---
title: 'Hands-On With Steam Frame, Half-Life Alyx Standalone + More - DF Direct Q+A'
source: 'https://youtube.com/watch?v=e101n0Y6XGA'
video_id: 'e101n0Y6XGA'
date: 2026-09-16
duration_sec: 3087
channel: 'Digital Foundry'
---

# Hands-On With Steam Frame, Half-Life Alyx Standalone + More - DF Direct Q+A

> Source: [Hands-On With Steam Frame, Half-Life Alyx Standalone + More - DF Direct Q+A](https://youtube.com/watch?v=e101n0Y6XGA)

## Summary

In this Q&A show, the panel discusses their hands-on impressions of Valve's Steam Frame headset, including its design, battery life, and the technical achievement of running Half-Life Alyx standalone via eye-tracked foveated rendering at 36 FPS. The conversation then shifts to a viewer question about frame pacing issues in Japanese games like FromSoftware titles, exploring whether these stem from language barriers or developer perception. Finally, they discuss potential future directions for DLSS technology and humorously speculate about IKEA collaborations in Elder Scrolls 6.

### Key Points

- **Steam Frame Pricing and Market Position** [00:00] — The Steam Frame is priced north of $1000, which is tricky compared to the Quest 3 at around half the price. Despite component shortages, Valve remains committed to VR as a passion project, with a niche but dedicated audience.
- **Steam Frame vs. Quest 3: Ecosystem Choice** [01:46] — The Steam Frame is a more refined device than the original Index, but its price point puts it in competition with the Quest 3. Owning a Steam Frame is a choice to stay within the Steam ecosystem rather than Meta's, requiring a Steam account instead of a Facebook account.
- **VR Market Viability** [02:56] — Despite low audience numbers, VR games are often profitable because budgets are kept in check and the audience is dedicated. Valve's support, exemplified by Half-Life Alyx, aims to push VR quality without compromise, benefiting the hardware ecosystem through API and driver development.
- **Steam Frame Compromises and Impressions** [05:23] — The Steam Frame has some compromises, like monochrome pass-through cameras and a battery life of about one hour standalone or two hours streaming. However, it is noted for being very light and comfortable to wear, with the battery situated at the back for balance.
- **Battery Life Trade-offs for Comfort** [09:29] — The battery is relatively small to maximize headset comfort, resulting in shorter standalone use. While some see this as a downside, most agree the trade-off for ergonomics is sensible.
- **Half-Life Alyx Technical Triumph** [13:38] — Valve spent 18 months porting Half-Life Alyx to Steam Frame. They achieved this by running the game at 36 FPS with interpolation and dynamic eye-tracked foveated rendering, which reduces resolution in peripheral vision while keeping the central vision sharp. This allows for the same graphics and level sizes as the PC version, including workshop support.
- **Frame Rate and Foveated Rendering** [14:36] — The 36 FPS is common in standalone VR and is interpolated to 72 Hz. The eye-tracked foveated rendering is a game-changer, allowing higher perceived image quality by only rendering where the eye focuses at full resolution, a feature not present in Quest 3.
- **No Compromises in Game Content** [16:41] — The port retains all PC features, including level sizes and workshop support, without downgrades. The only visible difference is lower resolution in peripheral vision, which is virtually unnoticeable in practice.
- **SteamOS and Desktop Access** [18:04] — The Steam Frame runs SteamOS, providing a full-fledged PC experience in VR. Users can access a Linux desktop directly on the headset, offering more than the simplified interfaces commonly seen in other headsets.
- **Half-Life Alyx Visuals on Steam Frame** [22:17] — The visuals are impressive on a mobile chipset, with minimal aliasing noticeable. The game runs smoothly at 36 FPS, and the experience is comparable to PC, though using a joypad can expose the lower frame rate during lateral movement.
- **Cyberpunk 2077 Benchmark** [23:10] — Running Cyberpunk 2077 on the Steam Frame via remote play shows a 12% performance loss compared to in-headset results. At 720p FSR2 balanced, it achieves 22.4 FPS, highlighting that the CPU is a limiting factor for high-end local gaming.
- **Steam Frame Position in VR Market** [25:09] — With Meta stepping back from VR, the Steam Frame is a niche but compelling device. It improves on comfort and streaming but doesn't make VR a mass-market proposition. The whitelist of 120 games is expected to grow over time.
- **Frame Pacing Issues in Japanese Games** [27:09] — The panel discusses why many games, particularly from FromSoftware, have poor frame pacing. They dismiss the language barrier theory, suggesting it's more about developers not perceiving the issue due to dev tools and remote setups, even when frame pacing is formally correct.
- **Developers' Blindness to Frame Pacing** [29:10] — Many developers fail to see frame pacing issues because they often work with debug builds, rely on remote workstations that introduce their own issues, and some simply can't visually detect the problem. This leads to shipping with unresolved issues, as seen in several Japanese titles.
- **Alternatives: Unlocked Frame Rates** [31:41] — Some developers opt for unlocked frame rates to avoid frame pacing issues, but this also has drawbacks, particularly without VRR. The panel appreciates developers who properly implement frame pacing.
- **Future of DLSS Beyond Upscaling** [35:14] — The panel discusses potential directions for DLSS, including neural rendering and latency reduction. Oliver suggests a neural restyling filter for cell shading or animated looks, while John notes that frame generation's latency could be addressed with future reprojection to the future.
- **Latency Reduction and Neural Rendering** [36:05] — Oliver highlights the potential of Reflex time warp to reduce perceived latency and suggests NVIDIA could fold Reflex under the DLSS brand. There's also interest in neural restyling filters for emergent game visuals.
- **Frame Generation and Latency** [38:15] — The constant need to buffer a frame for interpolation adds latency, which is the main drawback of frame generation. Future solutions could involve intricate reprojection techniques, though exactly how is unclear.
- **Upscalers Reaching Maturity** [40:14] — Modern upscalers like DLSS 4.5 and FSR4 are good enough for many users, suggesting the focus might shift to other improvements, such as innovative AI-driven features like DLSS VR or frame generation at arbitrary refresh rates.
- **DLSS Ecosystem and Reverse Engineering** [42:57] — The reverse engineering of DLSS features, like multi-frame generation on older GPUs, is expanding the technology's reach. Even non-NVIDIA hardware, like the BC250 mining card based on PS5 silicon, can run FSR4 through community efforts, which is astonishing.
- **IKEA Collaboration in Skyrim** [45:08] — The panel humorously discusses the Skyrim IKEA collaboration, with John recalling how players store items in Morrowind. They joke about potential IKEA furniture in Elder Scrolls 6, with Rich expressing boredom with DLSS mods on social media.
- **PSO Reborn** [49:18] — John discovers that the Dreamcast game Phantasy Star Online has been ported to the web browser, running at high frame rates on psoreborn.com, evoking nostalgia for the game and early online gaming experiences.

## Transcript

Hello there and a warm welcome to the latest edition of the Digital Fantasy Direct Q&A show. You have questions, we hope to have some kind of answers for you.
Joining me on the panel this week, first of all, Paul V, his last direct appearance for a couple of weeks. John Lindman, hello. Yep, that's right, Rich. I'll be away for a while, so not here.
and also Audie and Mark and Corey are doing a DF Direct, sorry, not DF Direct, a DFW Super Show on the way. I'm curious to see what they come up with. I've heard rumors that it might be a topic that you're not going to love.
I've heard those rumors. I think the audience might love it, but you probably won't love it. Okay. That sounds good to me. Also joining us, Oliver McKenzie, hello.
Yes, Rich, for good or for bad, you're stuck with me for the next little bit, I'm not going to worry. But yeah, it should be a fun episode today. Okay, yes. And, well, I guess we should just tack on with the content that we've got for you.
Okay, so our first topic isn't exactly a question as such, but we did this out on the Steamplane Embargo for various reasons. But here we are, we did receive a unit, we have spoken to Valve about the Steamplane.
I am going to take it off now because it is very, very disconcerting that you're not actually in a VR experience. So the reviews are in.
I'm kind of curious going to you first, John, what you make of the various feedback that Valve has had to this because it's an interesting device in very interesting times, you know, and certainly the price point has, well, I don't think at this point
any kind of price point would surprise anybody, but north of $1,000, that is kind of tricky in a world where the Quest 3 is kind of like half the price. And I guess the whole concept of launching a VR device right now, kind of crazy,
but that's the way they roll. I'm curious about your take on it. So that is an interesting thing, and it's a very Valve move, and this is one of the reasons I like Valve. They just kind of do their thing. They do what they want to do,
and the success of Steam has actually allowed them to pursue projects like this. I would say, in a way that a lot of other companies might not be able to. But obviously, the nature of the component shortage has had an impact on this,
and the price is not super cheap. But it's also, I guess, not dissimilar to the original index, I believe, was similarly priced. Is that right, Rich? I can't remember the exact price. It was very, very expensive back in the day, but it was a huge endeavor, wasn't it?
You had those lighthouse boxes of factories and stuff like that. And yeah, I mean, this is like a far more refined device on every level.
Yeah, curious though about the whole concept of launching a machine like this. Yeah, absolutely. So what's interesting though about VR is that it's sort of found itself into its little niche
where games released on VR are largely able to actually make money, from what I understand. I've spoken with a lot of developers and publishers in the VR space now,
and the feeling I get is that despite the low audience numbers, the people that are into VR are into VR, and the game budgets are kept in check in a way that allows them to still be viable.
And I think that's what's really important here. not every game is going to make back their money, but a lot of these VR games are, they're still reasonable in terms of how much they cost to make. And because you have this dedicated audience,
they've kind of continued to be able to do okay. And Valve, of course, itself, is kind of supporting the VR ecosystem. And as they told me when I spoke to them about Half-Life Alyx on the Steam frame,
the original goal of Half-Life Alyx was just to make the most ridiculous, like zero compromise VR experience that they could. and it was not just about the success of the game, but VR as a whole.
And that's also kind of the message I got with importing Half-Life Alyx to the Steam frame, which we'll talk about, but it's like the steps in the process of doing this ended up benefiting the hardware as well in terms of developing the APIs, the drivers, and everything else.
So we're kind of in a situation where it doesn't seem like a big mainstream thing, and I don't think it is, but it might be enough. and SteamVR is a really nice ecosystem.
It's a great place for VR content. But when you think about, like, all right, well, the Quest 3 is, like, at least half the price or even less, but it's also kind of like you're deciding between do you want to be in the Steam ecosystem
or do you want to be in the meta ecosystem? You know, it's like you're kind of paying extra money to get out of the meta ecosystem, so to speak.
Right? right? Like, you don't want to have a Facebook account to play your VR games necessarily. I've done it, but I understand why people would not be too enthused about this, and it'd be better to just use a Steam account for such things, so I get it.
But I think the Steam frame is for a very specific audience, and that audience will be happy, I'm happy, with what they're trying to do here. I think there's still some compromises.
I really haven't gotten enough time to test this, obviously. I will when I get back. But they did make some choices that make it feel like just a slight upgrade from the Quest 3,
and in some ways maybe even a downgrade, but in ways that don't matter that much. Like you were just wearing the headset a second ago there, Rich, and the pass-through cameras we commented on are monochrome, but like purpley monochrome, I guess.
Yeah, that's interesting because I bought a Quest 3 as well, and my prior experience was the Quest 2. And the generational leap from the Quest 3, from the Quest 2, is astonishing.
And part of that experience is the kind of AI, AR, sorry, not AI, certainly not AI, the AR side of things, the pass-through cameras and the quality of the image. I mean, it's not amazing on the Quest 3, but it does the job, right?
And you can use it in that sense as an AR device. And, yeah, it is basically just, I'd say it's better than the Quest 2, but it's not great. It's not designed for that sort of thing. Right.
I guess that's kind of like the thing that I kind of like about the Steam frame is that it was built for a purpose. Valve really love VR. That's self-evident. They're fully committed to it.
they don't care about the size of the market. This is a device that they want to make, right? And they've made it, and I think there's certain things that they need to refine about it. It was quite interesting to sort of read the summaries and the various reviews from the outlets that I trust.
And there's a lot of commonality in there in terms of like how, you know, there's some bugs in the experience in the VR layer, that sort of thing. And, yeah, I experienced some of those as well. But, you know, ultimately, it plays VR games,
and it plays them well. they've really put a lot of effort into the concepts of having your PC as a base station for beaming out VR to a dedicated headset which, well you know John that it's actually really good
already on Quest 3, but they've kind of like over engineered this solution for the frame and kind of like one of the key success points for it which I think is you know it's kind of like remarkable that this thing exists, you know the world where Sony has basically turned its back
on PSVR too. So yeah it's a really interesting device. We'll talk about Half-Life Alyx in a minute but that just is an extension of this philosophy which is kind of like VR at any cost really. We're gonna give you VR and it's gonna be great. Maybe the market
isn't huge but here it is. I think that sort of sums it up really. Just a sort of quick impressions on the device. I mean it is obviously the battery is in the back here and the whole thing is very
very light, it's very very comfortable to wear, it's comfortable to wear even if you've got grout, although obviously the dedicated lens inserts are the way to go. You can run native games on it, I did some Streets of Rage, that was pretty cool, Streets of Rage 4, I mean those sort
of types of game will just run extremely well on Steam frame, it's a low power device, you know, throwing Cuphead, that sort of thing, super super easy to run but great game, so do you have the
kind of virtual theatre experience that you kind of expect there. In theory you should be able to do some benchmarks on that because I am quite curious to see what the Qualcomm chipset in this is actually capable of versus say the Steam Deck. I suspect it's going to be significantly less
powerful but let's wait for the benchmarks there. But yeah I just find this whole thing quite quite fascinating. Oliver, you have used the Steam thing. You were at Valve when you did it.
What's your thoughts now that the reviews are out? Yeah, quite a while ago, I suppose it was a bit of an early version, at least in terms of the software, but even then I could see that it was a really nice headset, and
judging from the reviews, it does seem like a very good headset. The thing that stuck out to me most in the pieces that I read was apparently battery life is rather poor in standalone on the other side of the computer. Streaming, it seems okay. You know, three hours in streaming,
maybe an hour, hour and a half in standalone. That's not disastrous, but I think they clearly made some choices in terms of putting the battery in the back of the headset, in terms of making the battery actually a relatively small cell
relative to the expected draw of the unit that seems to be oriented around maximizing user comfort, which I think is sensible enough, over longevity of the device in actual extended use.
I think in terms of my own feelings on it at least from my own experience with VR headsets I think it's the best VR headset I've ever used not necessarily in any particular narrow sense but in terms of offering the most comfortable
complete VR experience I could imagine out of any device with integration of Steam standalone capabilities, great streaming a really high quality experience end-to-end in every domain that it tries to tackle
it seems to do a really good job but I think I don't know it's kind of weird when you're evaluating VR devices and I've noticed this grain across a couple of reviews that I happen to observe which is that there's one class
of people who's like hey listen this is a great VR headset this does all the things I want it to do you know it's reasonable relative to the other entrants in the space that's doing a lot of really cool things and it's interesting standalone interesting streaming device
whatever whatever obviously it's primarily a streaming device but it has similar capabilities that's kind of one bucket of reviews I mean, the second bucket of reviews is like, this is Valve, this really cool company, aiming to tackle this problem again.
Can they fix all the problems with VR? And they haven't fixed all the problems with VR, right? Is that a realistic expectation for Valve, for people to have a Valve? I don't know, but in some cases, people do put a lot of emphasis on Valve resolving underlying tensions with the VR market,
in particular with its software offerings, right? the fact that premium software experiences are available in VR, and arguably, like, the last huge one that people really looked for, you know, it was, like, Half-Life Alyx, you know, some five or six years ago, and Valve obviously hasn't offered a new game to support the launch of the theme frame.
So I think those are kind of two kind of schools of thought when it comes to theme frame, and I've seen different reviewers kind of mix and match between those two perspectives. But for sure, I think by the standards of the first reviewer, perhaps, I think this is great.
But it's not into the second reviewer who wants this to really solve these underlying tensions of the VR market. I think this does some things in terms of accessibility. I think it does some things in terms of giving people that cool standalone experience with PC games for the first time.
But it doesn't actually move the needle on pricing or a game library in such a way that it radically transforms the fortunes of the market.
Yeah, I just want to pick you up on the whole battery life situation. situation. It is quite slow, there's no doubt about it. It's like one hour. I think the sort of baseline figure you can kind of rely on is sort of like an hour stand and two hours streaming One thing that I would say though is that the fact that the USB port
is actually on the rear, it makes it quite easy to plug in when you're, you know, let's say you're streaming from your PC, it's actually quite easy just to, you know, plug in the USB and continue gaming there. But obviously at that point it sort of diminishes the whole
point of the enterprise having a really good head start, doesn't it? So it's kind of like, you know, not quite the perfect solution, but you know, what would be the solution? Like a battery that's twice the size?
How's that going to work on this? It's not going to work, is it? Yeah, so, hmm, I mean, let's talk about Half-Life, Alex, because I guess, infamously, when you went to see it, Oliver, you know, the first question for many
people was, you know, okay, so where's the standalone Half-Life Alyx and it wasn't there and it's kind of like, okay you're gonna have to steam it. Now John, you actually spent some time last week
with Valve talking about the story behind Half-Life Alyx on the Steam frame and it's kind of like, how can you say it, an engineering, a technical triumph really, what they've achieved here? Yeah that's right. I understand that they worked on
it for about 18 months and during much of the project they were unsure whether it would be something they actually shipped which is why of course they didn't confirm or deny it with you
Oliver because at the time they didn't know but they were able to keep going at it and they finally got it up to a place that they felt was playable and that did require some sacrifice for sure and
the two key things that allowed it to work, and it's not the best things, but it is what it is. One, obviously, there's the frame rate situation, where essentially they run it at 36 frames per
second, which sounds very low, but this is not uncommon in standalone VR. They basically have various methods for interpolating it up to a higher frame rate. It's very common in VR.
it is kind of frame gen like right I think that's the whole idea of it but it's they use different inputs to trace them this works okay it's not perfect but it's pretty good nearly the entire library of like PSVR was like this and
it's that's just kind of how it works but it has drawbacks or sometimes things can look like different frame rates at once depending on how you parallax the environment, so it's not perfectly smooth, but I think it gets the job done. The other trick,
I think, is a bit better, though, and that's their dynamic eye-tracked foveated rendering, which I think this is going to be a game-changer for standalone VR, because your eyes really are not able to see the areas that are not being rendered at high fidelity. By tracking where
your eyes go, they're able to change the render resolution in real time, but just for that area. but everything around where you're looking that you can't really see in your peripheral vision
runs at a much, much lower pixel resolution. And really, you kind of can't tell when testing this stuff in real time. And PSVR 2 actually has this as well. That's kind of the trick here.
And it's something that Quest 3 doesn't have. And so on Quest 3 games, developers did have to push it a little bit harder to get the resolution up to get decent image quality, were here, they can really dial down the peripheral areas and just focus
on making where you're looking sharper. So you actually can get slightly better image quality than you might otherwise have. It's perceptually higher, even though around your viewpoint it's actually lower.
And those two things basically gave them the headroom that they needed to make it work. Because the rest of the game is not changed. And that was what surprised me. They didn't downgrade it or make changes to
the way it works. And their reasoning actually isn't just about the experience. It was about they wanted to retain full compatibility with the PC version, including workshop support.
So everything that works on PC must work on the standalone version, and it does. And so you basically get the exact same graphics and the visuals and the same level sizes, like no compromises to the rest of the game.
It's really just the image quality frame rate stuff, which I think is an acceptable compromise in this case, and just being able to have this on a portable machine like this is pretty cool, I must admit, especially since it's all directly tied into your PC versions.
You can be playing this on the go, and then if you want to do wireless streaming with the dongle, you can do that later. And I feel like all of this stuff is super well integrated. It just gets into, it made me think as well about the nature of what the frame is,
because it is running SteamOS, isn't it? And that means you actually do have a full-fledged PC strapped to your head. If you want to pull up a Linux desktop on your headset in VR, you can do it.
And I think that is actually, like, really cool. Because these virtual desktop stuff, it's fun, it's interesting, but usually it was either streamed from your real desktop or it was, like, a simplified kind of interface.
but here you get the actual proper desktop. It's a feat to have that, and I think that Half-Life Alyx is one of those games that I still, from talking to people, I think a lot of folks have not been able to experience,
and I think with this price point, they're not suddenly going to be opening the way towards a huge new audience, but I do think it's an important inclusion here, and I think it is something that,
if you've not played it before, you really need to, because I still think it's one of the most interesting and important games released in the last decade. Everybody should experience it. I concur with that.
I have had time to play, Alex, for a couple of hours on the Steam frame in standalone mode. I'm actually sort of renovating my office at the moment. There's like another room over there,
which is basically now just a big open area that I could use to play the game in. And yeah, you've pretty much hit the nail on the head. Basically, the big cutback, if you like, is resolution.
It does look pretty chunky, I have to say. I have to say. You know, the resolution's lost. In high contrast areas, you know, the sort of, specifically the external areas, yeah, the aliasing is pretty off-putting, it has to be said.
You know, it's a necessary compromise. Also, yes, you're right, it caps at 36 frames per second and it kind of interpolates up to... The frame itself is set to 72 hertz
and then there's sort of like an interpolation step there. I had no issue with that. I didn't really have any problems with that at all, which surprised me, bearing in mind that, you know, going back to the origins of SteamVR,
it was kind of like 90 hertz or buff, wasn't it? But yes, it's running pretty well, I'll say. The only time you can actually see that the frame rate is lower is if you actually change the controls
to kind of like a joypad-style arrangement with like one stick to look around and one stick to move. And in that scenario, when you're sort of having like heavy lateral left-right movement,
it looks like 36 frames per second. But, you know, that's sort of just a byproduct of the effects, really. There's not much you can really do about that. The visuals are really good, though. I mean, I think part of what they were aiming for here is a seamless experience.
So basically, you can play on the standalone mode on the frame, or you can play streaming, and your saves will transition across, the content is the same. And from that perspective,
it works. And I think that's kind of like something that they absolutely needed to do to maintain consistency in the kind of in the ecosystem there. I mean, you can't have two separate versions of Half-Life Alyx, where they got essentially the same headset. So yeah, that's
really really impressive and the quality of the game is similar to all of the Half-Life games from Valve. It's kind of timeless really and the fact that VR hardware has kind of like remained
virtually static, one might say, means that it just kind of still looks good, let's speak. It's still a top-tier experience.
So, yeah, I mean, I'm just mega surprised they were able to carry it off, but they have done so. And it turned out that it was kind of like work that happened in parallel with the hardware development as well. So, you know, one thing's fed to the other.
I've got some footage here and you might notice that they're kind of beholden to the limitations of capturing VR here. So you might occasionally see some really blocky parts in the image, but what that actually
is is the associated rendering working exactly the way it should. So in those points the eye is not looking at that area of the screen, so it gets deprioritized and rendered at a lower resolution.
Do you see it in the headset? A little bit, but it's virtually unnoticeable. Only if you're sort of going out looking for it and coming up with scenarios to kind of like fool the high-tracking dealers.
That's a little bit of a lower resolution. But this is Half-Life Attic working on a mobile chipset. Holy crap, this is great. One hour of factory life here, that is a bit... Yeah.
I did see Rift. People were testing a lot of standalone games that were not VR in here, and high-end games don't work well, naturally. But I did see somebody playing Cyberpunk on it, and again,
frame rate's not super high, but somehow just the idea of a game like that running on what's basically a mobile phone chipset on a VR headset at all is somehow cool to me. You just think, wow, okay, that's kind of awesome that it works.
Yeah. So, funny you should mention Cyberpunk 2077, John, because, come on, it's the benchmark game of the generation, right? And so, naturally, you've got to give it a go. I mean, we're in Hugh B. Dragon's territory, I think,
because it's a mobile chipset, and Valve is positioning this primarily as a streaming device, right? So local play, almost like an extra. Yeah, anyway, got some gameplay running here,
but frame has low video output, so it's running through remote play and comparing in-helmet results, to remote play results on the benchmark, you're losing about 12% of performance here.
So it's not entirely representative. But arguably, when the game plays as choppy as it is here, it doesn't really matter. But if people want a benchmark for comparison purposes,
720p, FSR2 balance mode, Steam Deck 3-Sec, it got me a 22.4 FPS average. And Steam Deck will be a whole lot faster than that. I actually lost frames dropping settings to medium, and I think that's because maybe you're
shifting work away from the GPU, exposing the CPU, and CPU is definitely a primary limiting factor here. You can see that just by loading up a save, it takes ages to feed in and decompress that
data. So yeah, expectation management is the order of the day really. You've got to pick your battles, and as I said, I played Streets of Rage 4 locally, That was the first game I actually played on the Steam frame.
It's one of Valve's recommended titles, and it just runs beautifully. Locked 60 frames per second. It's awesome. That's the kind of experience that you should be targeting for local play on this machine.
Higher-end games, though, I mean, just got to accept that this is not what this is built for, but it does make the achievement of getting Half-Life Alyx running locally even more impressive in my books.
So, yeah, I mean, how do we summarize all of this discussion, Oliver? I mean, it's kind of like a product I guess any Valve could produce at this point. It looks like Meta are kind of stepping back from VR, doesn't it?
And that kind of leaves us in the realm of, like, niche devices, like John's Primax Ultra or whatever. Oh, yeah.
I'm encouraged that Valve has continued to iterate in this space, I guess if anyone is going to hold the torch for the future might as well be Valve because they in the position to control the platform ecosystem They have a really compliant
device. I think it could well be them for well to come. I don't really see anyone else pushing VR in that interesting of a direction at the moment, at least. Outside of the high-end headsets, I think the market
for a high-end Pimax is a lot more limited than for a SteamTrain. I think the SteamTrain is much more compelling. you know, mass market would probably be pushing it too far, but certainly a much more compelling mainstream-ish device for VR.
But yeah, I'm not really seeing anything with this particular effort that makes me think it's going to make VR a truly mass market proposition,
but I am seeing a lot of things that make this seem like a really cool VR headset. So if you're already interested in VR, I think this is a pretty safe bet if you're fine with some of the battery life limitations, especially as I think, like we saw with Steam Deck, hopefully we'll see the same phenomenon with Steam Frame, where we see actually a lot of development and a lot of compatibility on the software side with the device, right?
because I know at the moment there are something like 120 games that are whitelisted for Steam Frame to run sound on the device. I expect that number to creep up and up and up, you know, and for a lot of games to be pretty accessible on the device in the future.
Right, so let's move on to some actual questions. We're going to kick off with this one from CheaperOne. Hey, y'all! Exclamation point. How many technical problems are due to feedback not piercing the language barrier? For instance, farm software games often have poor frame pacing.
I think it's not aware. Do you think it's a poke from a Japanese source? Thanks for all you guys do. Interesting question, right, John, but ultimately I think it's fair to say
that the Japanese are fully aware of everything that's being said about their games. The form software situation, though, I think it's basically to the point where I have spoken to platform holders
that deal with various third parties and they have dealt with font software and it's you know basically it's they do their own thing the technical side of things it's not really their focus it is what it is this stuff is highlighted
to them it's down to the developer to fix it um yeah it is a bit of an egregious problem but i'd say that the frame pacing thing you know it's not a wholly from software type thing it's just had the blood of Dawn Walker come out with very poor 30 frames
to second. So yeah, I'm not sure how much the language barrier is an issue here, particularly in a world where development is a global enterprise at the moment and
even the likes of Capcom etc. have a lot of Western staff on board as well as I understand it. So yeah, not really an issue. Yeah, this is a tricky one and I don't think, so the frame
pacing issue we've always talked about is due to capping the frame rate below the Vsync or the refresh rate. So you're gonna see it if you're at 30 FPS in a 60
Hertz container or 60 Hertz 60 FPS in a 120 Hertz container if the game doesn't perform it correctly. And because 60 FPS has become so more prevalent this generation, it feels like we haven't seen it as often, but it does still creep up.
it's not just from software games. Obviously, DragonSong went to when it first came out from Capcom. It had a 30 FPS cap. Same problem. Blood of the Don... Wait. The Blood of Don... Oh my gosh, that's a stupid name. The Blood of Don Walker. I think
that's right, Rich. Okay, The Blood of Don Walker has it as well. So I've wondered about this because I've worked with a number of developers now, sort of on the side
for various things, and I've encountered these type of issues before, and I've sort of posed the question, like, do you guys see this? And what I found is that when it ships this way,
a lot of people do not see it. It is a subjective problem, and that a lot of people's eyes just don't seem to pick up on this issue at all.
And developers are used to working in scenarios where, one, the game runs kind of bad often throughout development, especially in debug tools and, you know, when you're actually developing it. Two, the tools often report results that are good.
You're getting the correct or smooth enough frame times and frame rates and everything seems to be good on the performance side. three, a lot of developers I found actually use remote workstation setups where they're
basically connecting to dev kits remotely and that stuff also introduces its own problems and other things like that. Basically there's a lot of things that prevent people from seeing the raw footage of the game and you say
well just play it normally and I think even when people do it's like a weird blindness for some folk. Like they just subjectively are not able to detect the problem, which is why not everyone picks up on it.
And I know that seems silly for the people making it, but I think it really is just that simple. If the tools are reporting results that are good, but your eyes aren't able to detect the problem,
you don't solve it. And I almost guarantee that's what happens in FromSoftware. Whoever's involved in this doesn't see the problem. They're not able to physically see it with their eyes and they don't understand the complaint is my guess.
I don't know for sure. Maybe there's other things going on in their engine that make this tricky or annoying or something. But I really think it's just something as simple as that. It's just that they're missing it.
Well, it's kind of become a moot point because they've come up with another solution, which is just to run unlocked. Yeah, most of the time it's true. And that also sucks. But it probably sucks less.
especially if you have VRR not that I think you should lean on it as a crutch but yeah some software and some others they really just don't have a clue when it comes to this stuff frame health
if you will so I really appreciate when developers do take that stuff seriously and a lot of them do take it very seriously indeed so yeah I mean famously we do have a video
up on the channel that shows that with like a very very simple patch to the game, to any form software game. You can run it at a properly frame-based 30 frames per second. The drawback being it's using like a
Sony library to do the job and it does incur extra latency presumably through extra buffering. And so the latency is higher. But, you know, again, thinking of going back
to the beta of Destiny 1, that had inconsistent frame-pacing and they fixed it, but they did make a point of saying that it wasn't easy to do that while
maintaining low latency. That might be another part of the issue there. I suspect it's even a combo there where people are largely not detecting or feeling the issue is significant
but they do see the numbers about latency and they're focused more on that. I understand the idea of keeping the latency down. That is important but I think you shouldn't do so at the expense of frame consistency, and I think both are actually possible.
Any thoughts on that? Yeah, I think in some cases there might be an issue where there's split priorities, like John and you indicated, where if you're getting lower latency with a broken FPS cap,
higher latency with a fixed one, maybe go with a broken FPS cap, and games, for most of their development, kind of run this drop. You know, they don't so maybe you aren't picking up on it as much if the game is already not hitting its frame rate target for the
Majority of development. Maybe you're just used to it. Then when you actually do hit frame rate target, maybe you're not seeing it as much I'd also point out that it's something that you know to discuss a language barrier problem I don't really think it's a language barrier problem because within different like Japanese teams even in studios like Capcom
You see different approaches being taken right depending on the team in some cases they have a broken frame rate cap. In some cases, they have no frame rate cap. In some cases, they have a correct frame rate cap, like with Onimusha on Switch 2, I believe.
Monster Hunter Wilds on Capcom also has a correctly frame-based frame rate cap. But conversely, there are titles like Dragon Slugma 2 and a million other titles that are not so well-behaved in various ways, right, that we have documented over the years.
So I think it's rooted primarily in different perceptions of the issue within given studios, perhaps between different teams or between different studios, etc., etc. But I have noticed it occurring a little bit more in some Japanese titles.
In particular, recently with some Japanese ports to switch to, but I don't really think it stems from language barrier. I just think it's something that comes from different priorities within different teams, and if you don't see it or you prioritize lower input latency,
you can be led down a chain of decisions that I don't really agree with, but ultimately, you know, that produces an image that, you know, kind of like John said, some people just don't see it. And I wish I was one of those people, but I'm not.
Okay. Let's move on. Next question here from Oksalima. Hello, DF Labs! Besides things like neural rendering and all that, what you quote-unquote strayed of DLSS would you be interested in?
Is it some sort of virus at this point? I would love to see performance of the newer models improve, but besides that, I'm not sure where else the tech could. What would be after frame gen and upscaling? Any thoughts?
So, yeah, obviously, NVIDIA has gone in one direction with a kind of generative lighting path with DLSS 5. And, yeah, I'm not entirely sure that it should be under the DLSS umbrella as such.
So Oliver, any thoughts on where DLSS, in terms of its kind of, I guess, the original AI-powered performing enhancing side of things should go?
Yeah, I would like to see some kind of latency reduction technology, which I know is under Reflex, but perhaps that could fall under DLSS. And I'd love to see more progress on the Reflex time warp concept they had.
I think Alex and I saw that at CF like two years ago, but it never actually shipped in any game to my understanding. It seems to have been dropped. NVIDIA is not talking about it too much. I don't know exactly what's the name of that.
That technology, especially if it could be made to work with frame generation, which I think the technology that we saw was not able to work with frame generation, could make games more responsive because I think as we try to decouple the update of the game
of any given rendered frame from the actual frames that the player is seeing, I think a technology like this is probably interesting, if not important. I think it would be cool to see it, and that would be a really cool feature to have under DLSS
or some kind of video technology. But on the other side of things, on the neural rendering side of things, I think you could push a technique, like not DLSS 5 necessarily,
DLSS 5 is trying to do lighting, but I think you could kind of do a neural restyling filter, Maybe that would be an interesting area to go with DLSS, kind of like a filter that translates one game's tile into another game's tile, so you could have, like, a neural restyling filter that implements cell shading into a realistic game, or a neural restyling filter that maybe gives the game, like, an animated painting look, right?
that I know some games have achieved in various forms using, you know, various art tweaks, asset tweaks, you know, various shingers, whatever, but you could actually achieve something that feels maybe a little bit more natural, a little bit more sophisticated, a little bit more emergent in that sense.
You know, I think you could push DLSS in a lot of interesting directions, basically. But I think the most proximate thing would just be something to tackle latency, but not actual latency, perceived latency.
I think that would be the way to go. I think that would be an interesting point. Right. Yeah. Yeah. From my perspective, it is about basically the thing, the quote-unquote thing that is holding back frame generation
from being everything it could be is essentially the fact that you're always going to need to buffer a frame. There's always going to be a frame that you're going to need to render in advance and then generate the interpolated sort of intermediate frames there.
That the thing that kind of like the major downside to DLSS frame generation which is to say that you know inevitably you got to add latency because of that Now the extent to which
that latency is important I think in some scenarios is overblown in a world where every game has a different level of latency and nobody sort of points fingers at the game running at 60fps saying this is running 30 milliseconds
slower to respond from this other game running at 60 frames per second but at the same time that is kind of like the drawback of range generation and i don't know how they're going to fix that in terms of like you know unless there is some sort of um uh reprojection into the future so to speak um
which would be super super tricky to do but with that said you know everything dlso5 is doing is super super tricky to do on a technical perspective i don't know where where else you can push dlso5
in terms of it being a performance-enhancing technology, which is kind of like what I associate with the brand, you know, deep learning, super sampling. Super sampling, I guess, can go either on sort of temporal
or spatial resolution, but I don't think it applies to AI sources. So I think maybe there is some sort of future in folding in reflex into DLSS because, you know, reflex is inherently a performance-enhancing technology.
Tricky one for sure. I think the other thing which is quite interesting is on the upscaling side of things DLSS 4.5 is really really good right but at the same time FFR4 is really good. At the same time I think the 4.0 transformer
model is really good with DLSS. So we've actually reached the point now where I would say for a lot of people the upscaling side of things is kind of good
enough and there's only innovative improvements to come and maybe there is an argument to be said for focusing on other things and I think the recent release of the 4.5
lady construction DLSS was pretty much what we wanted John, DLSS Oh man what kind of silly things can we do here
how about DLSS news motion and by that I mean you have the poll for monitor doing that to try to clean up motion blur in moving images. What if
you could do this on any high frame rate monitor? We can kind of do that already but if they found a way to implement this at the game level so like everything just like cleans up motion, uses a high refresh rate to do some sort
of backlight, not even backlight like black frame insertion or something but controlled by the GPU in a different way I don't know that'd be fun. We could do what about DLSS VR?
I know VR is not big, but what if every game you can just magically render out two frames and like appear so basically like bringing 3D vision back with doing it so like it's like a hassle free process.
I mean a lot of this stuff already exists in mods and other programs. The idea of being able to just like put on your VR headset and have the game be 3D would be awesome. Or DLSS sound.
I don't even know what that would be. I don't think I would know. I told you we'd know. You just said it. I think one of their serious answers would be, I think Brian Catanzaro said this to me and Alex a couple years ago maybe,
where he basically said, well, what if you had, and this is not a new concept, but what if you just had an automatic frame gem up to the refresh rate of your monitor from like an arbitrary input, kind of like we see with LFSG at the moment, I believe,
where every frame is going to be a generated frame. That would also be really cool if you had some super fast hardware to do that and just give you like a thousand FPS from your 60 FPS input. I don't know how good that would look,
but that is one way for DLSSFG to evolve beyond like the fixed integer. I guess it's not a fixed multiplier anymore, but it is kind of a multiplier-based approach,
integer multiplier-based approach to frame generation. Maybe you could achieve something new in that field as well. That would be kind of interesting. There is something else that we could discuss here, and it is basically the whole concept.
I mean, obviously there has been the concept of DLSS5 working on an AMD GPU, the extent to which it's actually properly working would require proper analysis, but, you know, there has been a port.
Secondly, the kind of moat that NVIDIA has developed around certain DLSS features seems to be disappearing. I mean, there are now mods that are able to run multi-frame generation on 20, 30 and 40 series.
This is really interesting from my perspective because it seems to me the case that there's a lot of reverse engineering going on that is expanding the remit of DLSS to older hardware in a world where people are basically being forced
through the AI hardware crisis to keep their old hardware, right? necessity versus invention. It's pretty amazing to see all of that stuff. And it's reached the point now
where the BC250 mining card, which is based on basically active PlayStation 5 silicon, it's got a working version of FSR4.
It doesn't have in-case acceleration, that particular APU, but basically it's been reverse engineered and made to work with reasonable frame times on the BC250 running Linux,
which is astonishing. Now, in theory, and maybe I should try and find a few moments to test it, the DLL that makes that possible should, in theory, allow FSR4 to run on RDNA1.
So, yeah, there's kind of like the progression of features that we might hope to see NVIDIA take DLSS. And then there's basically this entire movement that's working to take these AI features and make them run on as much hardware as possible, which is remarkable.
Anyway, let's move on. Okay, final question of the show. This one comes from Skyrim Fest 26 FPS Edition. And the question is this. Now technology has advanced to the point where you can play alongside an Ikea Kallax shelving unit in Skyrim.
what budget Swedish furniture would you like to see in Elder Scrolls 6 my money is on 4X DLSS 5 rendered lag tables John this has been I think
probably one of the most amusing and delightful brand collaborations we've seen in recent times two iconic franchises coming together what do you think?
it's cute this kind of stuff is fun I like to see it you know, obviously these games are often about storing items and collecting tons of stuff.
So, it's just a fun joke, what can I say? And I'm happy to see that, given where things are at these days, it's fun to see a little bit of levity in your updates.
So, it just reminds me how back in the Morrowind days, My younger sister was so into this I think I've mentioned this before But she played for like thousands of hours
And Morrowind But used to like I remember She took over Balmora In that game And killed everyone in it Or pretty much everyone in it And like basically took over the houses And made each house like
A storage for a different type of item Like you had the sword house You had like the shield house Stuff like that And she actually just slayed them all out Which was a really fun thing So that's the first thing I thought of when I saw this kind of like little joke here
was like oh yeah they're just kind of referencing the old storage aspect of these Elder Scrolls games and it is funny having a little calyx following you around because I feel like the calyx itself is rather iconic at this point. Most people have at least a calyx
somewhere. We certainly do. Yeah. I wasn't going to say I do but I actually have a predecessor expedite I think it was called Justice for Expedite why did they replace it? it's a retro version yeah I've got the retro version of the Kalex
well let's talk IKEA furniture obviously the Strandmon wing chair in Nordvala dark grey I think is you know it could replace some phones
in Skyrim that would be pretty good I think but beyond that I'm just enjoying this particular brand collaboration I thought it was really fun and kind of stuck the landing.
Oliver, any thoughts on IKEA and, sorry, IKEA, to give it the correct pronunciation, and game collaborations? Well, I have an IKEA and game collaboration right behind me
because that is an IKEA Billy bookshelf, I think. Oh, right. All my old consoles and such, I bought that when I moved into this apartment because I thought it looked reasonable and seemed sturdy enough,
and I only had so much money, so that's what I bought. And it seems to have worked out okay. Seems to have worked out okay. But in terms of this question, I mean, four times DLSS 5, I don't get another bed for anything less than 10 times these days.
Come on. Those are rookie numbers. You've got to point that out. Yeah, four times is pretty intense, but 10 times is where it's at. I'm sure people agree.
I've got to admit, I'm getting bored of those mod videos on social media now. It's becoming right-click not interested or right-click mute. Because as time progresses, they're just looking worse and worse.
I saw an Alan Wake 2 one the other day, which the author thought was really, really good. I mean, it basically just makes Saga look about 50 years older. I mean, what's the point of that?
The whole thing is, you know, let's see some actual game integrations at this point. I'm done with the DLSS 5 mods at this point. They're just looking more and more hideous at each passing moment.
So anything more you want to add to this particular discussion there, John? Well, unrelated, but while you were finishing up, I just discovered something, Rich. Do you know what I'm playing right now as we record? Well.
Phantasy Star Online in a browser. Wow. Someone ported Fantasy Star Online, the Dreamcast game, to a web browser. It's running right now at a very high frame rate inside Chrome.
I'm looking at it as we play, and I'm just, like, amazed. Like, this is the whole game that's here, online and offline. Like, what the heck? Like, I just thought that was really cool. Like, you can just sit here and play.
It's psoreborn.com. You just go there, and you got PSO running in your browser. Oh, man. It was back many, many happy memories. That's pretty cool, man.
I was just remembering that back in the day when I was playing it, there was a particular supplier that was, I think it might have been British Telecom, that basically gave you a reasonable price for two hours online.
And so you had two hours online, and then you reached the two-hour point, and then you had to go offline and then reconnect again to continue. Oh, my gosh. But that, man, that was an amazing game, wasn't it?
Sure. I was doing the same thing over and over again and never got old. Yeah, I'll have to take some footage of this and tell you right now. But I can't. Cool, right? It's not going to last for very long, but it's very cool.
This is outstanding. Yeah. Amazing. Okay, well, that is the end of the show. So please do like, subscribe, share all of that stuff, ring bells and whatnot. DS Supporter Program, of course, patreon.com slash digitalfoundry, early access to DS Tech Weekly,
high-quality video downloads, ad-free of everything we do and have done since pretty much late 2016. We're going to be coming up to the 10-year anniversary of our video download soon. And as the access to our website, digitalfoundry.net.
And yes, if you want to read everything that we talked about Half-Life Alyx and indeed the scene frame, that's going to be up there right now. and lots of other juicy content as well.
But that's all from us on this one. We'll see you soon in the next Direct. And I guess, John, we'll be seeing you in a couple of weeks. Yep, sounds good. Okay, see you soon.