The Magic Leap One Meta Impressions: They have the best AR on the market

I don't have a Magic Leap One. Part of the reason is because I am in the UK. The other part is that I don't have $2300 to spend. So I can't really give you my review of the device. But I can take all the impressions that others have done and mash them into a meta review. 

There have been a number of different perspectives on the Magic Leap One and I think it is important to split them apart. 

The first perspective is the least informative. These come from the quick hands on from consumer outlets. This includes CNET, The Verge, Wired, etc. They produced hands on articles from limited time with the device from the perspective of their readers who are generally consumers. The Magic Leap One is not a consumer device. The content, function and name reflect that clearly. Reviewing it through that lens is always going to leave something to be desired. But I don't think these hands on are useless. They give a good insight in to what the consumer device might be like. If you want an overview of these reviews, check out this round up by Tom's Guide . The TL;DR here is that they all generally acknowledge that it is the best AR around but it is not fully baked (hence dev kit) and not as massive a leap forward as the (impossible to live up to) hype claims.

The other group of impressions are from developers who have spent time with the device and are using it as it is intended. To create something. These reviews are what we should be looking at to see the potential in the device and to confront the ML1 on it's own terms. I'm going to be talking about them.

Please click through the links to the actual reviews. These people put a lot of effort in to writing these and my small summaries do not do them justice. 

From the developers

The TL;DR here is that developers love this device. It is the best AR platform currently available. But it isn't perfect. It is very much a Gen 1 device. It can be a bit buggy, some features are not fully baked and there is a ton of room to improve things. Never-the-less, it is a success at what it intends to be. A creators edition. 

Magic Leap One First Hands-On Impressions for HoloLens Developers by Bryan Crow

nr.png

This review is from developer Bryan Crow who has spent significant time with hololens and is therefore one of the more informed reviews out there. He is generally very positive about the device claiming better optics, faster, better meshing in most cases, better controller and better gesture support when compared to hololens. He is also excited about eye tracking possibilities while being lukewarm on position tracking which he claims is on par with ARCore/ARKit and not as good as hololens. Hololens is the best in the industry at this problem and seemingly hard to catch.  

While Bryan does't provide a good quote to sum up his feelings, the same publication (Next Reality) has later said:


Let's get right to the meat of the matter: Magic Leap One is not a flop. Not regarding sales (the target, for now, is developers), and not in terms of performance and overall experience. This is the best augmented reality I've seen, full stop.


Lucas spent a long time working on this review and is one of the most in depth I've seen. To cut to the chase:


Magic Leap One is an ambitious, well-made, but imperfect MR devkit that doesn’t quite live up to the hype, but is still the most complete and affordable mixed-reality (MR) computer out there.


The (impossible to live up to) hype aside, Lucas clearly likes the device but notices many issues that clearly peg this a dev kit. Issues include resolution (which most others praise so he is somewhat of an outlier here) and occasionally laggy tracking of the 6DOF controller. He is also concerned that the OS leaves you "lost" sometimes and that it was designed with a 3DOF controller in mind so it doesn't take full advantage of Magic Leaps feature set. Very few use eye tracking for example. He sees this as an opportunity as "Magic Leap’s developer community will likely outmatch Magic Leap creatively and become the real metric of what this device can do". So if you are a dev, you can really shine here.

Interestingly, he doesn't see field of view as a problem saying it is the best in class and that clever design really minimises its importance. He stresses that designers need to "create mechanics that leverage it while also hiding its limitations".

Tested: Magic Leap One Augmented Reality Review!

While not developers, I think the folks at tested fall nicely under the creator category. They really love the device. 


We think this is the most impressive augmented reality device you can buy right now


They were particularly impressed with the control. It is 6DOF and tracks all around your space since it is magnetically tracked as opposed to optically tracked like most other 6DOF controllers. The device does not have to see the controller to track it. They do say that it isn't quite as accurate as an IR tracked controller but it works far more consistently than an optical controller particularly without having a lighthouse set up like Vive.

They love ergonomics. Claiming it is lighter and better that the Oculus or the playstation VR headset. 

One thing they noted is that while Magic Leap claims it will remember your space and not need to map it again, it rarely could figure out it was in the same space. A number of other reviewers have noted this as well. This is something they can hopefully improve as that persistence is going to be very important to the consumer release. 

Tested had the best look at how Magic Leap does variable focus. They found two separate focal planes in the device. This is something that very few other devices do and it does work. But for this to be more impactful they need more focal planes than two. Not only that, the near clipping plane is too far away and meaning you can't get too close to an object. I personally believe this is because they don't have a third focal plane for near objects. We'll have to see what they can do with the consumer release as this is where variable focus really shines. 

Tested says that the spatial positioning is great. It is responsive in real time for both mapping and positioning.

All in all they were very impressed with the device but were very clear it is still early days for AR.


This is a great product for what it is but it isn't the dream of AR... We are not there, we are a decade away from that


Various Twitter Impressions

Not all creators take the time to write out long form reviews but many have taken to twitter to get how they feel about the device across. I've been collecting impressions for the past few weeks and frankly I can find very few negative reviews. It is always hard to get a full picture from twitter but I have tried to search "magic leap" as a keyword and periodically go through the list. I'm sure I have missed some but here are the impressions in any case. 

A theme seems to be there were many Magic Leap skeptics turned converts. This bodes very well for the device.

Luckey Palmer

There is always one hold out. Luckey does not like Magic Leap.

I think it is important to point out that Rony Abovitz has been taking shots at the VR industry for years. Silly and unreasonable shots at that. So there are a lot of reasons for Luckey to not like Magic Leap outside of what the device does. Putting Luckey's questionable morality aside, his impressions actually line up pretty well with the many other devs I pointed out above, with some exceptions. The big difference is the scale of the issues he sees. He seems to think that every little problem with this Gen 1 device is fatal while others see the potential. Sounds like an early VR device I know...  

He points out that the controller tracking is a disaster but the folks at Tested said the exact opposite, praising the controller. Other devs have seen some issues with tracking at times but the overall impressions seem to be that it is a decent controller with pretty impressive tracking. This is particularly interesting given that magnetic tracking in a form factor such as this is mostly unheard of. This is one of the first times someone has tried it. There is likely ample room for improvement.

He mentions the lenses are dark which others have pointed out as an after thought. Not one review thought this was a tragic problem. Many claim it helps with the visibility of the digital objects and is a net benefit. 

He is critical the OS which many other devs have praised. He seems to miss the point of the Lumin Runtime which he claims are are just application windows on the wall. Magic Leap has stressed that making square windows is actually an anti-pattern. The Lumin Runtime is about persistent applications that are specifically not just "Windows 8 style application windows". I'd be curious to see how he thinks computing should be done in AR if not done in this way. 

He also says things that are just flat out wrong. For example, he talks about the time "back when they were still hyping up scanning fiber displays". Magic Leap never talked about this publicly. Not once. We only knew they were working on this technology because of patent applications and leaked material. This is not something they were hyping and Rony has said that they would have preferred if nothing about their early days was made public but patents are public and funding is public so they had no choice. 

It makes him sound like he very much wants to hate this device and through that lens every small issue turns into a disaster. He updates his impressions by pointing out a reaction from Rony. Rony seems to be be (very obtusely) pointing out the negative bias that Luckey has while taking more cheap potshots at him. It is clear that there is a personality clash here. 

The whole really sad given Luckey's history with Oculus. It would be nice if they could support one another. I think it was best put on Twitter from an early days Oculus supporter:

In any case, no one can say Luckey is not experienced in the field. His points do have merit. Given others impressions, I'd say read Luckey's impressions but take the doomsaying down an order of magnitude. That brings his impressions closer to the mean and you can pull out reasonable critisisms of the device.