are horrific. We'll talk about that. Teasing hilarity ads, I wonder, could Capcom's reluctance to provide 30-40 FPS cap options for Pragmata and other titles have something to do with how their RE engine handles input? Could it be that providing them the frame rate caps would introduce the input latency they don't want to have? Well, that is an interesting point, because if they do cap the frame rate, they will introduce a latency hit that otherwise wouldn't have. It's just a case of how much you would actually feel it, certainly at 40 FPS. But Oliver, what do you make of this situation? Because we're advocating for caps of some description on these games, or at least optional caps.

Speaker 3:
[26:24] Yeah, I mean, jeez, I think it's mostly just a Capcom preference at this point. Because when you look back through their software, their games behave fine by and large when capped to 30 FPS or 40 FPS when we see them on PC. But if you go back through Capcom's history, going back quite a ways, they try to hit 60 and if they can't hit 60, they just run unlocked by default. There are some exceptions. Probably the biggest exception that I can think of that is actually properly frame paced when it is capped below its output refresh is Monster Hunter Wilds. That game, you can go 30 FPS, 40 FPS, 60 FPS, whatever. You can do unlocked modes, but it all runs basically as it should. And so I think that they can do it or at least have demonstrated that they can do it. And it may be just like this person is pointing out that input lag is perceived as too high. You know, and there are other solutions that they could go down. I mean, they could use VR handheld properly. I mean, it does not seem to be implemented properly at all. It does, like when it's at 55 FPS, it's not smooth in the slightest. So I think there are a lot of solutions, but ultimately it just comes down to Capcom wants to do what they want to do, and this isn't what they want to do. And it's kind of a little bit annoying because these are good ports otherwise, but you know, it's up to them. On this 30 FPS tick here, on the 30 hertz tick, I guess, I don't think that would really work because Switch 2 doesn't support 30 hertz output normally, and it wouldn't deliver the smoothness you're looking for. You basically see your 30 hertz update on your television presumably, and you'd see a very inconsistent update and inconsistent presentation between the frames that are presented. So I don't think that would work at all, but it would be interesting. I mean, maybe, you know, implementing a 30 hertz display mode, I mean, TVs certainly support it. One big problem there is you're just dealing with like 33 milliseconds of latency between frames, so anytime you're dropping, you have big problems there. And that also poses big problems for latency. So perhaps not the ideal solution.

Speaker 2:
[28:27] Yeah, when 4K screens first appeared, they were limited to 30 hertz. And just the lagginess, just moving a mouse pointer about was bad enough. But you know, gameplay was just just horrible. Similar to the when people start talking about, you know, cinematic experiences at 24 FPS. Well, I've tried it. You don't want it.

Speaker 4:
[28:45] No, it's a bad.

Speaker 3:
[28:46] You just don't want it.

Speaker 2:
[28:48] Alex, what do you make of this? Should they introduce caps?

Speaker 4:
[28:52] Well, I think there should be an option for certain. Just one option. 30 FPS cap, yes or no button. Do it.

Speaker 2:
[29:01] I mean, 40 FPS.

Speaker 4:
[29:03] If if Xbox 360 games could do it, they can do it, you know, like so, you know, I think I think it's been a long overdue thing. They've been doing it for a really long time in their console titles. And I think in the titles where it's like almost 60 pretty often, that this isn't a no one even thinks about it. But there's a lot of the things like the Switch ports or the PS4, Xbox one versions of their titles that were like between 30 and 40 FPS at best, you know, that's really where it should be. And I think the downside is of course you're going to be queuing frames a lot more and probably adding something, I would just presume something like 30 to 50 milliseconds of input latency, easily 50 milliseconds if not higher. But like the smoothness of the output is way higher on a standard panel. And in Switch 2's case without low framerate compensation and with the way VRR works, you'd probably want that. Although arguably, like Oliver said, why not just also have a VRR mode that is pretty explicit and actually rolls its own low framerate compensation? That would probably be the best combination of these things that would give even lower input latency than they currently have, because it wouldn't be queuing frames, I presume at that point. So yeah, that's probably the best way to do it.

Speaker 2:
[30:34] Okay. Yeah. Yeah. I think that VRR should probably be sorted if you're going to be having unlocked framerates. I mean, there will be situations where the framerate would go too low, really. But even so, sort it. That's at least sort the handheld experience. It is possible. Okay, let's move on. All right then, another question from teasing hilarity. Hello again, gentlemen, exclamation point. Did you hear about the rating leak and other rumors around Starfield for Switch 2, given the performance Bethesda has been able to offer for Fallout 4 and Switch 2 versus Starfield's launch performance on PlayStation 5? Where do you think a Switch 2 port might land? Oliver, so well, first of all, if they're doing a Switch 2 version, please do proper tech QA on it. We don't want to have any more situations of a title launching with crashing and stuff like that. It's just, I'm still sort of a bit staggered by that. But this is basically a GPU thing, which I think could probably handle. And then there's a CPU thing where there's a lingering doubt, right?

Speaker 3:
[31:39] Yeah, I would say that's broadly my thinking as well, though maybe with one caveat, I think that in series, it's a bit of a CPU disaster with a single threaded nature, single thread bound nature of Starfield's CPU performance profile, I think. Specifically in Aquila and New Atlantis, and those cities come up a lot, but like the other cities in the game, like that Resort Town or like the Neon, there are lots of other cities in the game that basically perform fine, and I would expect would probably be okay at 30 FPS on Switch 2, but those two cities in particular just seem to bog things down to a kind of ridiculous degree, honestly. After all this time, it really hasn't been sorted out that much. I feel like on PC or on console, it's still a major issue. But Starfield, in the open world, I think it could probably be okay. Those areas tend to be a lot lighter and they can run into like the 80 FPS, 90 FPS region on current gen consoles. I suspect those would be fine at 30 on a Switch 2. On the GPU side, it's also pretty hard in the GPU, or at least it has been historically, and I kind of wonder how much performance would scale if you just pumped pixel count down to like 360p, 540p or something, if you'd actually get the results you're looking for out of the game, or if it would be kind of like a Metal Gear Cell Delta situation where you're just dealing with some things that aren't really scaling so well, and then all of a sudden you're, things that aren't scaling with pixel count, you're in all kinds of trouble with shadow maps or foliage or whatever the case may be. It would obviously need DLSS, I think that goes without saying. I think we've seen that from Fallout 4, right? Fallout 4 was quite blurry before, it looks a little bit better, but I would really, they would really need to make the most out of their DLSS implementation in this game to make it work at all. But again, that really depends on getting this game to a really scalable state on Switch 2, which I'm not sure that it is. It's also not the most handheld friendly game either. So that doesn't portend very well.

Speaker 2:
[33:34] Interesting. Alex, what do you think?

Speaker 4:
[33:37] Oh, boy. I honestly don't think this would work that well due to the city portion of the game, where you do spend a good chunk of it there at times, getting missions from mission givers and the main questing. So there's that aspect of it. I think it would just be a very imbalanced presentation as a result of that. But that would be the first time a game with imbalanced presentation is shipped on a Nintendo console. So it wouldn't be something out of the ordinary, and it wouldn't be the first time. But that's the issue of the game with such issues. So maybe there is something to it. Like Oliver's case here, I would actually say just run it as high res as possible. At 30 FPS, just do that. Just eat the cost of it and just run the game at 30 FPS and just pray at that point. Because I really don't think this game scales well on the CPU, on the PC testing that I've done with it. And the thing is, I actually have never run it on a Steam Deck. I don't know what it's like there, but I can imagine one of the bigger issues is like, when you go between areas in the cities, it's not just the generally depressed frame rate, but there's zones that load obviously different parts of the cities when you go around them. And they really tumble and fall there on Horizon 5 3600. So I presume they tumble and fall even worse on something that has a lot lower average bandwidth and getting stuff off of disk. So, it sounds a bit rough, but maybe they'll do it.

Speaker 2:
[35:24] Right, mm-hmm. Yeah, I think they'll probably ever stab at it. I think if they limit it to 30 frames per second, they'll be able to at least have a cap option. Maybe not the, how many options was it on PS5 Pro?

Speaker 3:
[35:36] 24, yeah. If you don't count the PSSR toggle, then 12. But yeah, 24 and 12.

Speaker 2:
[35:43] Okay, yeah. So I think maybe, you know, go for consistency, at least have the option on by default. And I don't know. I think there is a question. I think there has to be some specific testing on using the DLSS Lite at a specific resolution versus the CNN model at a lower resolution. Do a proper quality analysis to see which one is objectively looking better or subjectively even. And go from there. The CPU side of things will be a bit troubling. But I think 30 FPS is doable for a lot of the content. And where it isn't at 30 FPS, maybe that, you know, we can expect that there should be some cutbacks, right? It's not beyond the realms of possibility or whatever to actually just get this running at 30. But there may be some compromises required. But, you know, you got to remember, this is a game where the engine is quite open-ended. You know, if you're going to be putting like 2000 sandwiches in your cargo hold, there could be problems. Okay, let's move on to the next question. This one comes from Wilmery or Wilmery. Hi, DF Gentlemen exclamation point. I haven't yet got a GPU that can handle path tracing in 4K, but I'm dreaming of one. There seems to be more and more games coming out with path tracing graphical options, and that makes the games look so good. I'm totally on board with team Alex. Aren't we all? We're actually not. Some people aren't. Some people aren't. What would be your personal recommendation regarding getting one? Buy a 5080 if one can find one somewhat close to MSRP, or just wait for next generation of Nvidia and AMD GPUs. Worth the wait? PS, can't wait to see what Remedy will do with Control 2. I'm with you on that one. Alex, it's the classic, should I buy now or should I wait argument really? But there is a specific use case in mind here.

Speaker 4:
[37:40] Yeah. I mean, if it's something that you really want, and want to do, and want to see if you want to invest the money into it, can I give you something that I would try out? Especially if you have a VRR display and a current Nvidia GPU, I don't know what you have, Will Mary. But maybe give the highest tier, what's the name of the Nvidia streaming service again? My brain is broken.

Speaker 2:
[38:07] GeForce Now.

Speaker 4:
[38:08] GeForce Now. Give these games a try on the highest tier GeForce Now, one month or 24 hour, or however long you can give it a play for the cheapest amount of money, just to see if you think it is visually worth the amount of money, because you're going to be getting almost 5080 like performance there, though with a very, very different CPU. I would imagine then like what a consumer set might use. Then if you really think you like that so much so, then consider getting a 5080. If not, then wait. That's what I would say because I don't want to say do this because I like path tracing so much. That's not cool. Buy things because Alex says so. But at the moment, I'd say give it a try. Since you don't have much experience with it, I think it is pretty fascinating and a lot of fun. But also do it around the games that are your golden cows. The things you care a lot about control to. Sometimes building a PC just for a game coming out and then using it for years after the fact, is a very satisfying experience. I built a PC for crisis back in the day. I built a PC right around the time I sure 2033 came out. That's a fun experience in its own right. Maybe timing it with that could be a lot of fun for you if you enjoy it.

Speaker 2:
[39:38] Interesting. Oliver, what do you make of this particular question?

Speaker 3:
[39:42] Well, I think a 5080 would be a good place to start because that's where we think the consoles will be roughly speaking. I mean, plus or minus 20 percent, whatever. Probably a little bit slower than that, but good place to be. I think if you're looking to emulate that console experience, I would think. And so in that environment, like getting a 5080 now seems to entitle you to a class of experiences that will be pretty well catered for your hardware for quite a while. I would think, I would hope at least. But, you know, I mean, among the path racing titles, I mean, there aren't that many of them, at least the big ones. They're generally Nvidia sponsored. You know, they come out from these custom engines and whatnot. It's hard to really draw up too general a plan because it does seem to differ quite a bit from game to game in terms of what that path tracing looks like and in terms of how intensive it is. I've seen recent reports that Pragmata is especially intense and has a number of issues at the moment with its path tracing implementation potentially, even unique to this game over and above Resident Evil. So it's very hard to offer one generalized recommendation, I feel like, but in general for high-end PC ray tracing titles, a 5080 at the moment is pretty good company to be with, I think.

Speaker 2:
[41:04] Yeah, I'm inclined to agree, but I'm also inclined to agree with Alex's suggestion to check out GeForce now because it's actually very, very good. It's a bit, slightly more latency, but you are getting a lot of performance. You can use path tracing there. You can use all of the features. It is a fully enabled Blackwell GPU. I'm not sure whether it is actually for all games because there was a kind of, when I looked at it, a partitioning system where some games were using the new GPUs, others were not, but it's still, the ones that matter would definitely be running on those 5080s there. Assuming the service is available where you are, of course.

Speaker 4:
[41:44] Yeah, that was an assumption.

Speaker 2:
[41:45] Yeah, that was an assumption, but that's actually a really good thing, really good idea because the image quality is pretty good, very good, I'd say. It runs well. You'll certainly get a good taste of what the full path tracing experience is, and then you plan your next move from there. Good stuff. Okay, let's move on. Final question. This one comes from SkyrimFest26FPS edition. What do you think of the idea of the quote unquote Series S version of PlayStation 6 being a Cloud Edition? It could even run on the PlayStation 5. This follows on from the discussion we had a week or two back, where we were talking about potential micro consoles or cheaper consoles, where we suggested that maybe there could be a Switch 2 console, a home console, or there could be a cut down or other version of the PlayStation handheld that also operates as a cheaper console. I've got great news for you, I have infest 26 FPS edition. Well, there's already a system in the Cloud that enables you to run PS5 games, and those PS5 level hardware will persist into the next generation, of course, and it's called the PlayStation 5 in the Cloud.

Speaker 4:
[43:05] Yes.

Speaker 2:
[43:05] So it's already there if you want it. I don't think it's perfect. I think the image quality is actually very, very good in the 4K mode, but the latency isn't particularly great. Maybe there could be improvements there, but that's basically another potential string to the bow, really. The concept of actually, well, your cheaper option is just not to own the hardware. It does tie you into a PlayStation Plus subscription, of course. That's the downside. But Oliver, thoughts?

Speaker 3:
[43:39] I don't think the cloud is reliable enough as something that you can have as a major pillar of a hardware launch. I mean, at least I don't really see it that way. I don't really think they're going to introduce the PS6 and say, here's the PS6 and then here's PlayStation and the cloud is our alternative lower cost entering point. I just don't think that works for a lot of people. I don't think it makes sense for a lot of people and it has some unnecessary complication in the sense that you still need that separate streaming hardware, a separate streaming device and ultimately it's not a good experience at the moment. PlayStation, I hate to say it in terms of its frame presentation, in terms of its latency. In my opinion, it's not very good on the GeForce Now side, maybe so, but I think they need to improve it quite a bit before it's really in the same vicinity as a console experience. I personally favor the idea of a Canis console. I think it's cool. I like the idea of real home hardware, real home console hardware that you can actually buy and put under your TV and it works in a very simple way. It's plug and play, works on an internet connection ideally. It could really lean on that existing Canis spec. It could just be like a mild clock bump, maybe run the same profiles, etc. It could be a very easy system to add into the mix. I think it's a cool idea. I'm less of a fan of these Cloud ideas. Even when they pan out, it's not really my preferred route. But I do think Cloud is taking off in terms of popularity, but it's taking off rather slowly. So I don't really think it's going to be quite all the way there for next generation console. I think Sony does have a product for those people perhaps, and that would be a PlayStation 5, but perhaps they could even have a lower price console in the Canis SKU.

Speaker 2:
[45:20] Alex, thoughts?

Speaker 4:
[45:22] Yeah. I don't think you should rely on Cloud. For all the reasons that Oliver and Rich said, and if you really want that thing, it already exists, you can try out that Cloud. I'm not a big fan of Cloud gaming. I think it really does need the GeForce Ultimate level for me to start wanting to consider it, and even then I'd point out issues. I'm a real snob. I know that. Not everyone's a snob though. So I think you could get away with it for certain people that just want to play games, and don't care about all these other aspects. And I think maybe there's a tier of that, though I don't think you should push it as part of your next-gen strategy. You should probably focus most on the hardware, and then on the side, if people want to do a bit of extra research about finding out if they can't afford a PS6, then they look for alternative methods, older PS5, optional streaming service.

Speaker 2:
[46:20] I think the concept of what amounts to a cheaper console for the next-generation period is still a viable question, however. It's just we don't really particularly think the cloud is the most versatile or desirable solution. So for me, it's basically between two. It is either going to be some kind of maybe the handheld itself, maybe an offshoot version of the handheld that is basically the equivalent to a micro console. The other possibility which nobody's really talking about is basically a PS5 super slim, if you like. Can they die shrink the existing 6 nanometer chip to whatever it is, 4 nanometer or 3 nanometer? I think the new consoles are on 3 nanometer. Which one is cost-effective? Which one can actually deliver a cheap console? Is it going to be the Cadence slash handheld micro-console, or is it going to be just a PlayStation 5 with a die shrink? Logic would suggest that it's got to be the die shrink PlayStation 5, surely. Less memory, a smaller die, surely that would be cheaper.

Speaker 4:
[47:34] I'd hope.

Speaker 2:
[47:35] PlayStation 5 is still going to be around. It's the question of whether the handheld SoC presents cost advantages or feature advantages that Sony wants to, for a more extended lifecycle, for that level of performance. I think it's a really interesting question that's going to be posed and I'm curious to see if Sony will have an answer to that. Because I can see the handheld being there to increase the overall size of the market to get more people invested into the Sony ecosystem. But if it's going to be expensive, like $600, that's problematic. If it's cheaper than that, probably not. I don't know. It's a fascinating question. But for Sony that doesn't have any wish, seemingly, to want to move its games onto other systems, it's going to need a way to appeal to a mass market, I think, which $500, $600 consoles probably aren't. Interesting question nonetheless.

Speaker 3:
[48:39] The one advantage Sony does have going into the next generation though, is the fact they've moved with 90 million odd PlayStation 5 consoles. They have a huge base of customers to soak and to sell games to and to sell PlayStation Plus subscriptions to. And they have a lot of used consoles out there, and they have a lot of your friends used PlayStation 5, might be a good entry point into the console ecosystem. And so I feel like they're going to really lean on those players as we go to the next five years, 10 years of cross-generation software. I feel like they're really going to lean on those players and use that to bump up their overall ecosystem profits as they have with the PlayStation 4. You know, there are something like 30 million, 40 million PlayStation 4 still in regular use. And all those customers lead to recurring revenue for Sony in a major way. So that at least is a position that Sony is in. That's actually a better position than Nintendo is in. We're trying to move obviously into new console generation in a much better position than I think Microsoft is in as well.

Speaker 2:
[49:40] Yeah, that's an interesting point. For reasons that will become evident in the next few weeks, I was looking for used PlayStation 5s over the weekend, and you can get one for like £250, which is very, very cheap. Okay then, so that was the final question there for the end of the show. Please do like, subscribe, share if you enjoyed it. Ring bells for notifications for you, potential algorithmic boosts for us, DF's supporter program. Become involved, join our Discord, it's amazing. High quality video downloads of everything we do, the chance to get DF Direct early. It's all good stuff. And pose questions for this very show. patreon.com/digitalfoundry, store.digitalfoundry.net if you fancy wearing clothes that have our logo on it. But that's all from us on this one. Thanks for watching and supporting Digital Foundry and we'll see you very soon.