DF Direct Q+A: Will AMD Deliver Its Own DLSS 5? Will CD Projekt RED Bring Back The RED Engine?

0:00Hello there and welcome to the latest edition of the DF Direct Q and A show. You asked questions, we provide answers to the best of our ability. It really is that simple. Johnny, we too go through some of these questions today. First of all, John Liderman, hello. You got questions? We got answers, maybe. Possibly. Yeah, possibly. We'll try. Based this conjecture, educating cases, a whole bunch of stuff, possibly some answers. Also, Johnny, we of course, Oliver McKenzie, hello. Yes, through a select, a sip between baseless conjecture and I'm educated to answer myself. Right, okay. Yes, probably the best position to be in. Okay. Right, yes. Before we crack on, just out out to the DF supporter program, patreon.com slash digital found fee, early access to a bunch of stuff, early access to DF Direct, the chance to pose the questions for this particular show, a bunch of stuff going on there, ad free website,

1:00ad free video downloads, all good stuff. Okay, so our first question, well, there's a bunch of questions actually in the wake of the launch of DLSS 5. There's actually not so much questions concerning the quality level. I think at this point, quite subjective, different people think different things. People coming to their own conclusions, and we've actually, of course, only have actually seen one proper game. The rest is all modding. But yeah, very interesting questions here coming nonetheless. Going to start with this one from Skyrim 26 FPS edition. How do you think AMD should respond to DLSS 5? Do they need their own version? I should they approach the next leap forward in graphics quality in a different way? Is there a different way? That's a pretty good question. Boot forcing it through selling more silicon felt unrealistic, even before AI hoovered it all up. Yeah. This question from CarVox. The Bond-Geor DF crew in the wake of the launch of DLSS 5, I'm curious where AMD goes from here, given FSR is quote unquote behind DLSS, does the panel think AMD will move to

2:02mimic Nvidia or continue on their own path? Thinking out loud, the feintime cost and delicate hand required to implement likely means few games will implement natively without Nvidia support. Then there is consoles Nintendo aside being AMD units as well. Will this feature realistically see significant adoption for developers if it requires that much potential oversight and resourcing? Merci, exploration boy. Going back to Skyrim Vest 26 FPS edition, what would you want to see from a DLSS version two? Do you think there is a ceiling with what seems to be a screen space finishing pass could offer? Do you think a version two might be improved by passing game engine data more relevant to its task rather than it's relying on information designed to help with upscaling? So a broad range of topics here, but fundamentally, I'm going to go to you first on this one, John. I mean, where do you think the competing vendors in the graphics base are going to go with this? Because fundamentally, as has been pointed out, it does seem to be the case that we've

3:03got a clear pattern that's been established at this point, which is the Nvidia produces a PC based technology and Intel and AMD produce their own variants of it. Do you think it's the same thing that's going to be happening here or something different? Yeah, I suspect that we'll see more GPUs taking this approach. I mean, just this morning, I was reading an article about inside the arm, Molly G2 Ultra and X GPU where they talk about this sort of exact thing. So they're doing it. There's no reason to believe it until an AMD are not also looking at this. But I'll be curious to see how far this goes, what this is used in. I'm still skeptical that a screen space style solution is the best approach for game rendering. I almost, I was thinking the other day if we're in a stage similar to when we first got things like screen space ambient occlusion back in crisis one where essentially they were trying to

4:03solve a problem that was still the computational cost was still too high to do it in any other way. So these screen space solutions ended up being a good, good option. And I'm kind of feeling like that might also be the case here. And that doesn't mean that neural rendering can't still be used in other ways. But I'm wondering if this is sort of like a stop gap where it's a way to get some of those advances, but they're applying it in this specific screen space way. But there's no doubt that all the other companies are almost certainly doing their own work on this as well. Yeah, I think I was just going back to the interview that Alex and I did in Berlin with Tap Tom Peterson from Intel. This was like five years ago when we were talking about Intel arc. And we, you know, we kind of had an idea that certain things would be going on and, you know, tap is hugely into AI, that sort of stuff. That's his key point of research there over until five years ago, he was describing a scenario where, you know, basically there would be

5:08a generative filter put on top of existing game graphics. Now his implementation, his use case there seemed to be that the user decides what's going on. But, you know, ultimately it's the same sort of thing, right, which is to say you've got existing graphics and then there's an AI pass on top of it. His particular scenario is sounded quite horrifying to me, I'll be honest to you. But it's funny that he actually mentioned that before we actually bought up the topic of frame generation. And, you know, he's sitting there saying, yeah, what if there's more than one generated frame? And it's like, well, that's exactly what's come to pass. So, you know, yeah, it's basically the case that all of these guys, I suspect, are thinking on very similar lines with some points of variation. I was thinking the other day, Oliver, that possibly one route forward would be bearing in mind the proliferation of Unreal Engine 5, bearing in mind, you know, you've

6:09got a system like Lumen in there. If I was AMD or indeed Intel or indeed Nvidia, I would be looking for some kind of way to integrate you or entering into the into the engine at a more basic level, at a more fundamental level that would hopefully overcome some of the limitations of screen space. That seems like a pretty, you know, decent first start. I don't know, what do you think? Yeah, definitely more integration into game engines, taking in more inputs, integration into engines like Unreal. I'm sure there are a lot of interesting areas people to go. I think that would probably be more epic provenance though when we talked about Unreal Engine specifically in terms of integrating those technologies and probably less the vendor specific stuff, but I don't know, could happen for sure. But I tend to think with like thinking about the limitations of what screen space should be in this instance, I kind of think the limit of this approach would be kind of somewhere in the vicinity of a skilled lighting artist, getting an image, just still image,

7:12finding lighting faults with an image and fixing those issues with a copy Photoshop that would be kind of like roughly the limit of this technology I think at present, because that's basically the input that DLSS 5 has, and that's the human limit. So you'd think maybe they could get something closer to there. I don't think DLSS 5 is anywhere close to that at this point, and I think a future version based on this technique could probably get substantially closer to that point. In terms of like a hard limit on what it could deliver, it depends on the kind of lighting signal that's present in the image you're getting, and to what extent like an off screen object is contributing to that lighting, or should be contributing to that lighting. For instance, are you getting a nice shadow from an off screen object that DLSS 5 can use to infer that there's an object there or something, right? That's like very helpful lighting signal. But apps that lighting signal then DLSS 5 shouldn't be able to build that image, that detail into the image. So you'd expect a technique that takes in more inputs from the game engine, or constructs a kind of like model of the

8:16scene, does something exotic like that, or just takes in more frames as input. That would be necessary, right? But I think there's probably a lot of optimization space in this existing technique to push things further. That's my expectation. Even absent something that's like, I don't know what Android video will do next, or neither render will do next. It's entirely plausible that they could go on a direction of taking in more inputs from the game engine. I'm kind of shocked that this works at all with just colored motion vectors. That's the crazy part to me, but I think they will push it in possibly new directions. I think they'll enter it in this technique. I don't know where that'll take them, but I would expect it if they did like a DLSS 5.5 or DLSS 6. That's just this technique, more data, more training, more parameters, whatever. It could probably be scaled somewhat further. But I do expect other vendors are working this technique, and it's not a novel area of exploration. I think I said something on DF that was something like this, because I was experimenting with video models. I was experimenting with video models for a couple of years before this whole thing broke, and you could do kinds of interesting things, video models. I'm sure lots of people are thinking

9:19about generative filters and things like that over game graphics. I don't actually think this is a new area of exploration. It just, and video has been the first to productize it and take it into real time with interesting quality. So I expect everyone's going to be working on some 8s. But yeah, I mean, how close those products are to delivery? I don't know. We don't really have a good re-reconstruction alternative from AMD at this point, even though it's been over three years since we were a construction debut. So it might be a while before any analysis is actually delivering a comparable product. Right. Yeah, I mean, I think we've all agreed that screenspace only goes so far. There's a really basic example of this, right? I mean, when I first loaded up the very first DLSS5 mod with control, you know, basically you go into a room where there's a very strong rate-raised reflection, so I'm showing something that's not on screen. And DLSS5 can't magic up a rate-raised reflection, right? It's limited by what is on the screen at that moment. And in theory, what you

10:22would want would be some kind of unified rendering pass that would incorporate what's happening in terms of like the transformation with lighting, including all of that information, which isn't never going to happen on a screen space. You could do 10 times, as we've seen with some mods, 10 times DLSS5 filtering, it will never be able to make up a detail from something that isn't on the screen at that point, unless the model becomes like astonishingly, like spatially aware, in a way that we're not seeing so far, which again requires deeper integration into the game engine. There's almost this todological thing where I can say, well, what if the lighting detail was already in screen space, but then all of a sudden at some point when we're talking about, what if the lighting detail was already in screen space, all of a sudden we've invented like a perfect rendering thing where like we've got all the rendering already, we've got all the screen space lighting already that we need in DLSS5 is doing nothing, but yeah, for sure, like that is a hard limit of this approach. And you do see that in some games like Final Fantasy VII rebirth where

11:22you have like a lot of conflicting lighting signals that are present in screen space, like indication that some surfaces are able to reflect the environment incorrectly, and DLSS5 doesn't really know what to do. Right, so in those cases, especially you're like, well, I wish this had a lot more inputs from the game engines that could figure this stuff out, but in other cases it works better. We didn't have a question about this, but I certainly have a question of, I guess I'm pretty to you first Oliver, I mean the whole mod situation has like ballooned out of control in a way and video I don't think ever anticipated. It's quite remarkable some of the stuff that we see. First of all, the concept that they would introduce compatibility for RTX 40 series and then 30 series and then 20 series and then AMD, you can run DLSS5 on an AMD 9070XT, it doesn't run very fast, but you can do it. That's like kind of staggering. There's a kind of like rich I really hear,

12:25Oliver, remember when quite rightly there was pushback against the concept of DLSS5 being demonstrated using 2GPUs, right? I mean, it kind of suggested tech demo more than realistic application. However, now that's been brought back, people have put this a mod that they would use to offload the DLSS5 rendering to a second video GPU. I'm getting the feeling probably a net positive for Nvidia because people keep talking about it, therefore the discussion about DLSS5 is continuing when I think if it had just been MBA 2K27, the interest level probably would have died off within a few days a week, etc. This thing is just keeping the news channels flowing in terms of DLSS, so it keeps people talking about it. But I'm curious what you make of all of these developments. I think one unintended consequence of this from Nvidia's perspective is that they've announced that they're going to be doing an official implementation for 40 series GPUs.

13:29This was never indicated beforehand, no idea whether it was even a thing for Nvidia before this mod happened, but the mod is basically forced yet to happen, right? Yeah, that is the case. And, you know, I mean, ultimately, the kind of math and optimization that goes into these models, I mean, they can have significant optimizations for specific architectures, but in terms of precluding a feature like DLSS5 from an architecture like I think it's 40 series, I think the rationale for that is fairly weak and is tied more to a very specific lock-in for certain architectures for commercial reasons. That's not something that we haven't seen before, and indeed AMD does their own share of under-lock-out to specific cards and architectures, so it's nothing too surprising there. But I mean, I'm glad to see it happen. In terms of the proliferation of DLSS5 across all of the architectures and all the GPU configurations and all the craziness out there, I think to some degree people are doing a lot of reverse engineering using

14:32AI tools, which has made a lot of this stuff a lot easier and a lot less labor intensive than it used to be. And in terms of like multi-GPU configurations, I think that's kind of cool. Like I think that it's kind of awesome to do multi-GPU with this stuff and to actually find a good use for a second GPU, because modern game rendering is not well suited to multi-GPU setups at all, but all of a sudden you get this expensive neural filter running on a second GPU that seems pretty cool. Another cost to latency is prohibitive probably, and I would think that in the future efficiency improvements will make this much less practical. But in the short term, I don't know, I think it's pretty cool. I could see why Android didn't expose that functionality to users, though I think it sends the wrong message. And ultimately, I think it'll be kind of a stigial when this model is that like, you know, I don't know, in a year we're looking at a similar model at like four milliseconds or something for 4K output at a 5019, or like this isn't, you don't need two of these. So, you know, I don't think it'll be as necessary at that point. I mean, I've not looked at this particular mod yet, but obviously, latency is going to

15:34increase. It's not going to be a fantastic ultimate solution for it, but it is just tickled to be nonetheless. But another thing that's happened is that the point at which the DLSS-5 filter is being introduced has been swapped to basically pre-super resolution. So basically, the newer rendering pulse is being done before the upscaling rather than afterwards, meaning that the frame time cost has dropped dramatically, but there's probably going to be a significant image quality in the process. But it's just kind of baffling how people are just sort of, well, not baffling, but kind of intriguing almost how this thing leaked, was reverse engineered, and now it's being sort of experimented on almost in a kind of like sort of cloud-sourced way. I've never seen anything like it. I mean, in a way, it just reminds me of whenever we get some kind of new features for like reshade or something like that, right? Where just a lot of people have fun fiddling with it because

16:36it's new technology and it does something unusual. So, regardless of one's opinion on it, it is neat to see like what people are trying and also the limits to which it can go. And I'm surprised at how certain, I've seen people applying it to like PS2 games, like Gran Turismo 3 and Torus Trophy, and the results are like fascinating because when done right, it doesn't really add that much, but it adds just enough extra shading and like shadowing that sort of enhances it ever so slightly, given that this, it's an interesting look, but it just feels like everybody's just having a, having a field day here, just putting it in everything. And of course, the usual suspects like, oh man, we got our GTA 5 Ultra Mega hyper-realistic mod with DLSS 5 extreme, you know, that kind of, tons of videos of that, often running it like 20 FPS, but hey, yeah. I'm also shocked by the amount of people that are

17:36doubling triple passing it and presenting that as like the standard output. I mean, that's pretty good. Yeah, it's just like not intended and also you're just abusing, I think you're abusing the model a little bit, but then people see that it's like, oh man, this is DLSS 5, it looks so bad. And it's like, well, you know, it's like being used totally tastelessly, but okay. I do find a lot of those older emulator games kind of interesting, because I've noticed that DLSS 5 really wants to, into it, a kind of like screen space reflection in those older games. Oh, yeah, I know that it's. Streets, they're always kind of glossy, but this kind of semi-glossed look. DLSS 5 has a very specific way it looks to construct what I'm calling like, the role screen space reflections, because obviously it's not like a discretized phenomenon in the same way that would be in conventional rendering, but it is kind of like that same visual appearance of a reflection. And it seems to want to do it all over the place, and it often looks, I think it looks good, and it looks sort of plausible, but often not correct for how that material should behave,

18:36if that makes sense. It kind of looks plausible like a generative image model would come up with something, but it's not quite correct, I think, to how the material would look if it had like proper pre-VR properties, but it's kind of cool regardless, even if it's kind of like, well, obviously not intended by anyone to look like this. I think the thing also about the second GPU mod, that kind of works as long as it's basically a past that's added to a final frame. So basically, you've got your final frame, it gets offloaded to the second GPU and processed. If the endgame is basically to integrate it more tightly into engines, then we're kind of like facing the same problem that we had with SLI back in the day, where sharing all of that data could currently between two GPUs is just not viable. So yeah, I mean, if that does happen, it's going to be a relatively short-lived thing, I think. Let's move on. Okay, so we got this question from Scott. Hey, Chaps, having looked at the crimson desert DLC

19:38trailer, I'm blowing away again. I'm sure a fair amount of this was content that was at least in planning stages before release, but it's crazy how adept they are at integrating wholly new mechanics at speed. They have to be one of the most agile and efficient devs out there. I saw someone point out that if there were a dev of the year award, they would be frontrunner for sure. I wholeheartedly agree, what do you reckon? Well, I think it's fair to say that the amount of updates, patches, fixes, response to community requests from the Pearl Ability has, I mean, I can't think of anything quite like it for quite a long time, if at all. What do you reckon, John? I don't know their secret, but clearly they have pretty good planning and production capabilities. It kind of caught me off guard because I wasn't really that aware of them during the Black Desert era. They've done an amazing job with crimson desert, not just producing content, but answering and responding to fan feedback, you could say, and they've done a lot of work fixing and improving

20:41the game. It reminds me a little bit of a CD project situation where there's very agile in terms of working on the game and getting it to where people are happy. And because it's such a big game, it kind of works. It's not a one and done kind of experience. It feels like the type of game like, like a Witcher 3, like a GTA where a lot of people just spend time in that world playing it. And it has such a long tail that they can continue to add stuff to it. And they're really efficient at doing that. Oliver, I agree with John. I do think there are some things about it. They're a little bit overwhelming though with the pace of updates and patches and the kind of changes in terms of the way they present things to the player. I think it, you know, like to some degree, listening to fan feedback can be helpful, but at some point you kind of need to put your foot down and just say, this is how it's going to be. I think to some degree, but I think we should maybe at some point, when the release schedule slowed down a little bit, perhaps one of us should take a look at the

21:44game and it's current state and see if any of the issues that have been raised a few of the issues that John pointed out in his coverage and Tom pointed out his coverage have been resolved or the extent to which they've changed. Because I think the game has gone through a pretty radical least by the pace of updates, at least as far as the pace of updates is concerned, set of changes. And I'd be very curious to see if those issues have been resolved, if the infamous Bug Hill, runs better now. I mean, I've done some testing recently than in a kid's perhaps, that's not the case, but perhaps they've made some other changes to like upskilling and things like that. Okay. Yeah, I mean, maybe we should do a redo at some point, check out how the game has progressed and maybe tie into the DLC launch. Because it's clear, I mean, they are doing some new stuff here, and it is, I do think it seems to be mostly tweaks about just trying to think here. I think it was when I was doing the steam of machine review. I did speak to Pearl Abyss, and I said, you know, your console equivalent setting still the same, because there's been a lot of work that's been carried

22:47out. It was, well, yeah, pretty much. And nothing really seemed to have changed from a sort of GPU and CPU perspective from that regard. So I think it was maybe features and stuff. I think the real issue is possibly if there's anything that's going to have to happen for the Switch 2 version that can be backported, just kind of like that the promise that's been mooted for oblivion remastered. So yeah, curious about that for sure, but Dev of the Year, John, thoughts? It's difficult because it's hard to quantify, right? Yeah, how do you define the developer? Because, you know, Capcom is Capcom, right? But there's a whole bunch of Capcom teams. I was going to say, but Capcom has been delivering with this like amazing consistency. They put out three crazy good games this year that are all pretty technically solid. And you know, they've doing all this within reasonable development times. So there's that. I mean, there's probably others if you think about it, but I will say that Pearl Abyss is doing some things different

23:52than usual. And they definitely deserve credit for it. Yeah, absolutely. Yeah. Okay, let's move on. Question here from Jotsuya. What do you think is the story behind Capcom's or other conservative use of their own RE engine in Onimusha? Having treated us twice this year with beautiful hair strand rendering and spectacular pathfacing, Onimusha is visually a huge step backwards. Those flatly lit interiors really sting. We know RE engine scales well, so why not offer higher end visuals, even if only on PC? Well, it's an interesting idea, right, John, but there is kind of like probably much more to implementing a feature than just turning it on because you used it in another game running in the same engine. Precisely and clearly they decided whenever they started development on this, they selected specific visual features of the engine that they wanted to pursue and clearly built the whole game with that in mind. So this was not a part of it. And I can respect that decision. They basically went for stable performance and image

24:56quality above all else, I would say that was like the focus. And yeah, it would have been neat to see that. I do think it could have been useful, but also as we saw with Resident Evil Requiem, the global illumination stuff when done in RT can be a bit noisy and artifacty and like sort of dark environments without a lot of lighting, which only Moussa has a ton of. And that could have been somewhat unappealing for a lot of people and it would have just doubled the work in terms of now you got to do both the RT lighting, but then also like a baked pass. So I mean, ask with anything the developers made a choice and that's what they went with. And I think it's, you know, it is more visually conservative than their last two games, but it's still attractive. The artwork is excellent and it runs very well. Get good image quality. I'd say they've done a good job occupying things conservative. Yeah, I think the other thing of course is, you know, basically Capcom have become masters of putting out games on an even sort of schedule.

26:03The throughput this generation has been astonishing compared to other AAA studios. So, you know, it's and of course budgets of key importance as well. So, you know, we can't expect every single game to use every single feature of the RE engine, you know, in order to you know, just sort of get an even sort of release schedule of games out there. And I think that's part of the genius of Capcom is that that scaling projects intelligently, they're delivering in ways that pretty much nobody else is anything to add Oliver. Oh no, just to say it was, you know, it's built on a certain set of visual features, probably keeping a platform like the Switch 2 in mind, which the game seems to target pretty well. And it might not be worth it to do like a bunch of duplicate art passes for what might end up being just like features for high-end PC features for maybe PS5 Pro, might not be worth it as John pointed out, the PS5 Pro, you know, does have pretty good RTGI,

27:05interesting looking RTGI in RE Requiem, but the image quality you take to get there does not satisfy totally with some noisy nests and whatnot. Uh, honestly, she's also, I don't know if this is correct, but to me it seems like a bit of a smaller scaling in terms of production, just in terms of the kind of production resources that will be afforded that total relative to a total like Resident Evil 9. I don't know if that's your feeling as well, John, but you know, could be a factor there as well. But also Capcom has a bit of a history with this because, you know, in 2024, they delivered Dragon Star looking direct lighting in that title overall and kind of mixed visual outcomes in that game. So it's not actually something that Capcom has committed to wholeheartedly this kind of shift to

28:09RayTGI's lighting. They've adopted it on a project by Project Basis as it makes sense, and they may continue to do so in the future, right, but for Anamusha apparently didn't make sense. Okay, let's move on to a question here from the Adam. Good to talk, DF Crew. We have the latest update to Red Engine. I had the crazy thought that CD Projekt Red might not actually be discontinuing the development of its own graphics engine. Could UE5 simply be used as a bridging technology, allowing CDPR to buy time and knowledge for a new Ed engine. One shouldn't forget the business aspect either keeping in mind that's having your own engine saves high licensing fees, especially given the popularity of CDPR games. Maybe the Red Engine upgrade for the Witcher 3 is our first sign of this as always. Thanks a lot. And wishing good health to Alex Amen on that particular front. John, I mean, this is the kind of thing that the kind of internal narrative that kind of like you want it to happen, but it's not really likely.

29:13Yeah, I mean, I think it's fair to say that at least this update and the stuff that keep doing on Cyberpunk probably keeps engineers working on that engine to some degree. So it's not like they have completely put it to bed, but I'm not sure we can read anything into this about where they're going in the future. And it certainly doesn't affect what you're for. They've made those choices for Unreal long ago. And that's how the games being built. So I don't think it's going to change anything on that. It's just that they have a very long tail on support for their older titles, which means they do need to keep some of the engineering talent around to work on that Red Engine stuff. It's really cool of them to see that they're continuing to expand it though, I must say. Yeah, absolutely. Oliver, what do you think? I think when you look at the kind of commitment that CD Projekt is made to Unreal, including like, you know, co-development efforts with ATHIC and committing to its technology, I think it's very unlikely that they switch, especially when you

30:16consider the fact that, you know, switching to Unreal is a bit of a process in and of itself. And I don't think you go down one direction, go down one path, and then switch mid-streams to another path. Even if there is continuing work on Red Engine, like John pointed out with the Cyberpunk updates with the Witcher 3's new path racing upgrade for its remastered edition, which comes up in a couple weeks, I think that's just kind of work that I don't know, might share some DNA with like work that's being done with Unreal, but I think it's ultimately work that will not have a path in terms of their core technology. I think when you judge the work they did on the Witcher 4, and it's demo from Unreal Fast last year, I think you can see they're achieving very, very good results with the engine. I don't think they particularly have a need to switch back to Red Engine, at least if they're able to overcome, you know, I guess some of the characteristic and real engine issues, but you know, they were showing an off at 60 FPS with OKMage quality in a PlayStation of 5's that's just a reasonable level of optimization even at the stage.

31:18Going back to the Adam's point here about, you know, high licensing fees, I think that's something that we can only speculate on. This was sort of announced as a partnership of sorts with CD Projekt Red technical talent, essentially contributing to Unreal Engine 5 for the benefit of the whole engine as I understand it. So, you know, I'm sure that the deal that CD Projekt Red has with Epic is probably on along different lines to people who are just treating it as an off-the-shelf tool to be used. This seems to be like a proper partnership, so we can't really speculate on the business viability of it. That said, obviously somebody at CD Projekt Red ran the numbers and ultimately decided that using Unreal Engine 5 was of long-term better benefit for their games and their business, so that's really why they went for that particular direction. It is a shame to lose the Red Engine though because, man, you know, the things it's done for gaming,

32:21for this industry, the technical levels that they've achieved with that. It's just been fantastic. Let's move on. Question here from Dan. So, in these recent interview with the Wall Street Journal, suggests they're moving away from traditional hardware, manufacturing, and moving more towards software and content creation instead, given that they have now outsourced production of their television sets to accompany in China, could they eventually do the same with PlayStation as well and adopt a 3D-O type approach, especially given that most of the chips within it are now simple off-the-shelf PC components, with no disk drive for Sony to manufacture, and Sony have already admitted that PS6 onwards and are no longer be sold at a financial loss. My prediction is that PS6 will be the final Sony produced machine, and the PS7 will either be made exclusively by third parties, such as ASUS, or will simply become a downloadable steam type app for Windows PC, especially if the backlash over the lack of physical disks continues exclamation point. There's quite a lot of extrapolation going here, Oliver, but I do think there is

33:26one fundamental point, which is kind of difficult challenge, which is that the viability of producing a mainstream priced console box has been getting more challenging generation upon generation, and especially with the crisis we're now facing in the hardware space, one might even suggest that it's looking increasingly unviable to make a console, even in this generation. I'm curious what you make about this. I'm not sure the extent of which it can be tied to Sony's other moves in other markets, mind you, thoughts. No, I think the whole pricing question, it really depends on the extent to which other vendors could get better pricing for a large of the same components than Sony could. Like, could, if Microsoft were to make a PS6 or something, could they source RAM modules at a much better price than Sony? Could they source SSD modules at a much better price than Sony? Those are the two key cost drivers that are going into the next generation, and I think Sony

34:30probably has good enough buying power that they can source those components for reasonable price. I think when you're thinking about third-party console sales, you have to think about what would a third-party bring into it in general? I think one avenue would be custom chips, but you don't really want custom chips for a console. You want one chip for a console, right? That's kind of the whole point of console development. What it be scale, I think I've already covered that, but would Sony not be capable of delivering the scale? It would be necessary for a successful console. Clearly, they've proven they can ship hundreds of millions of consoles in the single generation of that issue. What it be like something like distribution, I think Sony has pretty good distribution throughout the world. Maybe they could do like a custom, like an intended, like the IQ console for the Chinese market, right? Back and way back in the day, Sony could do a similar kind of region-specific console, although when you think about big markets, they've already tapped most of them, including China, you can buy a PS5 in China. I don't know about game availability in China, maybe you need to switch

35:32your console to a different region or something to get a pretty good selection of games, but you can buy a PS5 just about anywhere you want to go. I think the idea that Sony needs to move to third-party licensing for a PS6 or a PS7 or something just makes a very little sense to me. People always propose that it's a kind of, are often proposed that it's a solution for whatever ails the console maker, but I think when you break it down, I don't actually think it makes a whole lot of sense. For Sony or anyone else, that's why I don't think many people have followed in the 3DO example. Maybe for Microsoft, it makes more sense, but I don't think it's the right solution for Sony. What, why does it make more sense for Microsoft because of the PC connection? Yeah, because of the PC connection primarily, and because of the, I don't know, if it's a smaller volume product and you want to move it through more sales channels, I kind of get it more in that sense than when you're trying to ship a console that's going to reach like 100 million users, maybe. I think it makes more sense than that environment. Yeah, shipping a PC as console blocks

36:35kind of already opens you up to competition with some of the third-party OEMs. Maybe you'd rather have them on-side, I guess, with your own product. That makes more sense, especially when you're trying to merge those two ecosystems. Okay, John, I just can't see Sony adopting 3DO type strategy, and I got to admit the concept of a Steam type app for Windows PCs. Basically, you're putting yourself up against Steam to a certain extent, which is a battle that nobody's won, I guess, if they had like fully exclusive content. But typically, the response for PC users has been a firm no thanks. And you lose all of the benefits of running on a fixed platform, which is something that Sony has done exceptionally well with. What do you think about the whole question being posed here, though? Yeah, I don't see that necessarily coming to pass. There's no reason for them to do that, doesn't really benefit them in any way. As you noted, Oliver, Microsoft has more of a purpose in doing this, because they are a software

37:36company. First, they own Windows. Anything they can do to push Windows, and that sort of platform benefits them, whether they're the sole hardware providers or not. But with Sony, they want to sell people into a closed ecosystem. That's kind of the point. Nintendo is the same really. So, going, opening back up to the PC in that way doesn't make sense to me. And on the console side, I don't really see any reason in them taking a 3DO like approach either, because the hardware vendors aren't going to make the money necessary to make that even work. They would have to sell it at an inflated cost, and they wouldn't reap the benefits of the software. Yeah, I think the problem with the 3DO type approach is that you're actually adding an extra person into the chain that needs paying. So the cost is only going to go up, which is why I've often been thinking about obviously the graphics card market, where Nvidia can put out a Founders edition

38:38card. And yet, most of the cards sold are coming from a Sue's, MSI, etc. Somebody is making more, yeah, there's an extra person in the chain there that needs to be paid. And I don't see how that can be good for the end user, and the price is obviously have to inflate. Yeah, I think this is also quite interesting that it's a question that nobody would ever ask of Nintendo. It's never going to be like a video type situation, being asked of Nintendo. It's a different kettle of fish with Sony and the way they're sort of set up. Okay, let's move on. My final question is the show, and it comes from Jonathan Hamilton. Is it okay to bring up a merchandise issue? I've received my DF pint glass, and I'm very happy except for one small thing. It arrived empty exclamation point. Well, here is the Digital Foundry pint glass, which I've been coughing from throughout this entire proceeding. I'm sorry, they arrived empty. It's so sort of, because initially we were kind of thinking, oh no, this is going

39:43to be a bit of a sort of support problem. Do we have to send a new glass, but no, I think he was expecting some kind of DF themed beverage with it. Job. Yeah, I mean, we could add a beverage option, I guess, on the checkout. It's like, what do you want in the glass? Pick your poison. Then we have to figure out how to deliver that, and we have to come in something else, unless there's a CEO for the pint glass. We can only provide the receptacle. I don't think we're in the business of moving into the beverage use business at this point. Probably not. No. Do you think there should be a DF beverage? Well, it'll make it much heavier, and I also think the customs and import situation with transporting food or beverage across borders, internationally, is like a little dicey. So I think that's probably a bit challenging. Even if you were to just ship like a can of beer

40:45and something like that, you could go to think that's like a, that's like an elicit good. I mean, really, it is a good, that's age restricted and most jurisdictions. So that seems like a bit of a challenge for me, but yeah, I'm sorry I derived them, but I don't think there's anything we can do about that, unfortunately. Well, I was just thinking back to some of those CES to X you did with Alex and I should red bull, should we just put a can of red bull in it? Oh, she's right. I tried to, I'd extorted anyway. I tried to cut out the parts where I was checking red bull, because we recorded those at like 2am or 3am in the morning, but yeah, I missed a couple of the, I was, I was pounding back red bulls from the, from the bar there, pretty badly. Wow. Quite, quite astonishing. Yes, but if you do what a digital foundry pint glass, please do visit store.digitalfoundry.net. But that was the final question. Therefore, the end of the show, so like subscribe share all that stuff, bell ringing, etc, etc. Patreon.com slash

41:48digital foundry for the DF supporter program early access to DF directly access to some other content. Lots of good stuff happening there. All add free, of course. And yes, as I just said, store.digitalfoundry.net for our merchandising wears. But that's all from us on this one. I guess we'll see you on the next direct proper. See you later.