An appraisal from PSVR1’s Immersive Experience Specialist
Part 2 | On-boarding and the general UX
As someone who worked closely on the development of the original PlayStation®VR as Sony's resident Immersive Experience Specialist, Jed Ashforth has been very excited in the build up to PSVR2’s launch. Here are his thoughts about the Sony's new successor.
Last time in Part One, I detailed some of my thoughts on the PSVR2's hardware in general, including the pricing, the fit, and my admiration for the new Sense controllers (despite some niggles). This time around, we're looking at the onboarding process, and the general user experience (including some of the setup, messaging, optics, haptics, eye tracking, and the foveated rendering). As with last time, this isn't an exhaustive breakdown, but these are the things I'm excited to talk about from my perspective, and I welcome any thoughts, clarifications or discussion in the comments below.
Onboarding
Turning it on is a buzz: One of my favorite things about the PSVR2 is silly, but it’s this little head vibration it gives off when you power it on or off. It feels nice, and is somehow evocative of the PS3’s lovely orchestral tune-up that happened when you powered it on. It’s daft but I look forward to it at the start and end of every session, it feels simultaneously classy and kind of soothing, and reminds you of it as a unique, signature feature of PSVR2. That’s a good idea because developers are using it sparingly for now (big crashes and things whistling past your head mainly) so it could be easy to forget it even existed otherwise.
It’s all pretty seamless and streamlined: This is the most truly 'plug and play' VR experience I've seen, helped by the fact that all your necessary user and account setup was done when you first set up your PS5. It genuinely only takes a few minutes to set everything up as a first time user, and when guest users come along their setup is minimized to the bare essentials, and even those can be skipped and they can just use the device as a guest without any tailoring. Individual IPD (eye distance) calibration is gently encouraged, and straightforward. New users only need to line up their eyes in the headset (see earlier) and then perform eye tracking calibration as a second step. This calibration is streamlined (around 45s) but a little clunky visually; It’s watching a dot move about as you might expect, and while quick it’s still a dry little friction bump for new users, and no particular effort has been made it hide the process behind something fun. It seems a missed opportunity to not gamify this process and build up anticipation and deliver some VR wow, or at least give some rewarding feedback to the user. This is the first contact point for excited users remember. There’s a quick test at the end that lets you play musical notes by looking at them which is neat, but again it feels bare-bones and way less fun than it might have been. One can only imagine the Nintendo version of this process would have had buckets more charm. But it works fine for now, and it may well be updated and embellished into something more appealing over time.
Setting up your play area: This is kinda slick and snazzy and feels nicer to adjust than other boundary systems. It auto scans and proposes a workable space for you, which you can then adjust. It’s only a small step forward over how the other folks are doing it but it feels classy. It also seems much more reliable at recognizing and maintaining your play area than Quest2 currently, and sharing it with other accounts on the headset means you don’t have to guide new users through the process - you can just set up a guest account and let them use your boundaries. I haven’t yet come across any sensitivity settings for how close you trigger the boundaries, but I haven’t had to go looking as they seem to be judged well for the use cases I’ve encountered in-game so far. We’ll see how it copes with the fast aggressive movements of famous TV wreckers like Gorn and Drunkn Bar Fight in due time, I’m sure!
Unhelpful controller diagrams: This is another side effect of those surrealistic spherical controllers, but one of my emerging niggles is that there doesn’t seem to be standard depictions of the controller used by developers, and this can be endlessly confusing because every diagram looks like a space ball with a highlighted button somewhere in the middle of it. The orientation of the controller in those diagrams seems to vary with every title, some showing from the side, some from top-down, some from the hole your wrist sits in. Black buttons on a black background again. There’s particular confusion around the trigger buttons, and being able to tell the left controller from the right controller if you can’t identify them from the button names. Normally if you have to show a PlayStation controller on a tutorial pop-up or a controls diagram, there are very strict rules enforced by the platform to maintain consistency from title to title across the broader library. I think the diagrams for PSVR2 must have been late sharing out to developers maybe, as the current approach seems unregulated. Lots of the launch titles seem to use tiny button icons on their diagrams and instructions that will be a little too small for some to read clearly in the headset, and you can’t always lean in to get a better view. Developer-side thing mainly, this, but I’d expect standards to emerge quickly here to do away with these teething troubles.
User Experience
Pass-through: The pass-through camera is B&W, but it’s much higher fidelity than you might be expecting if you’ve only tried Quest2’s pass through. It passed an important test for me in that I could actually (just!) read a message on my phone through the pass through mode, which feels like an obvious bar for the tech to pass. It’s also pretty stable, none of the weird warping and stitching that happens around the edges of Meta’s version.
There’s a big, recessed button on the bottom right of the headset that puts you into pass-through mode and automatically suspends your game at any time, which is useful but I’ve got to say a week in I still have to take a moment or two to find the button with my thumb. Quest2’s Scott Summers-style visor tap is more immediate, I think, but it’s also far more prone to being accidentally triggered every time you tried to adjust the unit on your face. Feels like the button is always further in towards your nose than you expected. As a system function I’m not sure if it can be reassigned to a facia button on the controller right now, but I’d be happy to be able to swap places with the social share button which I use much less.
Tracking: On the whole the inside-out tracking is stable and reliable most of the time. The position of the tracking cameras combined with the layout of the IR lights under the controller’s surface seems to be well judged. There’s always times in VR where your hands are hanging by your sides and not in view of the headset cameras, and PSVR2 handles this fairly well most of the time. It doesn’t hurt of course that Sony have at least a decade of experience with tracking prediction for spatial controllers to draw upon, so their off-camera tracking feels reliable and robust – to a point…
The controllers profess to have 4+ hours on a single charge, but I have noticed that once they’re down to about 30% or lower, then drift and positional aberrations start to make themselves felt in the controller tracking. I’m assuming this is when some kind of power-saving mode kicks in and lowers the sample fidelity somehow; they start to feel a little more sticky and less reliable, especially when going out of camera view. Knowing how much post-launch improvement the engineering teams were able apply to PSVR’s vision-based tracking, I’d expect this to get some tuning attention in due course.
Eye Tracking: This seems really great so far. There’s been quite a few places where it’s felt like some kind of magic from the future (targeting enemies in Rez with your eyes is just fantastic, and feels like a super power has been added to a game I know so well). And then there’s been a few others where it’s applied to a game menu that can also use stick selections at the same time, and the two start working against each other and it takes a moment to realise it’s happening. Normally for a UI that can be accessed through disparate input systems, a game will see what elective input the user makes first, and assume their preference from that, and prioritize two conflicting inputs accordingly. But eye-tracking is essentially non-elective: you can’t assume the user is looking at something to select it, they’re just as likely to be just looking at it, reading the text, admiring the drop shadow on your button or whatever. They might be moving the stick around absentmindedly or because you’ve made it emit a nice noise or haptic effect when you roll over option panels. So it’s a bit of a thorny tangle when you’re supporting two disparate input methods simultaneously, and it’s just the sort of puzzle I love engaging with, so I’m keen to see how this all shakes out from a UX point of view. I think in the short term devs probably need to ensure there’s messaging or signalling to the user that eye-tracked menus are active, especially while everyone is getting used to the tech being there, because it's easy to forget it’s a thing with PSVR2 at the moment. Not every title uses it as standard for menu selections, and some ports of existing titles (the vast majority of the launch line-up – more on that next time) won’t have had the foresight, time or budget to even spot this could be a pothole for users when adding eye-tracked selections.
Foveated Rendering: – For those new to this tech, foveated rendering is a technique that allows the game to render at much higher resolution wherever you're looking, lowering the resolution elsewhere to compensate. It’s not easy to see from within the headset – the very point is that if it’s working correctly, you won’t notice the delay before you’re seeing the up-rezzed area of the screen being rendered where you’re now looking. That latency has always been the Achilles heel of foveated rendering of course. The power boost it can give is sure appealing – rendering detail only where the user is looking means what they see can be made to look super detailed and fancy at all times, without making the PS5 fall over in a wobbling heap. That’s a big win of course. Sony engineers have described a potential 3.6x boost in GPU frame times from the technique, so that unlocks a lot of rendering power, but seeing the lower resolution rendering for even a split second before it gets up-rezzed will give users that same feeling of ‘geometry-popping’ that’s familiar from videogames and Google Earth, and that can just destroy any chance of immersion because it’s such a regular ongoing effect the user keeps seeing every time they move their eyes. So it’s a grand prize, but it’s not worth shooting for at all if you can’t get the latency low enough to make it feel seamless.
The best use cases have always been dynamically applying it where the user needs to see more detail, like reading text or looking at close detail or the distant horizon – stuff where the eye movement is stable and predictable. That predictability is essential to keeping latency down, so you can start rendering the high detail before the user gets there. But going where the puck is going to be is hard with eye tracking, because of how fast our eyes saccade from one point of interest to another. It’s a very quick flick, and can take less time than it takes for the PS5 to render a single frame, depending on the distance the eye has to traverse across your field of view.
It’s very early days on this, but already we can see a number of different applications of the technology across the titles I’ve played, we'll take a quick look at two of the biggest titles which represent two different approaches taken on the development side, and using it to different effect.
Horizon: Call of the Mountain takes a fairly gently approach. Horizon’s detail drop outside of the hi-resolution focus areas seems minimal, there’s not that much difference between the dynamic areas inside and outside the fovea. In a way, it feels somewhat extravagant knowing that even the bits you can’t look at are still looking great. That’s a big win for the social screen (TV) view and streaming, as it presents a quality experience all round, and is no doubt helping people to be wowed by how good it looks on videos and streams. The vistas are expansive, stunning and so often in view that the engine (Unreal 4, running in 60hz reprojection mode), seems to have understandably been tailored more towards the epic spectacle, sometimes compromising on scenery detail up-close as a trade-off. There's debate that more aggressive foveated rendering in the title might have been horse-traded for better looking oranges in the marketplace, but I'd imagine the sheer quantity of textures and details here from some of the viewpoints must be a significant constraint.
The foveated rendering is mostly undetectable in practice, sometimes there's minor detail pop-in as you look around as well which I found more noticeable, perhaps another sign of managing so many textures at any given time. The foveated switch-over does seem noticeable with pop-up tutorial panels which can seem a little murky when they first appear, but during action and movement all of these things melt away.
Having said that I’m only a couple hours in and most of that has been standing around admiring the scenery and exploring the implementation. We'll see how it holds up as I get deeper into it. Firesprite have been making excellent VR experiences for a long time now, and here they have made a very inspiring showpiece for the capabilities and magic of the new hardware seemingly without leaning too heavily on the foveated rendering to do it. It’s super-impressive to me as an overall mix despite the visual compromises in various areas. I think a bunch of difficult choices come with a scope this large, and the choices the team have made here gel into a very impressive overall showing. It feels very epic, and very on-brand for PlayStation. If Sony can work this magic here, it's maybe the pudding of proof that AAA titles like God of War, The Last of Us or Uncharted could be perfectly feasible in the future, and that's certainly going to be a big win for VR adoption.
Gran Turismo 7 uses foveated rendering very differently. There’s a distinct resolution difference that visibly scales between the area where the user is looking and the falloff elsewhere, and it can be quite visible on the streamed/social output. I worked on racing games for nearly a decade, and there’s a lot of science and understanding around where the user will be looking at any given time during a race. Traditionally we use this to work out the best position to array all the necessary data around the player’s point of concentration, so they don’t crash while they’re looking over at their map or up at their lap time for example. We’ve always understood that it’s a largely horizontal, smooth-pursuit movement of your focal area that snakes along with the track, and is fairly easy to predict in general, interrupted only by quick saccades of your eyes to check your instruments. In a real car cockpit, all of the vital information is arranged conveniently on the dash for easy access anyway, so in a lot of ways VR racing games seem like an ideal fit for more aggressive foveated rendering, and an easier case than most for developers to understand where the detail focus is likely to be needed at any given time.
Again, even though the difference can be visible on the social stream, in-headset none of this is apparent at all to the user. The cockpit always looks sharp, solid and very readable (and genuinely proves to have much more utility to the player than the flat cockpit instrumentation you see in non-VR mode) but I have to say the distant vanishing point still tends to feel a little on the blurry side. It feels surprising and noticeable to me at the start of the first race in each session I play, but once I’m immersed in the action I just don’t notice it, it’s certainly not a deal breaker in terms of quality
Head haptics : Another great addition to the hardware. We actually investigated this for the original PSVR, largely to see if the vibration could provide vestibular noise that would alleviate any discomfort symptoms. We were curious to know if some vestibular feedback, even the headset vibrating, would be better than just having an absence of the expected feedback you’d get from, for example, driving a VR car around a corner. Would it help soften the mismatch between our expectations from reality and what the VR tech could deliver? Our tests were non-conclusive at the time, and caused a few headaches, but there were a lot of other factors we didn’t know about that we do now, so it’s really interesting to me that they’ve included this, and I’m curious how it gets used in the longer term. At the moment it’s being used sparingly, to enhance the impression of a bullet whizzing past your head or a dinosaur stomping nearby, for example, and I think that’s a wise approach to take; it keeps it feeling special and surprising, and it mitigates against any potential risks that might arise from vibrating the user’s head too much. And as I mentioned earlier, the startup and close-down ‘buzz’ is such a nice reminder of how cool and unique a feature it is. It’s a lovely combo to close out with – user confidence because they can feel the headset has shut off, even while they’re part way through removing it or putting it on the shelf, and their last contact with it leaves a good impression, reminding them of a signature ‘magical’ feature. Such a neat little touch and I loved it. 😊
VR reset and orientation: Sony kept the same legacy reset control as on PSVR, but made sure the button was now easier to locate and press on the new Sense controllers, which is smart thinking. I also thought it was smart that the pass-through on your TV also acts as a really solid tracking anchor for the headset cameras, if you care to leave it on while you play.
Bright lights in the room do seem to occasionally scramble the tracking when you're (unknowingly) aiming straight at them from inside VR. I had to angle a couple of lamps away to fix some early issues with aiming that I thought were game-side rather than system-side. The Quest 2 has a similar wrinkle of course but I've never had to troubleshoot the lamps into a different direction. So it's a simple fix to try if you encounter tracking issues.
However, I have noticed the age-old specter of orientation assumption in a few titles, which is kind of crazy when you’re no longer dealing with outside-in tracking where facing your sensors is important. Where a game or experience assumes that your forward orientation is relative to the play-space, rather than what you want to be ‘forward’ when you reset your VR view, you can end up with weird side-effects. I’ve played a couple of titles where loading screens seemed to be absent, only to find they were rendering behind me because I had my back to the TV. It’s surprising as well because one of the titles that practices this, The Light Brigade, led development on Quest2, and this orientation assumption should just not be a thing for inside-out tracking like Quest2 and PSVR2 at all, it’s usually a holdover from titles that came over from desktop PC VR. Very curious, my only thought is that it could be linked to how they’re making use of the user’s available play-space, which might benefit from knowing a definitive forward vector. You could imagine this would make sense if they wanted to ensure you have maximum room to dodge left and right, for example. I guess doing this during the loading screen is an opportunity to return errant users back to their center point. I missed a bunch of the story as a result, so I have no idea why ‘Dark Souls in WW1’ is happening to me, but ‘just because its cool’ seems to be reason enough for me, and I’ve been having fun with it.
Final Assessment
Well, a 'final' assessment is impossible for PSVR2. The software underlying the experiences will continue to be updated, the tracking quality will continue to be refined by Sony's software engineers, and the user-facing aspects of the onboarding and systems operation will no doubt be revised a few times in the months and years to come, just as is the case with any modern hardware launch. Aside from a couple of niggles and wrinkles, though, the experience out of the box is hugely compelling and impressive, even for a serious PC VR-head like myself. There's so many improvements over PSVR1 that the migrant fanbase from that platform will feel like they've made a more-than-generational leap in the capabilities of the device, the controllers, and the overall user experience it offers. Quest 2 owners will be similarly impressed no doubt - even the best mobile XR-2 experiences are going to look visually weak pitched against PS5's heavyweight Zen-2 processor and custom RDNA2 graphics architecture, of course, but let's not forget that the 'HDR' OLED screen also delivers a visual experience unlike any other headset, and it can make a startling impression. All tuned-in and on-your-head, this is some of the best VR I've ever experienced, hands-down. The Sony quality can really be felt, and the expectation of future releases to the quality of Horizon Call of the Mountain, Resident Evil Village, and Gran Turismo 7 teases an irresistible future for the device. But only for gamers. PlayStation is, of course, for the players.
As an all-purpose VR headset, it obviously falls short of being essential because it only lives within the PS5 ecosystem; for now (and the foreseeable future) we won't be using this for anything more than playing games on PS5, as amazing as it would surely be for creativity and productivity purposes. There's a dream in my pipe that Sony will continue to be PC-friendly and eventually unlock the system as PC-compatible, but that would mean opening up their walled garden to let Steam come in and play with their customers, and I can't see any business sense behind them doing that, for now or in the longer term. It feels like a shame. But that trade-off means PSVR2 can put all it's effort in offering a best-in-class experience for that singular, but predominant use-case - playing amazingly immersive VR games. On that front, it feels like it won't be beaten for a good while, and the recent teases around a PS5 Pro in the next couple of years will see a viable upgrade path in the future that offers even more processing grunt, much like PS4 pro did for the original PSVR. I'm excited to see what comes next, both in terms of future software announcements, and in terms of revisions and updates to the current user experience the device offers. But mostly at the moment I'm still just excited to spend more time with it, which for someone who spends an awful lot of time with their face stuck in various VR boxes is honestly the highest of praise.