Rift S is very cool! It takes concepts that have been around for years and puts them into a fully functional product for the first time. Sure, sure, I see people complaining about how Rift S is worse than CV1 concerning audio quality, display characteristics, and ergonomics – some of the tradeoffs are real, some are imaginary, and people should really wait for it to come out before passing final judgement. All in all, it is going to be a great HMD.
For about 70% of the population.
My IPD (interpupillary distance, the distance between my eyes) is a hair under 70mm and slightly skewed to the right side of my face. One of my best friends has an IPD of 59mm. I don’t know what your IPD is, but both of us were perfectly served by the IPD adjustment mechanism on Rift CV1, a mechanism that was an important part of our goal to be compatible with male and female users from 5th to 95th percentile. Anyone within the supported range (about 58mm to 72mm) got a perfect optical experience – field curvature on the focal plane was matched, geometric distortion was properly corrected, world scale was at the right size, and pupil swim was more or less even. Sharp imagery from edge to edge of your field of view was the norm. The small handful of people with an IPD outside that range would not get a perfect experience, but at least they would be in the right ballpark. IPD skews in different directions by gender, race, and age, but we managed to cover almost everyone, and we were proud of that.
This is not the case with Rift S. Like Oculus Go, it uses two lenses that are set about 64mm apart, perfect for a perfectly average person. Everyone who fits Cinderella’s shoe will get a perfect experience, anyone close will deal with minor eyestrain problems that impact their perception of VR at a mostly subconscious level. Everyone else is screwed, including me. Imagery is hard to fuse, details are blurry, distortion is wrong, mismatched pupil swim screws up VOR, and everything is at the wrong scale. “Software IPD adjustment” can solve that last bit, but not much else – it adjusts a single variable that happens to be related to IPD, but is not comparable in any way to an actual IPD adjustment mechanism . This is the main reason I cannot use my Oculus Go, even after heavy modification on other fronts.
Does that mean Rift S should have had mechanical IPD adjustment like CV1?
No. That is the easiest way to look at things, but it misses the point by focusing on a particular solution to a broad problem. HMD design is all about tradeoffs, and finding the right balance is very difficult. More important than IPD adjustment is IPD Tolerance, an all-encompassing set of factors that determines your ability to use a particular VR headset both at rest and during active motion that slips the headset around on your head a little bit. There are a lot of ways to approach IPD Tolerance, here are the four most common:
1. Mechanical IPD Adjustment
An oldie, but a goodie. This is the way most VR headsets have tackled IPD tolerance over the decades, driven largely by microdisplay-based architectures with narrow eyeboxes that required perfect positioning to work at all. It allows the HMD designer to focus on other qualities like FOV and field curvature. Tying physical adjustment to the distance between virtual cameras is a great way to ensure the scale of the virtual world is correct, and a single HMD can be adjusted to fit a wide variety of users. It does add a small amount of weight and a moderate amount of fragility. This was the approach used in Rift CV1, though we tied the adjustments together in a way that slightly impacts people like myself with asymmetrical faces. Good enough!
2. Custom Sizing
This is my personal favorite, and will dominate the VR industry in the long run. Every adjustment on an HMD adds weight, bulk, complexity, cost, and fragility. If the headset is fitted to match the end user from the start, you can minimize and sometimes eliminate the need for adjustments! This is a common principle behind almost all wearable goods, from shoes to clothes to watches to sunglasses – it is the reason your wardrobe is not dominated by one-size-fits-all unisex jumpsuits.
Given the small size of the VR market, it is hard to justify fully custom sizing for every single HMD sold. Luckily, you can get huge gains out of even a handful of options! I am not a huge fan of Magic Leap, but one thing they did right was achieving IPD Tolerance by making two different versions of the ML1, one for wide IPD and one for narrow IPD. Focals by North is taking a similar approach with their smart glasses, but have dozens of different sizing options.
I won’t mince words. Rift S should have done this. The logistical overhead of managing a handful of different SKUs with slightly different plastic pieces holding the lenses at slightly different distances would have allowed Rift S to keep costs low and expand the addressable market for VR without cutting out new and old customers alike.
3. Perfect Collimation
Nothing is perfect, but . It is possible to design optical systems with a very wide sweet spot/eyebox and little need for physical adjustment, but not without tradeoffs. Playstation VR is a good modern example, using solid and heavy lenses with a fairly long focal length to create an wide enough eyebox for most users. This comes with a penalty to weight, cost, size and panel utilization (and thus resolution), but it is hard to argue with the results, especially for a headset that regularly sells for $199. The Rift DK2 also had a large-ish eyebox that allowed it to be a useful development tool, but there is a reason some users created aftermarket modifications for lens spacing – we knew it would not be responsible to launch DK2 as a consumer product, and it took nearly two years from the launch of DK2 to make something we felt comfortable pushing as the One For All solution.
4. Everything is Fine
One of the most common approaches to IPD Tolerance over the decades, and one of the many reasons VR has always failed. You can’t solve engineering decisions with marketing, but that won’t stop PR professionals from trying. Doing nothing and claiming Everything is Fine does nothing to keep customers from churning out and failing to engage as a result of blurry images, headaches, and eyestrain. Some people will only see these effects once they start actively using their headset – more people will be within an acceptable range when standing still than during active play, because the same factors that allow for a wide range of IPDs also allow for more active playstyles.
The most obvious sign of this strategy is when any company refuses to admit what their supported IPD range is, especially companies that have proudly marketed it in the past – providing a simple number that would allow prospective customers to make smart decisions would also give customers and press the ability to objectively calculate the number of people being cut off (or, alternatively, just how far the acceptable quality bar has dropped). Much easier to treat it as some kind of insignificant, unfortunate, unavoidable minuscule problem with no demographic-based distinctions whatsoever.
Why are you so bitter about Rift S in particular? What about the other bad HMDs?
Thanks for asking. Yes, there are lots of other headsets that don’t tolerate a wide range of different IPDs. Many early Windows MR headsets, in particular, were prominent members of the Everything is Fine club. There are a lot of headsets in the SteamVR ecosystem that totally blow, as well. I have been quiet about this for two reasons: First, I don’t really use those platforms. Yes, I own almost every HMD ever made, and I keep up with developments in the VR marketplace, but I am completely tied to the Oculus platform. I buy my games from the Oculus Store, I launch my games through the Oculus UI, and I develop VR applications for the military using the Oculus SDK. I have a lot more interest in Oculus because I want to continue doing all of these things.
Secondly and more importantly, Rift S is the only way to use the Oculus PC ecosystem. It is the singular option, a full replacement for the now-discontinued CV1. Anyone who can’t use Rift S is going to be effectively locked out of the ecosystem, including people who have invested thousands of dollars into their content library. For Windows MR and Steam VR, on the other hand, bare-bones headsets are just part of a broader ecosystem that also includes headsets that do support a wide range of IPDs. I have no problem with the existence of low-cost headsets that only work for 70% or 80% of the population as long as they are flanked by options for the rest of us. Oculus Quest is great, but it is not a PC headset – it is an option for people who want mobile VR and can’t use Go, and that is all.
I spent much of my later tenure at Oculus working on supporting headsets from other vendors, in part to avoid this type of situation. As things stand, I find myself shunned by an ecosystem I spent most of my adult life helping to create. To be fair, Oculus did not show up to the GDC announcement of Rift S without a plan to address the concerns of users who find themselves with no future in the Oculus ecosystem. If anyone falls outside of the carefully unspecified IPD range of Rift S, they might want to stock up on the original Oculus Rift while they still can!
I love most of you Oculus guys. Keep trying to cut a few more tools for Rift S IPD+ and Rift S IPD-. Fingers crossed for a future version spaced a hair under 70mm, skewed to the right.