Magic Leap One first impressions review: it’s not a leap, but a step

I thought it would have taken me a lot before I would be able to try the Magic Leap One, but luckily this has not been the case. Thanks to a friend and a bit of luck, I’ve been able to test it and so I will be able to give you my impressions on the device… are you interested?

Last Saturday I visited my former startup partner Gianni Rosa Gallina. If you follow me since a while, you should recall that I had a startup focused on full body virtual reality called Immotionar: we mixed VR headsets and Kinects to let you use your full body in VR. We had full-body room-scale VR in 2014, we were so cool. But while we were great on the technical side, we weren’t good at doing business and marketing (I only started my blog in summer 2016), so in 2017 we closed the curtain on our crazy adventure.

It was a nice ride, and now Gianni, that is an amazing developer, is working in a big company and I’ve started a new adventure in trying to establish a consultancy agency called New Technology Walkers. Anyway, a startup bond is something hard to break, so Gianni and I kept in touch and so on Saturday, I went visiting him and his family. At a certain point, I asked him: “oh, you said me that you have tried Magic Leap… how is it?” and he answered with “You are going to tell me in a while“. And then, as a magician, he took out a Magic Leap One and let me try it. I was amazed.

Thanks to him, I’ve been able to spend half an hour with a Magic Leap One on, trying some apps and demos. And so, I’m now able to give you my first impressions on the device. I’ll try to be the most objective possible about it, trying to ignore all the fluffy hype of the company that has promised us the impossible some years ago and that made me critical or ironic towards Magic Leap various times. Ready, go?

Composition

I guess that you already know this, but I’ll repeat it the same: the Magic Leap One system is composed of three parts:

• The actual headset, the Lightwear;
• The computational unit, the Lightpack;
• The 6 DOF controller, the Control.

The fact that the headset does not contain the computational unit, but it is just basically a display, makes it very light. This comes anyway at the expense of having always a cable that runs from the headset to the Lightpack. The controller is instead a remote that reminds one of some VR headsets like the Vive Focus.

Appearance and design

Magic Leap One is cool to wear. Deal with it. In my opinion, this is one of the greatest innovations that it has introduced in the AR ecosystem. Yes, it is still not as cool as a pair of Rayban sunglasses, but shooting a photo with a Magic Leap on, everyone seems cool. That’s why, after its release, I’ve seen a lot of people posting selfies with this device on and many of them also keeping a photo with the Magic Leap One as the profile pic (Charlie, I’m talking about you :P). That also helped a lot the marketing of the product, that seemed immediately “the cool thing to own”… until the first reviews came to light.

With HoloLens or Meta, there is not the same sensation: they look more like tech gadgets or lab glasses. I’m not saying that they are bad, but they have not the same style of Magic Leap. And I think that if we want XR to be widespread, we need to make it stylish, to make it trendy. That’s also because Mister President walked the runway at Shanghai Fashion Week: to show that VR can be a fashionable gadget.

Walked the entire runway at #GQ show with a #ViveFocus on at the #ShanghaiFashionWeek today! Coat was designed with a #VivePro by #Pronouce! #VRandFashion @htcvive pic.twitter.com/l1ULhtcMt7

— Alvin Wang Graylin (汪丛青) (@AGraylin) March 28, 2018

Comfort

IMHO, the other great innovation of Magic Leap. There is a great attention to ergonomics: the device comes with various pads and add-ons to configure it so that to fit perfectly the face and the nose of the user.

And have I said you that is light, haven’t I? It is DAMN light if compared with HoloLens… when I took it in my hands, the first impression I had, has been “Is it true? How can it be so light?”. Then Gianni reminded me of the Lightpack and I got it. Yes, of course: all the computational units are in the external computer, not in the headset, that’s why it is so light.

And thanks to the above two points, the headset is very comfortable.

But don’t forget the “no free lunch theorem“: all of this must come to a price. The first price is having a mini-PC attached to your pocket: this is usually not a big issue, anyway, since it is quite light (unless you don’t have pockets to attach it to… in that case, you have a problem). The second price to pay is the cable between the mini-PC and the headset. If you don’t set up this cable in the right way, it is a great nuisance all the time and it ruins your experience. If you setup it in the right way, letting it pass behind you, along your back, then it is only an occasional issue. In the end, I haven’t loved that much the cable and I hope for a wireless solution for Magic Leap Two. There is a reason if John Carmack has said that they discarded this design for the Quest on purpose after various focus tests.

A final note on the setup instructions: I really don’t get why they have made a glass that should be kept not perfectly horizontally, but slightly oblique so that it points towards the floor. This way, it is easy for everyone to install it in the wrong way.

Visuals

When talking about visuals, there are two questions that everyone asks:

1. Do the virtual elements seem real as in the video of the whale in the gym?
2. How is the field of view?

Let me answer both questions:

1. Forget about whales and elephants: the virtual elements do not seem real at all. They are just a little bit better than on HoloLens: they are there, in your environment, but they have some kind of transparency… they seem a bit like opaque ghosts in the room. This is a problem of the technology and Magic Leap couldn’t do anything to make it better. Furthermore, the computational power of the device only allows scenes without too many polygons, so you can’t run a high-poly photorealistic scene. Even here, anyway, we are at a level higher than HoloLens’s one, thanks to the NVIDIA Tegra X2 contained in the Lightpack, that is more performant than the chip contained in the old HoloLens. This lets you experience scenes that are a bit more complex than the ones that you can live on Microsoft’s device;
2. The field of view is 40° Horizontal and 30° Vertical. This means that it is slightly bigger than HoloLens, but nothing special. It is still like a window to the augmented world. But Magic Leap One is great in masking its limitations: more or less, when the FOV window ends, it begins the frame of the glass, and this triggers a nice mechanism in your brain, that accepts the fact that there are virtual elements only if you look through the glass. Let me explain that better: in HoloLens, you have a transparent glass all around your vision, so you can see all the real world, and in a certain abrupt region, the virtual elements pop up. This is annoying because there is this confusing augmented window out of nowhere. On Magic Leap, you are wearing glasses with a thick frame, so you don’t have all your vision. And the augmentations more or less happen only in the region that is inside the frame, so your brain thinks “what I see inside the glasses frame is augmented, what I see outside it is just the real world” and this makes more sense to it. When playing with the device, I had a sensation similar to the one that I have in HoloLens only in the lower part of the augmented window, that for my vision was too distant from the frame of the Magic Leap One and so the above considerations didn’t hold, leaving me unsatisfied about vertical FOV. So, FOV is not exceptional at all, but I appreciated the clever masking mechanism.

Multifocal display

Magic Leap uses a waveguide display, exactly the same type of HoloLens. And, exactly as in HoloLens, it is possible to view the lights of your rooms reflected with rainbow glares on the lenses.

The innovation is that it is a multi-focal display with two depths of field: near and far. The system works this way: your eyes are tracked continuously by the system and when it detects that they are crossing on a near object, it activates the near focus display and when they are relaxing on a distant object, it activates the far plane. Theoretically, it should mimic what happens in real life, so you should never notice this happening, it should feel natural.

I wanted to absolutely see it working, so in the end, I had an idea: while playing Angry Birds, I put the sling near my face, exactly on the line of sight between me and the constructions of the pigs, that were far away from me. This way I could choose to look at an object and see the other becoming blurred. From this experiment, I noticed that:

• Two focus planes are not enough for this mechanism to feel natural and we need more;
• When the focus-switching mechanism works, it is great, everything feels real;
• When the focus-switching mechanism glitches, so it selects the wrong focus plane or selects the right one but with a noticeable delay, it is really bad. My eyes crossed and my brain got crazy for one second or such everytime this happened.

Audio

Magic Leap One comes with stereo integrated audio. As I have said various times, I am not an audio expert at all, so I can’t judge about its quality. Anyway, it worked and worked well. There is also a 3.5mm jack to add your external headphones (like the ones announced at LEAP Con), but I haven’t tried it.

Controller

The controller is a 6 DOF remote, that somewhat reminds the ones of some VR headsets. But it is much bigger, it is more or less big as my hand. It features a touchpad, a button on the top that turns it on/off and also acts as a system menu, and two triggers, one for the index finger and one for the middle finger. It is tracked thanks to magnetical fields tracking and communicates with the headset via Bluetooth.

I didn’t like the controller at all. One of the main reasons is that the touchpad is not clickable, you can just use it to slide your thumb and nothing more and this is the opposite of what happens with all the XR controllers out there (Vive, Vive Focus, Lenovo Mirage Solo, Gear VR, etc…). So a lot of time, I tried to confirm stuff by pressing the trackpad, but with no luck and this irritated me a lot. Maybe creating an interface that was similar to one of the other XR glasses would have been a smarter choice.

The two other triggers are confusing as well because they seem to be there without a clear logic. I mean, on the Oculus Touch you have a middle finger trigger, that is used to grab stuff by closing your hand and the index finger that is used to trigger actions: when you use a VR gun, everything is more or less like in the real world. Here you have the two triggers the same, but because of the different shape of the controllers, they didn’t sound intuitive to me at all. During the experience, I never got how to use them at a first glance. And a UX should be intuitive… if I have to follow a tutorial for each experience, maybe there is something wrong with it.

Regarding the controller tracking, I haven’t found it exciting at all… sometimes I found the controller pointing at a slightly different direction wrt the expected one and during all the experience, I found it being tracked in a non-ideal way. I mean, my brain told me that something was wrong with it all the time, even if I can’t explain what (maybe a little lag, or a slight offset). I read that metallic objects can ruin the tracking performances… maybe there were objects causing interferences in the surroundings, who knows.

Anyway, at least there is a controller. With Hololens there is not, and air-tapping is a pain.

Tracking

The 6 DOF tracking of the headset worked fairly well in my experience.

When reconstructing the various environments, I was able to see the meshing system at work and I noticed that it was better than the one of HoloLens: the triangles of the mesh were smaller, the tracking was faster, there were fewer holes and the resulting mesh was also less chaotic. So, it is good, but don’t expect it to be perfect: the mesh had holes and had also an offset wrt the real world sometimes. For instance, when playing Angry Birds, I noticed that the mesh of the floor was actually like 5cm above the real floor.

The nice meshing is the reason why people are all astonished by Magic Leap physics system: if you let an astronaut fly in your room, he can fly behind your table and he gets completely obscured by the table. Then you go looking behind your table and you find the astronaut there, in the position you expected it to be. That works very well because the meshing is indeed good. On Hololens something like that was already possible, but the mesh was more rough and less adherent to the real world so that the illusion was broken more easily.

The problem of the tracking is that the positional tracking is a little more unstable than the one of HoloLens: virtual elements seem all to jitter a bit, and this is especially noticeable when you go closer to them. This is really annoying and ruins a bit the magic to me. With HoloLens, instead, virtual elements are really fixed in place as if they were real.

UX

Dear UX designer of Magic Leap, you had one job. And you did it bad.

The UI of Magic Leap One is all full of cartoons and nice drawings, exactly as its website, and all of this is nice and surely more good-looking and more-relaxing than the cold Windows 10 interface of HoloLens. And again, I appreciated the attention of the Florida company to the visual experience that the user should have, this is really great.

But the interfaces… oh, the interfaces are a complete nonsense: it is like every time there should be an action to be done, you have to spin the Wheel of Luck and pick the interaction that randomly comes out. If I have a 6 DOF controller in my hand, I expect to do everything with it, pointing at menu entries and then clicking my trigger to select the menu item. Instead, this doesn’t happen… ehm, no, I mean: sometimes it happens, sometimes not. When I triggered the Main Menu of Magic Leap One, I immediately pointed my controller towards “Dr. Grodborts invaders”, but I saw nothing happening. I looked at Gianni and he said to me “no, you have to swipe your touchpad to select menu items“. SWIPE MY WHAAT??? It is a circular menu in 3D in front of me, why should I swipe on a touchpad to use it??? I should point at elements! Then, in another experience, I selected a menu item by actually pointing at it with my controller, but then in the resulting submenu, I had strangely to swipe on my touchpad to select the sub-item in a completely incoherent way from what I just did before.

via GIPHY

Gianni said, “eh, you have to get used to this interface”. No, I don’t have to get used to anything: a UX is like a joke, if you have to explain it, then it is not good.

What really made my cry has been the use of head tracking: when you select an app in the initial menu with your controller, then the system shows you a 3D gizmo of the app that you have to position in your room USING YOUR HEAD. So, you grab things with your controller, but then you have to move them with your head. Seems logic, no?

(Image by Know your meme)

I got so frustrated by using the UX system of Magic Leap, together with the flawed controller design, that in the end, I was on the edge of throwing angrily the controller on the floor. This has been the thing that ruined my experience the most: more than FOV, more than polycount and all the things that we have all talked about in the last months.

Apps

I’ve tried a few apps:

• Dr. G Invaders: it didn’t work because the room was too little. Dear WETA, in Europe we don’t have all the big houses with giant living rooms that you have in the States, so please let us play your game anyway;
• Create: I found Create to be a very enjoyable app. You can add 3D cute characters or objects to your room and see them interacting: you can add corals to your floor, you can add an astronaut flying in your room and so on. It is a nice sandbox, that seems like an evolved version of the Hologram app on HoloLens;
• Tonandi: a musical experience that made me completely bored. I know, everyone else is praising this app, but I just saw some nice eye candies while a music I didn’t like was playing in my ears… I expected something more. Some visuals are nice, though;
• Wayfair: you can see some 3D models of rooms with furniture and move the pieces of furniture. How amazing… I closed it after 40 seconds, I guess;
• Angry Birds: the experience I liked the most. It is the classic Angry Birds experience, that is of course very enjoyable, ported to your room. The graphics are nice, the gameplay is entertaining, and everything has been designed to fit quite well in AR;
• Helio: it is the 3D browser of Magic Leap. Before using it, I had to follow a tutorial (-_-), because even here the UX is completely unintuitive. also because the mechanic to take 3D objects out of the web pages, is another UX mess. Taking a 3D object out of a webpage was an interesting mechanic, but the actual implementation ruined it. Furthermore, it is not something original to Magic Leap, since I’ve seen Microsoft doing it in VR some years ago. I’ve been able to extract the swimming whale out of a web page, and I have to say that it is absolutely not like the one of the promotional CGI videos, but it is any way beautiful to be seen. From videos, it seemed terrible, while actually, it is nice, especially because it is something very big that you see in your room.

Problems

There are some things that further impacted on my experience:

1. The glasses got warm during the time I used them, and I started to feel a non-pleasant warming sensation on my forehead;
2. The headset OS didn’t appear as stable and sometimes the software glitched or halted for some seconds;
3. The computational power is bigger than HoloLens, but not that big enough and when there were too many objects in the scene, the OS lagged a bit;
4. I felt eye strain really quickly. HoloLens makes my eyes dry too, but not this fast and not this much;
5. I had motion sickness. I mean, I’m used to playing games with standard locomotion and I love VR rollercoaster… and this thing gave me motion sickness, that is something that in room-scale AR should not happen. I thought that was just nausea given by something I have eaten, but Gianni confirmed me that he had the same sensation when he tried the device. That’s really bad.

Other considerations

There are hands tracking and eye tracking, but they are almost never exploited.

Final impressions

Magic Leap is 不怎么样, it is a not so special device. I think that in the end, Palmer Luckey was right:

It is slightly better than Hololens in some ways, slightly worse in others, and generally a small step past what was state of the art three years ago – this is more Hololens 1.1 than Consumer AR 1.0.

As someone that has already used HoloLens for some projects, I didn’t find Magic Leap One either magical or a leap at all. It is more like a HoloLens Pro. It has better FOV, better computational power, better environment understanding than HoloLens, and it also has acontroller, hands tracking and eye tracking. But it has also a worse tracking stability, a less coherent UX and an unriper ecosystem. I think that the definition of HoloLens 1.1 is a bit too critic, maybe HoloLens 1.5 gives more justice to Magic Leap. Anyway, after I tried it, I also found it more underwhelming than how I expected it after having read all the possible reviews about it. And my ex-partner agrees with me. I think that some key issues, like the bad controller tracking, the bad UX and the motion sickness can really hurt the user experience.

On the other side, I think that it is also remarkable that a new startup has been able to create such a product coming from nothing. Yes, they got billions, but managing money is anyway a hard task, especially if you have to fight against Microsoft and Facebook in the AR realm.

And controller apart, the device itself is not bad. It is remarkable to say that most of the problems that I found (jittery tracking, terrible UX, mediocre apps, etc…) can be fixed in future software updates. The impression that me and Gianni got from these tests is that the device features so many gross errors (like the ones of the UX), that it seems having been rushed out for the release, maybe because of investors’ pressures or because they knew they had to release something before the next gen of HoloLens could be announced. It was not ready, but it was released anyway because waiting for it to be complete would have taken too much time. The good news is that these issues can be fixed, but the bad one is that it will require months, months in which it is difficult to say that Magic Leap is a wonderful device.

The fluff of the company regarding a mind-blowing device that makes you feel virtual elements as real is all fake: this has only been useful to create the hype to sell more devices at launch. But the headset per-se is a nice AR devkit, that lets you buy a device that is a bit better than HoloLens for a slightly lower price. But take in mind that to make a comparison with VR, maybe we are at Oculus DK 2 stage, where everything is a mess, there are few apps, the emulation of reality is mediocre and the UX is terrible.

Should you buy it?

I have mixed impressions on Magic Leap. For sure it is not a disruptive evolution of AR/MR, so there’s no need to rush to buy it, but it is also a nice devkit. Personally, I’ll continue waiting some months to observe the evolutions of augmented reality, both on the smartphone side and on the glasses side, where in the next months we should see something coming from Microsoft, Samsung, and Apple. This will let also give the time to Magic Leap to fix all the software flaws of its device. So, if you are not in a hurry, my advice is to wait.

Anyway, if you want to buy an AR headset to start experimenting with spatial computing, Magic Leap can be something that may interest you, because it is a good devkit and costs less than HoloLens. Furthermore, the company is eager to find developers, so you will have all the possible support.

If you have to work for an external enterprise customer, instead, at the moment probably I will still advise using the HoloLens, because the ecosystem is more complete, there is behind a major company like Microsoft that we’re sure that will be there for the upcoming years, there is a clear business licensing model, there are also tools for industrial environments, a full-fledged SDK and such. It’s a safer choice, IMHO.

Other resources

This has been just a brief hands-on session. If you want to read a more complete review, I advise you to read the one of Lucas Rizzotto and to watch the video by the TESTED guys.

I hope you liked this little review of mine and that it has clarified a bit your ideas regarding the Magic Leap One… and if this has been the case, please share it on your social media channels and subscribe to my newsletter to keep updated with AR and VR articles like this one 🙂