Passthrough AR on XR-3

Photo by Road to VR

XR-3 isn’t just for VR though. The headset includes a bevy of sensors which give it both inside-out tracking and full-color passthrough AR including LiDAR for real-time depth-sensing.

While the cameras on the front of the headset don’t quite match the resolution of the displays, they’re definitely higher resolution than what I’ve seen through any other passthrough AR headset, and they cover nearly the entire field-of-view of the lenses.

While pure passthrough video works very well, the headset can of course also project augmented reality visuals into the world around you. To demo this, Varjo spawned a 1:1 scale car in the room I was standing in. Combined with Ultraleap hand-tracking and the headset’s LiDAR, I could reach out with my real hands to open the car’s doors, and even see my real hands occluding the augmented imagery which made it feel all the more real. Granted, the edges of the hand-occlusion are still too jagged to make it all perfectly seamless.

While the headset’s passthrough AR capabilities are impressive (including some absolutely magic VFX tech which maps the surrounding room onto the reflections of virtual objects), I did notice a bit of latency, especially between the passthrough view of the real world and the virtual objects layered over top (ie: they didn’t feel perfectly ‘locked’ to the same space). I may be a bit spoiled with Oculus’ ‘Passthrough+’ tech on Quest 2, which offers a remarkably low latency experience even if the view is much lower resolution than what you can see through XR-3.

XR-3 Ergonomics

Photo by Road to VR

The other big change to Varjo’s XR-3 over its predecessor is the ergonomics. XR-3 is purportedly lighter, but it’s still a fairly bulky headset at 980g. Luckily the XR-3 has a brand new headstrap that’s much more adjustable.

The major parts consist of a top strap to keep the headset up, and a rear strap that swoops down low to keep it balanced. Both have dials which you can turn to tighten them as needed. On top of that there’s dials right near your temples which allow you to tilt the visor up or down without putting additional pressure on your brow or face.

Photo by Road to VR

All in all it feels like an improvement, and I especially like the ability to tilt the visor; everyone has different brow and cheek topology, and this should make the headset adjustable to a wider range of users.

– – — – –

Varjo is burning the candle at both ends. On one side, they’re improving their hardware and delivering better visuals with each iteration. On the other side, they’re reducing the cost of their headset to the point that it’s becoming increasingly viable for more than just Fortune 500 companies. It’ll be years until we see this kind of visual quality in a mainstream headset, but this glimpse shows not only that it’s possible, but that it’s definitely coming.

Newsletter graphic

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. More information.

Ben is the world's most senior professional analyst solely dedicated to the XR industry, having founded Road to VR in 2011—a year before the Oculus Kickstarter sparked a resurgence that led to the modern XR landscape. He has authored more than 3,000 articles chronicling the evolution of the XR industry over more than a decade. With that unique perspective, Ben has been consistently recognized as one of the most influential voices in XR, giving keynotes and joining panel and podcast discussions at key industry events. He is a self-described "journalist and analyst, not evangelist."
  • wow

    The annual fee is what deters me. I’ll stick to pimax for now.

    • Brooke Hedgepeth

      I am able to obtain 85 US dollars/h to complete few services on apple laptop…~r2225~ I’ve certainly not realized that it is achievable however my top mate was earning $25k within a month easily completing this leading opportunity as well as she had convinced me to try…~r2225~Look further updates by visiting following article >>>

    • Bob

      I agree with this. The $3200 price seems significantly more bearable than the price of their previous headsets but the annual fee is still a major deterrent.

      It is expensive and if it was just the $3200 alone without the annual fee, I’m sure many individuals (dare I say “hardcore enthusiasts”) would jump at the chance to pick this up. I’m sure they’ll get there eventually and correct me if I’m wrong but I did read somewhere that Varjo were interested in expanding into the consumer market at some point. If things continue to go at the rate that they are now, in terms of their hardware iterations, it’d just be a matter of time before Varjo enter the consumer market with a product that doesn’t require an annual business fee.

      And rest assured – myself and the enthusiasts will be looking very forward to that day! Hopefully!

    • david vincent

      Why would you buy an enterprise VR headset, especially since SteamVR apps don’t take advantage of the headset’s unique foveal display system…

    • tomchall

      its an enterprise device, no 3 month RMA wait for an index at this class of device

      • wow

        I bought my pimax from a reviewer. It would have really annoyed me if I ordered any product and had to wait 3months to 1 year especially when I’m used to 1-2day shippi g from Amazon. I wouldn’t have bought a decent headset I had to wait.

  • Hmmm, this pupil swim seems like something that needs to be eliminated. I get a bit of wobble/distortion when I move my head around on my Quest 2, and I’d really like this kind of thing to be fixed asap.

    • kontis

      Valve’s engineer once posted on Reddit that serious comfort issues like this one was what drove them to embrace fresnel lenses, despite their other… glaring… problems.

      • RFC

        From valve’s Alan yates (originally posted on Reddit in relation to the Vive gearvr lens mod):-

        “frensel lenses were specifically designed to minimise some dynamic distortions that we know can cause discomfort and motion sickness. The frensel lenses were not selected for low mass, low cost, hiding subpixel structure, filling SDE or any of the other crazy conspiracy theories I have read. They were the only practical lens technology for hitting the overall set of optimisations we wanted, especially minimising eye-position dependent distortion with a single element. They are not “cheap” lenses and need special equipment to make well. They are lower mass than the “equivalent” non-frensel profile lens, but that is mostly a happy coincidence, if a conventional lens could achieve the same performance in the axes we care about we’d happily tolerate the small mass increase for the reduced stray light and easier moulding. Our goal was to have lenses that worked well for everyone, from the least sensitive to the most easily nauseated. Some people just don’t perceive pupil swim, at least not until you tell them what to look for, and some people once they see it can’t unsee it and it ruins all HMDs with swimmy optics forever for them. Most concerning is that swimmy HMDs cause nausea at an almost subconscious level, you don’t need to perceive it for it to make your experience using the HMD unpleasant.

        We also knew using frensel would mean accepting more stray light (aka “god-rays”) and would make the lens much, much harder to manufacture. It is technically difficult to make frensel lenses with low stray light by injection molding, you also have to be careful about spatial frequencies and a bunch of other important details. The Vive panels are brighter than other HMDs so we actually did quite well to keep the stray light to the levels you experience in the Vive lens. This is one of the many knobs HMD makers can use to deal with the various trade-offs in the optical system; turn down the brightness, soften the contrast, adjust the lens MTF, reduce the FOV, shrink the eye-box.

        There were a lot of design compromises in the Vive lens. Just whacking some magnifiers in front of a panel and calling it good will “work” to some degree, but devil is in the details if you want good performance. The entire point of a HMD is to produce stimulation of your visual system that is as close as possible to the natural light field you experience viewing the real world. At practical consumer cost points and with the technology available right now the lens is near-optimal for the panel and the objective function it was designed for.

        Why do you think almost every high-end HMD since our Steam Sight prototypes were demoed uses a frensel based lens design?

        Now I don’t for a moment suggest there aren’t better optical designs possible. We already have better ones, no doubt others working the HMD space have also caught up and likely have their own high performance designs for next generation HMDs too. I also recognise that some people care about different aspects of lens performance than others. But if you are wanting a “better” lens for your current HMD just realise there is a lot to try to optimise for at once and there is a great deal of prior art to understand before you can truly design something objectively better.”

  • DuxCro

    One day this visual quality will come to consumer headsets. Probably sooner than we think.

  • Rudl Za Vedno

    60+ PPD? Now that’s impressive!

  • MosBen

    Cue the people saying, “Absolute fail. It should be less than $1,000. Then they’d sell a lot more units and make more money.” because they don’t understand what an enterprise product is.

    • Sabrina Dearman

      I’m creating $88 every hour to complete few services from home…~gi281~I never thought like it is achievable but my close friend was obtaining $24k within a month doing this top work and also she had influenced me to join…~gi281~Look more info through following web-site >>>

    • Gail Perkins

      I’m being given $88 per-hr for completing easy work staying at home…~au178~I never realized like it would possible however my top pal was obtaining $26,000 within a month completing this top task also she had influenced me to join…~au178~View extra guidance on going this web-page >>>

  • Geoff

    In this hot market, they will have to enter the consumer space or somebody else will, knowing that this is no longer a concept product.

    • mpaforoufakis tsiou

      VR enterprise market today and in the next 10 years makes more sense than the consumer market

    • Lucidfeuer

      This is still pretty much in a concept product space, having “focus display” as they call it and even external tracking/passthrough system is still experimental even it should’ve been implemented from the very beginning of this VR paradigm.

  • Ad

    Are these using SteamVR or are they already openXR and their own times? I know they have their own SDK.

    Also how comfortable is it? I imagine this is easy to swap out pads, but I still expected some kind of luxury headstrap.

  • guest

    Uh, does it get bricked if you don’t pay your second years annual fee? And speak of bricks, do bricks more or less 980 grams?

    • Bernard Cozier

      I guess they can program an expiration date in the firmware until payment is made and can download a new firmware.

      • Lucidfeuer

        That’s not how it works legally (thank god), it’s like software update: if you stop being subscribed you just don’t get newer update but can of course continue using your current one

  • Manu

    We need a consumer version :) without fees
    Kindly support contacting Varjo

  • Bernard Cozier

    So what if a company decides they no longer need it after one year, they going to be forced to pay the $800 annually? How? On a canceled credit card?

    • benz145

      No they would stop paying the annual fee and some services would probably stop working.

  • Ben, I tried VR2 in the beginning of 2020 and I remember being disappointed by heavy lens distortions when I moved my eyes. So this seems to be a problem that haunts Varjo headsets since a while…

    • benz145

      Thanks Tony, next time I see one of the older models I’ll double cheak as well to compare.