You all surely know that two days ago Magic Leap has surprised everyone announcing that in 2018 it will release its first (dev-kit) device, the Magic Leap One, and showcasing on its website some images of the upcoming device. I’m saying that you surely know because the web has been literally flooded by comments and posts about this news. Magic Leap is a very secretive start-up and we are all so curious about what it is doing that every little teaser on it becomes viral.
A lot of posts were enthusiastic, while others were full of hate. I admit that mine were of the second kind. In this article I want to tell you what you need to know about this device and express my opinions about it, just to explain to you why I’m not that hyped.
If you missed the news
If you were working with Elon Musk on spaceships that can take us to Mars and you spent the last week on Mars, maybe it’s better that you read some articles to keep pace with all the rest of the XR innovators. All the major XR magazines have published one or more articles dedicated to Magic Leap reveals, but actually, they’re just mash-ups of two sources:
- Magic Leap official website (link), that has been completely renewed to show you photos of the device and to tell you more about it;
- The long interview + hands-on on the Magic Leap One that Rony Abovitz, CEO of Magic Leap, has granted to Brian Crecente of Rolling Stone magazine (link). Since this article contains a description of the device and some first-hand impressions, plus the story of Magic Leap as seen from the eyes of its CEO (and Abovitz is a great storyteller), it is really worth reading. You MUST read it.
Once you’ve read all those articles, you have 97% of the public knowledge you could have about this device.
In this article, I’ll just prompt you some highlights and give you my commentary.
What has Magic Leap given us?
Magic Leap has created a huge hype because after years of secretive work, it has finally revealed us its product. Oh well, maybe. Actually, they just gave us:
- Some photos of the device, highly photoshopped. I and other people thought they were just renders, but Magic Leap has written to Ars Technica that they’re actual photos of the device. Considering that Brian Crecente has tried the device and he has not said that those photos are completely different from what he tried, I think that this is true. So we now know how a Magic Leap glass looks like;
- Some photos of the device worn by people, highly photoshopped;
- Some characteristics of the device: which sensors are on, some demos they’re working on and why the device is so special and its future vision (more on this later);
- A nice story of these years inside Magic Leap: if you read Rolling Stone’s article, you can read a very interesting story from the dream of Abovitz of emulating reality to current times. A great story, I have to admit it;
- A lot of fluff on the possible uses of the device, his vision;
- A clear release date: 2018.
The problem is that, in my opinion, this is not that much, considering what they haven’t given us:
- While Crescente has been able to try the device (so there is a device somewhere), there is no public showcase of it. There’s no video showing that the device is actually working and working well. Furthermore, Abovitz has chosen a pop magazine like Rolling Stone and not an XR one like Upload VR or Road To VR because it dreams about his headset becoming mainstream. This means that it has not been reviewed by someone whose job is reviewing EVERY DAY AR/VR stuff. In the article sometimes the journalist compares ML (an AR device) to virtual reality headsets and this has made all of us understand that he was surely not the best person to write a review on it;
- A final design of the device. The website shows us some photos, but there’s also written that from here to the release date, everything can change. So, they’re just showing us what there is now: look, that is Magic Leap… maybe;
- Its complete specifications: we have none of them. We can just infer some of them from patents and from Crecente’s impressions;
- Its price: we absolutely don’t know it. Someone imagines that it will be above $1500 (a past TechCrunch report seems to suggest that, too);
- Its actual release date. They say surely 2018. But they also previously stated that it should be 2015…
So, we have just some photos, some opinions, and a vague release date. In my humble opinion, it is not that much. It is not much more than we had before. The only difference is that now they seem more confident in releasing something this year.
What are its specs? How can we compare it with HoloLens?
Magic Leap One is composed of three parts:
- The AR headset called Lightwear;
- The belt computer, called Lightpack. The Lightpack is the processing unit of Magic Leap, that is connected to the headset through a cable;
- The controller called Control. It is very similar to the one of GearVR;
From what we know, Magic Leap is very similar to HoloLens. In common with HoloLens, it has:
- The fact that it is an AR glass (and all that comes with it);
- An array of microphones;
- Surround audio;
- The ability to scan the environment and detect walls, planes, desks, etc…;
- The ability to remember the rooms you’ve been in, so that when you return in a previously visited room, you still find the holograms that you left there.
It is different from HoloLens because it has:
- Lightfield technology, that allows you to see the objects in a more natural and realistic way (more on this later);
- Eye tracking;
- A slightly bigger FOV;
- A 6DOF remote controller;
- A better design and comfort;
- Various versions and add-ons to make it comfortable on your head (e.g. there will be prescription lenses mounts for people that need glasses)
- The belt computer, that moves the external processing power outside the headset that so is lighter. Abovitz claims that the little smiling box has the same processing power of an Alienware laptop, but in my opinion, it is a bold claim. If this is true, it means that Magic Leap has far more processing power than HoloLens, that can just process few polygons for each frame (ca 20K for Unity applications).
Why I’m not hyped
First of all, as I’ve said, they have given us very few information. In all these years, Abovitz & Co. has had a wrong attitude towards the community, giving us very few info, continuously postponing deadlines, insulting others AR/VR devices (claiming that they’re dangerous), making weird presentations and showing us fake or non-informative videos.
Because of this, my trust in them is very low. And surely it has not increased because of they showing us some photoshopped pictures, some pictures of a design that can change completely in the future (so… why are you showing it to us?) and some bombastic words about this not being a “dev kit”, but “a device for creators” (different words to just say the same thing) and other fluff like that. This means that until they’ll give us a video of a working device and a review by some XR journalist, I won’t believe them. They say that they’ll ship in 2018, but they already missed previous deadlines. So, I’ll just wait for facts and ignore all their words.
Then there is the belt computer. Every time I had to put something attached to my belt, like in the case of OSVR headset, I always found it a total nuisance. In my opinion, it is a step back from Hololens all-in-one device.
Furthermore, as you can see from our previous paragraph, we’re making a comparison between Magic Leap and HoloLens, that is a device from two years ago. And things are too similar. The field of view of Magic Leap is just slightly better than the one of Microsoft’s device. According to Oliver Kreylos calculations, it is just some degrees more:
Not really. I don’t know how long that person’s arms are, and “half extended” is about as fuzzy as it gets. Also: VHS tape? in 2017? Seriously?
Anyway, a VHS tape is 7 3/8″ wide. That’s about the span of one of my hands. HoloLens FoV, for me, is two hand spans at arm’s length across. This is one hand span at half arm’s length, which would end up being the same. I have longer-than-normal arms, so let’s say this guy’s are shorter.
Based on all that and a boulder of salt, I’m going to say it sounds around 40° horizontal, which is in line with earlier estimates based on ML video footage. HoloLens is 31.5° horizontal.
So they took $1.9B to arrive at what Microsoft has released two years ago. 40° of horizontal FOV is very far from immersive, from the revolution in AR they have clamed for years. Abovitz says that they already know how to fix that and that it will be fixed in future iterations (another vague statement… we all startuppers say the exact same thing about every issue our product has).
The problem is that Microsoft has not spent the last 2 years playing with dolls: if you have subscribed to my newsletter (if not… use the form at the end of this article!), you’ve surely read in my weekly XR update that they’ve already announced how they plan to DUPLICATE the actual field of view of HoloLens and have showcased in public hand tracking and AI future functionalities of the device. This means that when HoloLens 2 will be released in 2019, it will take a huge leap forward, with hands tracking, object analysis and recognition (thanks to Microsoft Cognitive Services) and improved optics. If Magic Leap One will be released in 2018, considering current release cycles of XR technologies, the next iteration will maybe be released in 2020/2021, so after Microsoft’s new device. 2019/2020 is also the timeline for the release of the Apple Headset and Apple is a company that is very good on all sides: technical, design and marketing. So, either Magic Leap engineers are able to make something so disruptive that the Magic Leap One will be able to keep the pace of the competition of this enormous companies, or Magic Leap is going to have a hard time to survive.
They could survive better by being humble and asking the community for support, but this won’t ever happen. Magic Leap has always been driven by some marketing craze, promising lots of things that who knows if they’re even possible. They could make the same error of Google Glass: Glass didn’t fail (only) because they were a crappy product, but mainly because they were a prototype accompanied by a huge marketing effort promising enormous things that the device wasn’t even remotely able to fulfill. This difference between promises and facts has been the main reason for Glass’s failure and risks to be something taking also Magic Leap down.
What I actually found interesting
I’m not saying that I’m not giving a heck, either. There are some things of this announcement that actually excite me.
First of all, we have a release date that seems more realistic, so maybe we’ll actually have a new startup in the AR realm… a new entrant, with a new device. I want to believe that they’ll actually ship something.
Then there is the attention to design. Microsoft has made an excellent work with Hololens, but the design looks awful. Abovitz wants to create something that people could wear every day in the streets and while the Magic Leap One absolutely isn’t something you can wear in public places, it has surely a better design than all other AR glasses. It looks less a device for nerds and more a device for consumers. It is cool to see and also very light and comfortable to wear. And also all this attention towards spreading the word to creators, influencers, pop stars (they made Beyoncé try the device), pop magazines makes us understand that they truly believe that they can appeal the general consumer. Even the website is clearly not a website for nerds: it focuses on what you can do with the device and not on numbers and specifications. So, they’re already focused towards mainstream adoption and that’s awesome.
The controller is 6DOF. This is possible thanks to magnetical field sensing (the same mechanism that powers the Razer Hydra as a redditor points out) that is a bit less precise than optical tracking, but more immune to occlusions. The use of a remote is not that immersive, but the fact that they implemented a 6DOF remote inside their first dev kit is a great choice since HoloLens interactions with fingers or gaze are damn uncomfortable. If first apps will be mostly UI based, a 6DOF controller will be perfect for them.
View post on imgur.com
(The green circle highlights the “electromagnetic coil assembly” used for the magnetical tracking)
But the thing that excites me the most is the lightfield technology. I’ve talked about lightfield and how awesome they are in my article about holographic displays, but if you don’t want to read it again, I’ll make you a very short summary. Our eyes just see light coming from the environment about us: we see through light rays that arrive at our eyes and these rays have a direction, a color and a strength that our eyes and our brain process to understand the world around us. If we were able to re-create exactly all the light rays that would come from a virtual object that we put into our environment in AR, with the correct direction, intensity, etc…, our brain won’t be able to understand that the object is virtual and not real. The problem is that creating such a precise lightfield is almost impossible. What Abovitz and his engineers have discovered is that it is not necessary to re-create exactly this lightfield, but we can create an approximation of it, because from one side the eye is far from being perfect, so a lot of information gets lost before arriving at the brain and from the other side our brain is very good in filling holes. This is a bit like the MP3: we can reduce a lot the size of the audio files because we can remove some information and still not notice the difference.
So, Magic Leap technology creates a digital approximation of the light incoming from the virtual object that so appears as real. In comparison, HoloLens doesn’t have this technology: it just shows the representation of the virtual object from the point of view of the left and the right eye… it doesn’t recreate how exactly the light emitted by it should be. Magic Leap technology allows its device to solve the vergence-accomodance problem since the objects appear as being at different distances (near, middle, far) exactly as the eyes would perceive them: HoloLens has only one focal plane, while ML has multiple ones. This makes vision more natural and more comfortable for our eyes. They want to re-create exactly the same vision we humans have… so, for instance, they have a demo where the more you get close to the character, the more you see details: this is exactly what happens in our everyday life. Furthermore, the objects appear as opaque and not semi-transparent as in HoloLens. This kind of display is the huge innovation of Leap Motion and most probably the technology that has taken them to get so huge investments. And there is to say that according to past rumors, Magic Leap has been forced by investors to release something to the public, because they actually have under the hood some even more awesome technology (initial rumors from 2014 talked about the possibility of having retinal displays, that is images projected directly onto the retina), that they haven’t still been able to miniaturize and so it will be implemented only in future releases. This is the real thing that excites me about Magic Leap.
Anyway, I have to say that other glasses, like ODG, will implement lightfield too. And that almost all AR and VR companies (including Oculus) are working on multi-focal optics. So, even here, they have a lot of competition.
This are my impressions about Magic Leap: I’m both skeptical and interested. I advise you to register on the form available at the bottom of their website to get further info about it… let’s see together what they’ll be able to provide us!
(Header image by Magic Leap)
The post Magic Leap announces the Magic Leap One for 2018: what you need to know and why I’m not hyped appeared first on The Ghost Howls.