Although it’s expected to launch this year, there’s still no firm release date on Magic Leap 2. However, the company has begun sharing details on the headset which suggests the launch is approaching.

This week at the SPIE Photonics West 2022 conference, Magic Leap’s VP of Optical Engineering, Kevin Curtis, took to the stage to share a bevy of new details on the headset. Attendee Nataliya Kosmyna shared portions of the presentation alongside some portions which cropped up over at the AR XR MR subreddit.

During the presentation Curtis detailed Magic Leap 2’s bevy of sensors, optical stack, Android foundation, and more.

The most interesting details to come from the presentation are perhaps about the headset’s dynamic dimming capability, which is a first among commercial AR headsets.

Dynamic Dimming Lenses

Image courtesy AR XR MR

Curtis shared that Magic Leap 2 can adjust the light transmission of its lenses from 22% to 0.3%. The former being something like sunglasses and the latter being closer to welding goggles. This wide range ought to make the headset usable even in very bright outdoor environments (though it will of course come at the cost of dimming the world around the user as well). Dynamic dimming is paired with a brightness range from 20–2,000 nits; combined, these capabilities should make the headset significantly more flexible than its predecessor, and similar headsets, when it comes to varied lighting conditions.

Also noted in the presentation, the dimming capability can refresh at 120Hz, and is “segmented” as well. The slides state this means that Magic Leap 2 lenses can “enable black,” presumably by selectively dimming only the part of the lens where black is needed in the image. On traditional transparent AR headsets it’s impossible to have ‘black’ as a color because black is the absence of light but the lenses have no way to stop light from passing through. Without access to the entire contents of Curtis’ talk, we don’t know how precisely the dimming capability can be segmented so it’s difficult to know if this will be a comparatively groundbreaking capability, or something more limited.

The Apple Effect: Magic Leap Founder's Previous Company Launches App on Vision Pro

Curiously, dropping to the minimum 0.3% light transmission might even make Magic Leap 2 useful for fully immersive VR experiences where the real world is largely dimmed to make way for entirely virtual content. It remains to be seen if this is a use case the company is actively aiming for.

The Headset

About the headset itself, Curtis shared that Magic Leap 2 will come in just one size—a change from Magic Leap 1 which had a ‘large’ and ‘small’ variant. The reasoning behind having two sizes for ML1 appears to have been driven largely by a small eyebox and the lack of IPD adjustment, requiring two different sizes of the headset to try to cover a suitable range of the IPD spectrum. For Magic Leap 2, Curtis says the eyebox was doubled in size, apparently making it large enough for the company to move to a single headset size.

According to Curtis, Magic Leap 2 weighs just 248 grams (0.5 pounds). That’s a nearly 22% reduction over the original headset’s 316 grams (0.7 pounds), while furthering its lead in weight over HoloLens 2 which comes in at a much heftier 566 grams (1.2 pounds). Granted, Magic Leap 2 still relies on a tethered connection to a ‘Compute Pack’ which gives it a big advantage in the weight department over fully self-contained headsets.

ML2 will also include eye-tracking, with two cameras per-eye. As far as we know, that’s up from one camera per-eye on ML1, which could mean greater accuracy. Eye illumination is provided by six tiny LEDs which can be seen embedded in each lens.

Squad-based Shooter 'Breachers' Gets Ranked Competitive Multiplayer in Latest Update
Image courtesy Nataliya Kosmyna

It isn’t clear yet if Magic Leap 2 will have the same varifocal capabilities as ML1. Given the diagram shared by Curtis (further above)—which appears to show one waveguide per red, green, and blue color (instead of two per color as with ML1)—the feature may have been scrapped.

The headset will include a 12MP RGB camera for user-facing photo capabilities, like taking pictures & videos, scanning barcodes, and streaming video for ‘see-what-I-see’ use-cases. On-board audio is also confirmed.

Image courtesy Alessio Grancini

Two key things we don’t know yet are the resolution and field-of-view of Magic Leap 2. According to Curtis, Magic Leap 2 has “double” the field-of-view of ML1 (which is 50° diagonal), though we expect this means double the area, not double any of the linear dimensions. The company has previously shared this comparison of the ML2 field-of-view vs. ML1 in which we can see most of the gain comes in the vertical direction.

Image courtesy Magic Leap

Compute Pack

As for performance, Curtis indicates the Magic Leap 2 ‘Compute Pack’ will have 2–3 times the GPU & CPU performance of Magic Leap 1, including the addition of a dedicated co-processor for handling computer-vision operations, which he referred to as the CVPU (computer-vision processing unit).

Curtis didn’t specify Magic Leap 2’s processor, but the company has strongly hinted that it comes from AMD. Interestingly, Magic Leap 1 was based on Nvidia’s Tegra chipset, which was fairly novel for this kind of device.

AMD is also a somewhat novel choice for Magic Leap 2 as most devices in this category use chips from Qualcomm, including Magic Leap’s main competition, HoloLens 2. You may know AMD as a creator of desktop & laptop processors as well as GPUs, but the company also has a significant foothold in the console market as the longstanding provider of chips in Xbox and PlayStation consoles. Magic Leap 2 could represent AMD’s first significant entry into the XR space.

Hands-on: Apple Upgrades Personas for True Face-to-face Chats on Vision Pro
Image courtesy Nataliya Kosmyna

Magic Leap 2 also touts “more memory, storage, and battery life vs. competitors,” but we don’t have many specifics at this time. From the materials we have we know that ML2 will have USB-C charging, Bluetooth 5.1, and WiFi 802.11AX (AKA WiFi 6).

Continue on Page 2: Controller, Embracing Android »

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."
  • VRFriend

    Completely overengineered crap.

  • What will be the point of Magic Leap and HoloLens 2 devices when Meta release Cambria with full colour passthrough? Magic Leap’s field of view is as bad as the HoloLens. If Cambria has clear/full colour passthrough (which seems likely based on short clips) and comes in at under 1k, with all its other benefits, buying a Magic Leap or Hololens would be stupid (VR and AR, plus double the FOV, eye tracking, hand tracking, a whole suite of VR applications, similar size, etc. etc.)

    • dk

      cambria and similar headsets won’t give u the perfect fidelity of the real world ….at this point
      and the high angular resolution varjo is pretty big and super expensive

      • They will easily give you a good enough passthrough image to be used for everyday stuff, gaming and work AR use-cases. Magic Leap 2 seems like a total waste imo. I can’t think of a single good reason to get it over Cambria. Some might say it’s smaller, but it’s still a pretty chunky pair of glasses, attached with a wire and a compute box you have to carry around too, so that’s not a strength of Magic Leap imo.

        • Cooe

          Spoken like a true idiot that’s never actually used a proper AR headset themselves… -_- …

          Passthrough, even the best full-color stereo passthrough possible atm (what Cambria will bring) isn’t even REMOTELY close to being a “full replacement” for normal vision. And ESPECIALLY in dangerous work environments like a factory floor! (The quality of depth & parallax reproduction isn’t even in the necessary UNIVERSE yet, let alone in the immediate ballpark!)

          This is like complaining about how people kept going to the movie theater in the 50’s & 60’s once they had SD TV’s at home. Sure, they are both ways to “watch a movie”, but the quality & tenor of the experience aren’t even REMOTELY close to “being the same”.

      • kontis

        Neither will Magic Leap with those tiny pinholes greatly reducing that real world view.

        • dk

          I was talking about fidelity …looking at your phone or monitor or 20/20 vision or having vergence accommodation of the real world …….but yes blocking a part of your vision is a disadvantage of ml2 compared to H2

      • brandon9271

        I’d rather have lower fidelity and a bigger FOV. Plus, I think most consumers would prefer a MUCH cheaper AR/VR solution over a much more expensive AR only solution.

    • This is basically what I think. Sure, Cambria’s passthrough isn’t going to be as clear as just looking through a slightly tinted less, but given that either way you’re gonna have to wear chunky glasses on your face then I’d rather just get the all-in-one device that is great for both VR and AR. And, as long as the image of the passthrough is clean and stable, I would be totally fine with that. I’ve managed to use the current meh Quest 2 passthrough just fine as is, and Cambria’s solution already looks to be a major step up. Personally, I see absolutely not smart reason for any sane person or business to shell out for a Magic Leap 2 over a Cambria.

      • kool

        Wh I’ll le I don’t want a mk2 I don’t think a vr passthrough is better for all day ar. Cambria might do some cool stuff with ar but it won’t be an all day device until some glasses with at least 100 fov can be worn comfortably. I think some glasses with a phone as the brains will come first. It’ll have to rely on bluetooth, wires are unacceptable for all day use.

        • Magic Leap 2 isn’t an all day device either. Anyone who believes any current AR glasses are all day devices is living in la la land. We’re still a ways from that yet.

          • kool

            I dont really mean all day, I meant a device you can actually use during the day. Like walking down the street or taking a crap…you know daily stuff.

          • Ah, I get you.

    • Christian Schildwaechter

      Passthrough gives you only the focus of the cameras, unless they not only figure out how to precisely track the eyes, but also adapt to the focus of our eyes’ lenses in real time. Which is almost impossible. The human eyes do a number of very fancy things to aid with depth perception, for example there is a constant micro-stutter with the eyes slightly twitching to look at objects from different perspectives. None of these things will work with passthrough.

      And while the FoV in which Magic Leap can actually augment reality may be small, the FoV on what you see of the world through the non-display parts of the glass can be a lot larger, and even extend beyond the display itself. Just like a pair of sunglasses with very thick frames, Magic Leap is open to the sides, so objects in your peripheral vision will still be visible, while Cambria will probably lock you out of everything that isn’t captured by the cameras.

      Add to that the reduced weight, and devices like Magic Leap could be much less intrusive when worn all day doing regular tasks. And that is the direction that AR is heading for, smart glasses that you barely notice. The current generation is very far from that.

      Realizing AR with VR HMDs and camera passthrough currently has a number of advantages like really being able to block out external light, much higher FoV for graphics or much higher contrast. But it also severely restricts the human vision, and we will have to see if it is even bearable for long term usage. Some people prone to get motion sick could simply be unable to deal with the minimal latency and lowered framerate the cameras introduce. We will know a few months after Cambria is actually available.

      • Early on Lynx wanted to add varifocal lenses to the cameras but decided that the technology was not ready.

        • Christian Schildwaechter

          Even if the lenses were available and allowed matching the eye’s focus, they probably wouldn’t be fast enough. A (young) human can shift the focus from extremely close up to infinity in less than a quarter of a second, unmatched by artificial lenses. Looking around the room, you can focus on several objects per second, partly because the brain estimates the needed focal length from the blur. The lenses are already (roughly) focused when your gaze reaches a new object, and only then could eye tracking attempt to match it.

          So you should probably avoid driving a car through heavy city traffic while using passthrough AR. For most other use cases the response time isn’t all that critical, it might be more of a comfort issue, but focus time is only one of the eye’s features that (currently) do not work with passthrough.

          For no longer young humans passthrough AR might even restore some lost abilities. At 17 the eye accommodation allows to adapt to different focal lengths, matching about 15 diopters. By the time we reach 45-50, we are down to two diopters, and at 70 only one Diopter of focus range remains.

          So AR/VR with variable focus is most relevant/useful for younger users anyway. Above a certain age you actually benefit from everything in VR (currently) being shown at the same focal length, as you will see things in VR that would require wearing multifocal glasses in the real world. Maybe this can be adapted for AR too, making passthrough AR the ultimate corrective eyeglasses.

          • I agree with most of what you said. Just a quick heads up: PoLight TLens has a response time of 1ms. Of course, you need to add other latency, from eye tracking etc.

    • Malkmus

      Well to be clear ML2 does not have the same FOV as Hololens 2. HL2 is 50 degrees and ML2 is 70. That being said, Cambria is going to be great (I hope) for office use, but there’s now way a passthrough device is going to be usable in factories, construction or medical where precision of the real world view is critical and should the display ever have a glitch or turn off you’re not blinded. The Lynx R1 however has a better chance of bridging that gap as it has the open peripheral view so you don’t lose total vision in case of an issue.

    • Andrew Jakobs

      Let’s first just wait until Cambria is actually released or have shown more details of the production model.

    • Hivemind9000

      Until passthrough can give you a 200+ degree field of view (with natural looking parallax/depth perception), AR headsets like these and Hololens are going to be far safer in an industrial situation (which is where they seem to be focusing now). Situational awareness around robots and other people/machines is critical.

      For anything else (especially entertainment/games), I agree, passthrough is going to be better and cheaper.

      • Natalie Hill

        I am earning 85 US bucks per hour to complete few tasks from home.~er481~I’ve certainly not realized that it can be possible however my confidant pal was collecting $26,000 just within three weeks just working this best offer and also she has convinced me to join.~er481~Visit for more details going this web page.

        >>>> https://https://­t­.­l­y­/googlestar

    • Lucidfeuer

      Meta is the enemy of VR and consumers but…you’re right.

      AR headset is a “fake” product, there’s absolutely no use for it because the underlaying tech are decades away from being product-viable, VR pass-through being of course the way to go (and should’ve have been since 2016).

      It baffles me how most companies from Apple to Google are trying their hardest to fail, whatever basic analytics, prospective and strategic misreading led them to invest in AR instead of VR…

    • The point is wearing something that looks like glasses. They may not look like it now, but these devices will eventually turn out that way. They will eventually become glasses in form factor.

    • Cooe

      Only an absolute freaking idiot equates even the BEST full-color stereo 3D passthrough with ACTUAL unimpeded human vision from external light sources… -_-

      This is like wondering why people still went to movie theaters in the 50’s now that they had SD NTSC TV’s at home… They might “both be ways to watch movies” but the quality & tenor of experience just isn’t even REMOTELY the same!

  • kontis

    it doesn’t matter what amazing features a HMD with FoV lower than 80 has.

    Look at this amazing car. Its wheels are broken, but it’s a beautiful car with the most amazing super comfortable seats.

    • Malkmus

      I mean Hololens 2 with only 50 degrees is considered an industry leader so not sure your analogy makes much sense.

      • Brianna Young

        I have even got paid $17230 in no more than three weeks through working easy jobs from apple laptop.~tq034~As I’ve lost my last post, I was so disconcerted but finally I’ve came across this online project now I can figure out how to receive thousand USD just staying at home.~sw069~Each individual can certainly start this best work and can get extra greenbacks online by exploring this site.

        >>>> https://­t­.­l­y­/tTrG

    • Cooe

      Translucent display AR ≠ fully isolating traditional display VR, you absolute freaking dumbass… -_-

  • Keopsys

    Does anyone care about the Theranos of AR ?

    I would never buy anything from this company.

    • Lucidfeuer

      They started as the Theranos of AR, but they were forced to actually produce some “non-revolutionary” headset to be credible, so now they’re just an actual AR company (probably still a front for speculative-laundry)

    • Cooe

      Theranos never shipped a working product. Magic Leap has done that TWICE.

  • XRC

    Got to spend time using Magic Leap 1, it was impressive after all the negative reviews. Look forward to trying Magic Leap 2, but the reported omission of dual focus is sad, because it worked well. Attenborough “Rise of the dinosaurs” was stunning on ML1.

  • Very interesting details. I’m curious to see how the dynamic dimming performs well. These features are usually great on paper and then they have only decent performances when used

    • Sven Viking

      I’m just glad to hear Magic Leap talking about actual specs rather than the usual Magic Leap stuff.

  • Lucidfeuer

    It’s funny that what started as a tech money-laundry scheme had to eventually produce a prototype and product under the scrutiny and skepticism, and now they’re actually making interesting experiment and research (that is unless I missed similar advancement in the field, and they’re just copying).

    • Malkmus

      Bizarre that people think this was any sort of scam (much less a money laundering scheme). Consider that no investors have ever complained or sued them. When that sort of thing happens people tend to make a lot of noise. Look at Theranos and Wework, full of lawsuits and investigations alleging fraud. If you want to know the real story of what happened I recommend watching the XR Talk Youtube show which is made by ex-Magic Leap employees and they discuss in detail the history of what investors saw along the way to producing ML1.

      • Lucidfeuer

        People didn’t think it, you don’t seem to understand how speculative money laundery scams like Bloombox or ML work, they’re just companies like Theranos or Wework that didn’t mismanaged they front and therefor didn’t get sued. Now I don’t care about you’re either naive, stupid or astroturfing opinion, but it was spectacular how obvious of a scam it was when they started talking in Wired for example, and it became clear to any analysts that they didn’t have any product or technology (let alone as “revolutionary” as they dubbed it), until eventually, they just had to go and make product for the front to be credible. The result was that it was nothing revolutionary, nor even innovative, and barely sold, but still serve as an “overvalued” speculative fund like Wework or Theranos, which were either mismanaged or unveiled.

        • Malkmus

          Bloombox was also sued by investors. So your analogies still don’t make sense. You don’t scam a bunch of people out of billions and no one care to make a peep. You don’t launder billions and not catch the attention of the feds or SEC. That you think a simple fact like this is “astroturfing” says a lot more about your inability to asses the situation as you clearly view everything through a paranoid lens. Guess what, the truth isn’t always as juicy as you want it to be. And sometimes investors and the companies they work with can both fail at what they’re trying to make, but that’s doesn’t make it a money laundering scheme. Now if there was even one publication in the last decade to expose this scam, even so much as one whistleblower you would have some credibility here.

          • Pablo C

            Since you are deffending this, I´d like your opinion of the first device itself (ML1). I guess if it´s a good device, that would settle this discussion.

  • Tom_Craver

    How is dynamic dimming ‘news’? Hasn’t it been obvious from the start that they’re blanking out real world light where they draw virtual pixels?

    • Malkmus

      It’s news because no other optical AR device (Hololens/Nreal/Vuzix) has that capability. And the inability to display black has been one of the biggest drawbacks of AR.

      • Tom_Craver

        “News” ought to be “new”. I guess the specific percentages of dimming may be news…

        • Malkmus

          I’m curious what other headset you think already does this if you don’t consider it new.

          • Tom_Craver

            Not saying others have it. Just that it’s been in every Magic Leap headset – that’s how their “magic” works.

          • Malkmus

            This does not exist in ML1 either. I think you’re either misunderstanding what it does or are unfamiliar with ML1. This is actually a pretty big deal for AR headsets — if it works as intended.

          • Tom_Craver

            Oh? How do you think they were making virtual objects appear ‘solid’ over a lit visual background in ML1? Doing that literally REQUIRES selectively blocking the light from the real scene. That was a huge part of what made ML1 unique, and was their big ‘secret’ ingredient. All that’s happening here is that ML has revealed the magician’s trick.

          • Malkmus

            The dual focal planes were their “secret ingredient” in ML1 which they’ve now ditched because it wasn’t good enough. And have you used a ML1? The number one complaint after the small FOV is that the virtual objects are transparent. Anyway, don’t take it from me. Ask literally any Magic leap dev on Twitter or their official forums and I guarantee they will tell you the same thing I’ve been telling you. Or just read any of the several interviews with the CEO who also says this is a new breakthrough for not only Magic Leap but the whole industry. Im not really sure why you’re trying to even debate something that’s so obvious, unless you’re trolling.

  • ZeePee

    This is honestly, very exciting.

    We can have our demands of what FoV we need etc, and be dissapointed at how slow AR is taking, but the fact of the matter is, we are now on the cusp of what we have been waiting for.

    This device is amazing. It sounds like it could be very, very close to what we want.

    ML3 is probably the headset we desired for ML1. We’re almost there. This tells us that in 3 or so years, we will have true AR glasses.

    Until then, VR pass through will probably do a very good job. I bet Meta’s project cambria is significantly better than ML2, purely because of the FoV. The sensor quality and tracking capabilities / latency, is probably going to be good enough to the point where it is as compelling as a true AR headset, but with 100°+ fov.

    I imagine Apple will get there around the same time as magic leap, and we might see something from meta at that point too. In 3 years time. This year is the beginning of consumer AR due to VR pass through, but 2025 is the beginning of true AR consumer glasses.

  • At this rate they might actually fullfill their wild promises… in another decade or two. Meanwhile, AR is moving towards VR with pass-through. There is no better way to properly mask AR elements so they appear less ghost-like and work in all lighting conditions.

  • Pablo C

    I missed the first! how was it?