Hacker News new | comments | show | ask | jobs | submit login
Magic Leap is real and it’s a janky marvel (techcrunch.com)
100 points by allenleein 6 days ago | hide | past | web | favorite | 65 comments

> its first product is a somewhat janky piece of magic

> Magic Leap is about as impressive a piece of augmented or virtual reality hardware as I’ve seen. Other companies may have better fields of view and a more compact device, but they lack the variety of content that makes Magic Leap’s offerings shine. The early partnerships the company has inked have, indeed, paid off.

the author nearly breaking their neck stretching to find some positive way to frame this train wreck of a company/product

saying the hardware is the most impressive they've seen, but because the software from partners is good?

HN needs a moratorium on TC paid advertising pieces / TC in general

You’ll actually be hard pressed to find any press anywhere that thinks the hardware is bad.

> Other companies may have better fields of view and a more compact device

it's right there in the original review my friend, even the paid-for TC journalist has to admit the hardware is inferior and "janky"

Who on Earth approved this trailer? https://www.youtube.com/watch?v=45ZHNq9_7eY

"Ha ha, you'll be such a dork with our product on your face!"

That does not seem a wise marketing line.

There are two ways a marketing message can deal with a widely held negative conception about the product: ignore it, or try to convince people it's wrong. The former is usually the safer course. Trying to convince people they're wrong is really hard, but if you can pull it off, it can pay off big.

Pepsi's "Pepsi Challenge" (https://en.wikipedia.org/wiki/Pepsi_Challenge) campaign from the 1970s and '80s is the canonical example of how to make an "actually, you're wrong" campaign work. Pepsi's market research found that the prevailing opinion of Pepsi was that it tasted too sweet, so people turned to Coke instead, just assuming that they wouldn't like Pepsi without ever really trying it. So Pepsi filmed commercials showing people taking a blind taste test of both Pepsi and Coke, and preferring Pepsi before being told which was which. The commercials worked -- they convinced people to give Pepsi a try, and enough of them found they liked it to drive up Pepsi sales for good.

The "you'll look like a dork" fear strikes me as widely enough held to be one of the biggest challenges VR and AR have to overcome to break into the mass market, so it's not illogical that at least some marketers would decide they have to take it on directly.

Part of my point is that I don't think they pulled it off, not even close. I don't think the storyline was "Ha, he looks like a dork, oh, he sure showed us!" it was "Ha, he looks like a dork... wtf just happened to reality?" The attempt at a humorous payoff at the end fails to address the dork aspect. "I guess he wasn't a dork, he really was fighting real robots"?

You are right. It was not a good ad. Felt very awkward and forced.

Similarly the trailer shows one of the device's biggest flaws at 0:42-0:44. The limited FOV of AR devices breaks the immersion in a way that the limited FOV of VR devices doesn't. Any augmented object disappears from view if you aren't looking directly at it, but unlike with VR you can still see that area of the real world so the missing graphic is immediately noticeable. So complements to the marketing team for being truthful in how people will react to the device and its true capabilities, but it is just surprising that this type of thing was approved.

I'm surprised they'd use a single-player game for a trailer. My first thought for concisely showing the benefits would be something more like a Mario Party-style game, with four people but split across two physical locations.

The "game footage captured on device" is super misleading, same thing that Hololens did. The field of view is actually a small portion of your entire display, it doesn't cover the entire lens.

Ahhh but they are 'converted' by the end; which is presented as a simile on how realistic the gameplay is.

> Who on Earth approved this trailer?

Ha, if you're surprised about that, check out what they approved here https://www.youtube.com/watch?v=w8J5BWL8oJY

reminds me of the first cars i'd ever seen with massaging seats at the north american international auto show.

Less of a massage, more so the sensation of sitting on some large, undulating, lovecraftian leather mass. There was the accountable whine of a team of motors, and once every few minutes a sharp audible pop could be heard. God bless Mercedes for trying to cover this all up by blasting classical through the car speakers though.

In my work with the HoloLens, I found that it's really best when you're augmenting the real world rather than displaying generic graphics. I know that sounds kind of like a tautology "use an augmented reality device to augment the real world?!? What a novel idea!" But if you look closely at the extant apps on HoloLens, you will realize they are mostly VR applications that happen to be rendered in your living room. To wit, the content is not connected to the space in any way other than proximity. You can move the content somewhere else and it's the same content.

Where I've found AR to really shine is in IoT settings. Take the real world with smart objects and add relevant data on top of them. That's the easiest thing to do, as the data is readily available and the UX is much more obvious than other potential use-cases. The real world becomes an anchor for the content and makes it much more believable, even when you lose the augmented graphics in the periphery as you move around.

There may be other ways to augment the real world other than IoT. GIS and routing might be one, though the tech isn't solid enough yet to put in a fast moving car, on roads with wide varieties of weather conditions. Medical assistance (though not training, again, that's VR). Logistics and shipping--previous experience working in a grocery warehouse, I saw that new pickers struggle the most with finding things, and the most successful pickers have memorized the locations of things.

ASIDE: I'm also very interested in augmented audio interfaces, spatialized audio through bone-conducting headphones, for a much better virtual assistant experience. I have two small kids and use Google Home a lot while I'm getting them ready to go somewhere or making them dinner. Having different reminders and features located in different parts of the house, with proper fall-off of volume, would be a huge improvement on our daily routine.

What does this have to do with Magic Leap? I would be very happy if all that Magic Leap had to offer was "slightly better HoloLens". The HoloLens has been underwhelming for a lot of people, but I think even today at basically 5 years old, it still has a lot of untapped potential. ML is slightly cheaper than HL, which is always nice, and it's running an Android-like OS, which might bring more developers along (though personally I very much enjoy UWP on both desktop and MR Windows). We are so close on so many different concepts that incremental improvements can't come fast enough.

> The HoloLens has been underwhelming for a lot of people, but I think even today at basically 5 years old, it still has a lot of untapped potential.

I was able to play with a Hololens a few years back and although very impressive, the field of view was the showstopper. On the plus side, its spatial tracking is impressive.

The Hololens viewport is equivalent to a 27 inch monitor held at arm's length whose position is locked to your head and not your eyes which move far more than your head. So you wind up moving your head around a bunch with this heavy headset. Talk about fatigue. Then add to that you can't walk up to something to get in close for a look because the viewport is at arm's length, anything in virtual space between your eyes and the viewport is out of frame and invisible. So you can't lean in for a closer look. After a while of wandering about while gyrating your weighted head around, you quickly tire of it.

AR devices like these are pretty damn impressive and quite useful beyond entertainment. But they need to shed some weight, increase the viewport and therefore the immersiveness. Though, the holy grail and killer feature they really need is eye tracking. The key is to eliminate that awkward disconnect between the virtual world and reality by keeping things in line with your eyes and not your head.

> The Hololens viewport is equivalent to a 27 inch monitor held at arm's length

You must have really long arms. It's a lot worse than that. But that's why I said it's best when you augment real world objects. The focus becomes the object, it maintains the object-permanency on its own, the graphics come along for the ride.

> Then add to that you can't walk up to something to get in close for a look because the viewport is at arm's length

That is a per-application choice, though Microsoft does officially recommend a near-plane of no less than 1m. I've personally been able to make experiences that were comfortable at 0.5m, if the graphics were small, and again, anchored to some real-world object.

Yes, better devices will make better experiences. But we have the HoloLens (and Magic Leap) right now. And clients are paying now. There is work that can be done.

This was about three years back I think. The measurement was a rough estimate made by walking up to a 27 inch imac until the view port disappeared which was roughly the size of the 27 screen. Perhaps I was being generous.

I would love an app that would render a 3d model in the real world. Imagine looking at the site of your new house with the latest plans from the architect and able to visualize exactly how it would look in it's surroundings.

The Ikea app can do this for furniture. I know hololens can do this for larger objects but I don't know enough about its "app store" to know if there's one for architecture. http://elevr.com/experimental-still-lifes-and-landscape-inte...

I recall seeing one, once, but I think it was just someone's college senior project. It looked very compelling.

Reminds me a lot of the introduction of the Iridium phone.

"Worldwide coverage! No more dead zones! This will change communications globally!"

Then they saw the phone. And its antenna. And learned it only worked outdoors. And cost $2,000.

~$6B in satellite costs.

It wasn't a failure (I still use mine once in a while), but it sure wasn't what people were expecting.

The second corporate incarnation of Iridium, which bought the whole network out of bankruptcy, has been extremely successful, however. It's the only truly global network that will function equally well in mountainous regions of northern Afghanistan, on the Antarctic polar ice cap plateau, in the middle of the south Indian Ocean, or elsewhere. The architecture and technology was way ahead of its time, with Ka-band satellite-to-satellite trunk links between satellites following each other in polar orbits. And voice codecs built with 1996-1997 technology that can fit intelligible voice into about 3000 bps.

I agree that the late 1990s marketing which was trying to sell Iridium phones to international business travelers was ridiculous, however. Once the new Iridium corporation settled on the actually useful market of government, military, disaster response, maritime, expeditions and M2M data they've been wildly successful.

The second and third generation handsets (9555 and Extreme) are much more reasonably sized and closer in user interface to a modern cellphone.

> It's the only truly global network that will function equally well in mountainous regions of northern Afghanistan, on the Antarctic polar ice cap plateau, in the middle of the south Indian Ocean, or elsewhere.

Interesting. I only know a little about the market and products and always have been curious (and on the occasions I needed it, good info was hard to find): What about names like Intelsat, Inmarsat, Globalstar ...?

Globalstar is an epic turd, network architecture wise.

Right! Nothing like buying an already-deployed $6B satellite network for pennies on the dollar to bootstrap a successful company ;)

When iridium started, cell coverage was just about nonexistent. In the time it took to go from idea to launched satellite networks, cellular expanded so much that the opportunity vanished.

If they had gotten things running just a few years earlier, it may have had a chance. Probably not a great one, given the slowness of iteration cycles of satellites vs terrestrial.

But is Techcrunch real anymore? 5 comments on their article and 4 of them are spam?

Not really, they are just another PR company at this point.

Are journalists that were given early previews released from their NDA obligations? I'd be really curious to hear some of them explain if the released product is indeed what they were shown.

so, it's a more compact HoloLens, with better content/games?

I dunno, would you describe every AR platform as "Hololens but a little different"?

One thing that got briefly mentioned in the article is that the Magic Leap has an external computer, which you can put in a pocket or clip to your waist. It runs hot and it might cook itself if you block the airflow to the fans. Also, I've never heard of anyone needing to wander around for 4 minutes before the Hololens started tracking properly. On the plus side, Magic Leap has two different focal planes where HL only has one. Apparently this helps a lot in making users feel a sense of depth to the images.

You definitely have to wander around for Hololens to map the room in some cases. Have you used one?

And this is definitely very similar to Hololens. In fact the main difference seems to be the field of view. Other than that, and the fact that you wear the brains on your hip instead of your head, how is it any different?

Well that external computer seems like a pretty different user experience to me. And having two focal planes can make a big difference. And no I haven't used either system.

I guess what I'm asking is, can you imagine an AR platform that you would consider to be different? Or will you describe every AR system from now on as "very similar to HoloLens"?

Yeah I probably would because really Hololens more or less jumped to the endgame. The only thing that really needs to be improved in it is the terrible field of view.

Everything else was very good - tracking, image quality, room scanning, etc.

Oh and I guess the input system sucks. That's one other big difference of Magic Leap.

The computer being on the hip should make the headset much lighter/cooler than the hololens, which should make it less tiring to wear for longer periods.

>so, it's a more compact HoloLens, with better content/games?

No wireless. Less space than a nomad. Lame.

People actually bought the iPod.

Looking back I wonder if iTunes might have been the secret sauce for the iPad to work, rather than the iPod itself.

For sure. I had a Rio player and an iPod. iTunes + iPod = music search, discovery, purchase, download, and transfer to device, all at once. For the Rio player I had to do all those steps manually.

3DS did this in 2011 and it was better, too.

Faceraiders was great! It's a shame it didn't seem to go much of anywhere though for later games, it's a shame it and the Vita's more odd features didn't get more use.

That demo did look janky. Still promising in my opinion.

Paid PR piece, anyone? And still the writer is reaching so far they're going all the way around the world back into their own ass trying to praise this thing.

<rant> I'm so fucking done with MagicLeap. Their head conman CEO, who is, absolutely and by all accounts, an absolutely fantastic showman and salesman, has taken so much money and so much hype that it's not going to end well.

Pretty much EVERYONE who has tried the magic leap headset has basically said 'meh'. It blocks around 85% of the incoming light [1], for fuck's sake. That's the first of a very long list of 'meh'.

Why am I so pissed off about this? Because it's going to crash the AR industry (already on its knees because of an overhyped VR industry) and set it back years. Already, and for years now, no one wants to invest in an AR hardware startup because "how could you possibly beat Magic Leap?

Fuck those guys. And the dumb investors who let themselves be fondled into throwing billions? Shame on you! </rant>

The same thing happened to lightfield imaging with Lytro (who took "only" a few hundred million in comparison). Lightfield imaging has been stagnant ever since they took too much money and sucked all the oxygen out of an entire industry.

[1] https://www.kguttag.com/2018/09/26/magic-leap-review-part-1-...

I completely get your rant, but I don't see it the same way with the VR stuff. There's REALLY good VR out there... I have an HTC Vive Pro and one of my primary uses of it is for training as I'm currently in flight training, and the immersion that I get using flight simulators in 3D is just stunning. When I'm not doing that, from exercise games, to social games, to first person shooters, I can't think of playing video games not in VR.

But between the headset, computer, and everything that goes along with it, I've spend > $4k on this setup. So I get it that this is not the typical VR experience for everyone, but the issue in this space is one of price. I can't believe that if that experience could be had for, say, $500, that we wouldn't be inundated in VR stuff. Contrary to what I've seen in AR; even the top of the line stuff is still just kinda "meh"

What flight simulator would you recommend I use it with?

Depends on what you are looking for. For me, systems fidelity and ability to custom design scenarios is most important, so for that I recommend Prepar3D (Lockheed Martin bought FSX and kept developing it).

Even multi-million dollar level D simulators don't quite do a good job at recreating the flight control feels, so I've given up on that altogether. But some features of the flight model --- things like P-factor, adverse yaw -- are better modeled in X-plane, and visually (at least out of the box), it beats the competition.

But the most immersive experience has got to be Prepar3D coupled with payware aircraft from PMDG. I know real Boeing airline pilots who actually have used it for training, and vouch for how well the systems are modeled.

It really depends on your interests and skill level. There are now several with good support. There's some that give you access to flight controls by using existing VR controls and are generally a jump in and fly experience and then at the other end there's stuff like DCS which will require some additional hardware and a fair amount of learning (but totally worth it, especially with the Tomcat on the horizon)

I think there is plenty of money being spent on AR R&D right now. We are pretty sure Apple is. Michael Abrash discussed it during his Oculus Connect keynote a few weeks ago. Snapchat has certainly been public about what they want to do. This is the post-phone platform and it is really, really big. A few billion up in smoke isn't going to stop it.

The crime here is Magic Leap hyped the hell out of their product, took tons of money, was extremely opaque about it, and then severely under-deliverd. Exact opposite of Valve and Oculus.

Oculus could have shipped CV1 a couple years sooner, it is essentially identical to DK2. There was a lot of good will (and developer energy) squandered during those two years. To give just one example of where the major players have misstepped.

> Why am I so pissed off about this? Because it's going to crash the AR industry (already on its knees because of an overhyped VR industry) and set it back years. Already, and for years now, no one wants to invest in an AR hardware startup because "how could you possibly beat Magic Leap?

Found the guy who read the Palmer Luckey "Tragic Heap" article.

I'm not allowed to give details, but from my own perspective, neither the VR or AR market are as fragile as you make it out to be.

> already on its knees because of an overhyped VR industry

Is it though ?

It looks like a product that is on the cusp of breaking into the mainstream gaming market.(barring 1 or 2 big pain points)

Movement and Full wireless capabilities (in a desktop grade headset, not Quest) are the 2 big pain points for me.


I do fully agree with you on the Magic Leap though. I find it unbelievable that it received the amount of funding that it did.

Have you actually tried it? It’s mind blowing. Way better than I ever expected it’d be. To be fair, I haven’t tried any other AR system.

By many accounts, Hololens is better in most respects, and it's a few years old!

Agreed 100%.


Full disclosure I'm a bit of an MS fan boy as they have paid my bills for going on a decade. That being said, I want a solid AR product more than I want my team to win [so to speak].

By comparison, I tried Hololens in a decently controlled demo and was blown away. The idea that Windows 10 UWP apps would 'just work' also seemed to promise an immediately useful product, and mature dev pipeline. Then I started seeing fluff pieces for Magic Leap which were also hit pieces against Hololens. This was when Magic Leap was little more than a gobble of wires in a basement and a demo no one was allowed to see. It made me mad but really just confused me.

Fast forward. The Magic Leap launches and it's lacking against most of Hololens specs and experience. The worst part of _that_ is it doesn't force MS to innovate to compete on price/features. Which makes me afraid that MS will keep it priced for the industrial clients that are currently adopting it and not deal with this nonsense in the consumer space and who can blame them?

Now a new round of paid pieces and promises of content contracts are rolling out. Even more upsetting cause when you _really_ dig into the details the _honest_ answer from anyone seems to be that Magic Leap is not really for even alpha/beta adopters.

Hololens 2 is supposed to be improved across the board and add a 70 degree FOV. Not perfect but a step towards the 130 degree sweet spot I'm waiting for.

Folks smarter than me explained that Hololens locked the FOV to 35 degrees because the math it's expotentionally harder when you start dealing with the natural curve of the eye + the curve of the headset lens + real world overlays. So I get it's a tough problem to crack if you don't tether it to an i7 gaming rig like the Meta 2.

If Hololens can just get down under $1k I think MS will own the market for awhile. _If_ they don't drop the ball....something they do with such frequency that it makes me wonder if the Illuminati is real.


Do people really hate Microsoft _so_ much that they'll let someone defraud the industry, and pay for pieces that are on the verge of lying just so they can support something else?

This cost 2.3B? 2.3B?! Its projectors and a couple of wave guides glued together. You could go to the moon with that cash.

Capitalism needs a reboot.

No, it costs $2,300. They have developed the tech to do multiple stacked waveguides (12 I think?) but this device only does 2, probably for price reasons. Anyway the company is valued at $4.5 billion which has nothing to do with how much it will cost, only the amount of profit they think they can make from devices and patent licensing.

I think you misunderstood what he was saying. Magic Leap, as of now, has received $2.3B in the comments[1]. This is mentioned in the article's subtitle.

[1]: https://www.crunchbase.com/organization/magic-leap

Yes but what they're buying is a share of the company and its profits. No one is paying $2.3B to develop the tech.

Not for price reasons. Its because the more wave guides you super glue together the more image disruption you get. It's allready blocking a significant proportion of the light, not to mention the splitting out of colors.

Any idea what they are doing differently from the Vuzix Blade?

Vuzix Blade is an "eye-mounted tablet". It has a basic orientation sensor in it, but it's really bad and most of the apps I've seen don't use it. You've basically got a low-power Android device in the corner of your eye, with an interface designed around three buttons.

It has a front-facing camera, but it's used strictly as a web-cam for "see what I see" calls and QR code detection. It does not do any environmental analysis.

About the only things they have in common is "they look like goggles".

> You could go to the moon with that cash.

Not quite, but you could get low Earth orbit.

To compare, ST3, the light rail project in the Seattle area, is costing about $54 billion. That could take 1 WA State resident to the Moon and back.

Other similar devices (Hololens, Oculus Rift, Google Glass, etc.) have had similar costs. Remember that not all of the costs are hardware design or even production, they need to build a developer ecosystem, software, do marketing, etc. The costs involved in bringing a new hardware device in a new category to market are much higher than a CRUD web app... CastAR raised "only" $15M, which is enough for some software startups to operate for years, and crashed and burned.

Hell, you could go to Mars with that cash! After all, a rocket ship is basically just a tube with fuel strapped to one end and people strapped to the other.

Right! And a computer is just a pile of sand in metal casing.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact