After all this time (and money) this is really the best they can do? This has to be some sort of early prototype thing. Surely they will never ship this to market in it's current form.
I think the FOV estimates are pessimistic by a good margin, your eye looks around edging very well, as you can see in binoculars and other low FOV devices.
The rest is absolutely spot on though, without IPD adjustments any head mounted display is going to have real problems.
I'm skeptical of the analysis because it seems the author has a singular fixation on constantly criticizing ML, and apparently works for competitors.
He may be right, but whether or not ML1 is a success really won't depend on whether or not stuff like the original concept of fibre optical scanning light fields will work, the ML1 will succeed or fail based on consumer experience and content availability. And I really don't think this Kremlinology-based approach could capture it anymore than people speculating on what the Apple watch or iPhones would be like based on photos of the device and patents.
I suggest you spend some time reading his pretty extensive set of high quality articles on many companies and aspects of AR.
This is an industry where outsiders have a very limited understanding of the serious serious problems in front of AR.
It's also an industry that is taking advantage of the very limited understanding of just about everyone not in it to hype things far beyond their capabilities.
If you care about AR in even an offhand way you should spend some time self educating and Karl Guttag's site is a good way to get started.
It's very interesting field but there are tremendous challenges in front of it and it serves no purpose to just assume the companies involved have actual magic leaps that solve the problems.
It's also entirely possible they offer partial solutions to some problems, no solution to other problems, and still produce am experience worth paying money for.
That's a natural argument extrapolating from the phone, electronics and laptop market.
I think it is wrong in this case, however. As I understand it, the problem is "half-working" in the case of VR/AR/etc isn't something like less convenient that early adopters can simply put up with. Half-working VR has ability to sicken a person and injure the vision system.
That is a pretty bad article about VR that might have been excusable in 2013 but is not on firm ground in 2017 much less 2018.
Broadly, almost everything in it has been either not born out (vision concerns) or less of a problem then first assumed (motion sickness) once the limitations of different users were understood.
Many people using VR find that the vestibular concerns are not an issue over time as long as the framerate of the system is high enough (90fps+). People actually do get their "sea legs" over time and many (largely gamers so far) ask for locomotion systems that were considered terrible ideas even 2 years ago.
I've read his site, but he focuses way too much on aspects of AR that have little to do with consumer needs or concerns. A "worse is better" AR display that violates pretty much every concern he has could win the market as long as it was done really well in terms of comfort and applications, the same way the limited Apple Watch pretty much dominates the market today, even though it falls far short of the kinds of magic people thought it would be capable of originally.
> whether or not ML1 is a success really won't depend on whether or not stuff like the original concept of fibre optical scanning light fields will work, the ML1 will succeed or fail based on consumer experience
You can't separate "consumer experience" from the technology like that. This is a case where the technology tightly constrains how good the consumer experience can possibly be. The consumer experience absolutely depends on whether fiber scanning displays or other new technologies can work, because tha changes the possible form factor and capabilities of the device.
I worked at Intel in 2010 for a summer. I remember going to a presentation about why the iPhone was so good.
It was so good because they figured out how to do more with limited hardware. As an example, if you wanted to see an image on the iPhone, apple stored a version with just as many pixels as the screen, the windows treo pulled the full image. The difference was that one felt snappy and one felt slow.
My point is that your comment sounds like what everyone said back then... you need more or better hardware. Maybe, maybe not.
And yet this analysis from someone who hasn’t even touched the actual hardware or even has specs to base his analysis on is painting a completely different picture than the people who have reported on it since December. People are right to be skeptical of this just as they are magic leap.
I think the AR experience will be far more dependent on rock solid tracking and pose estimation, ergonomics of the headset, fashionability of the headset, ability to work in many light conditions, and killer initial apps, than display resolution or FOV.
Focusing too much on the display itself completely ignores the other aspects that inhibit adoption. An AR display that had a resolution of 100x100, could only display a green HUD, but was completely invisible inside a normal pair of sunglasses or glasses, would completely trump a super high quality fibre scanning display that was incredibly nerdly and unwieldly looking.
AR is pointless if you can't take it outdoors, if it's in-the-home only, you may as well go with VR.
FOV is pretty important. It's difficult to feel like you are in an augmented reality if you have to tilt your head just to see the full object in front of you.
Maybe you could get by with little helper things, but that's just a fancy HUD and not really AR.
Outside of entertainment, who cares if you feel like you’re in the environment vs being able to do useful things with it when you want to or need to? Using AR purely for entertainment doesn’t seem like such a revolutionary thing, really, especially when we have ok VR. I see some potential advantages for sure, but enough to make it a mass market nust-have device? I personally don’t see it.
Much of the affordances of AR will require a higher FOV - irrespective of use case.
E.g.: Tagging objects, directions and buildings with tracked information, following characters in the real world etc.
Long term a complete immersive solution - where virtual objects persist in the real world, will absolutely be more useful and compelling than both VR and limited FOV AR. Such a solution would literally be able to selectively replace and amend elements of reality as needed / desired. Although we can't fully predict it's uses, it seems self evident there would be many more than a poster sized slice of AR will allow.
I agree that the incidentals like tracking and form will matter a lot.
I think however there is no "proof of existence" of any software or capability that would compel even early adopters to use a 100x100 display, even with all the other physical issues being polished to a high sheen.
Looking down at your phone or smartwatch seems far more likely.
The "maybes" that people frequently suggest require a staggering amount of unwritten software infrastructure (sometimes up to and including sentient AI).
Its success will depend on whether it sucks or not. At this point all evidence suggests it sucks and ML's secrecy and doctored demos do little to refute that.
That’s the opposite of what the very thorough hands on Rolling Stone article had to say about it. Why are we even speculating about the user experience at this point?
I took a look at his site, and at first was inclined to agree, but a few things have convinced me that you’re incorrect. First, he’s been on ML’s case a lot lately, but his site goes into a ton of different companies and products. Even more than that, he’s harsh on any marketing bullshit regardless of the source (loved his treatment of wristwatch projectors), so maybe the issue is that ML has an unusually high noise:signal ratio. He seems to pull no punches, but going back years he also seems to really know his stuff, and is proven right, over and over. He’s covered tons of devices and companies too, so there’s nothing singular there.
Given that I’d wonder how much of his site you looked over before you came to your conclusion about a “singular” focus, which is demonstrably untrue. I’d also like to know which ML competitor he supposedly works for, since I couldn’t find a hint of that.
Sorry for the really basic question but why is the narrow field of view a problem?
This wasn't ever supposed to be like Google Glass, a device that you wear all day as you walk around. You put the glasses on for a specific experience.
If I'm putting them on to play a game, or read interactive motion comics from Marvel, I'm fine looking through a smaller window. I used to stare at my GameBoy screen for hours and its only a couple inch rectangle.
There is a pretty significant difference between the expectations of a GameBoy screen, in which your focusing your fovea on the screen, and a visual projection in AR. The biggest issue with narrow FOV in this case is AR strives to make items appear as if they are in the real world. The moment you slightly turn your head and item is clipped by the projected FOV, yet you can still see the real world where the AR item "was", breaks the immersive illusion.
With VR, even a narrow FOV doesn't have this problem: As you rotate your head and a virtual item moves out of the FOV, it's simply clipped into blackness... no visual dissonance as you have with AR item vs. real world it lives within.
Edit to address the article: I think the fixed IPD is the real deal-breaker here. People have greatly varied IPDs. Some VR HMDs, such as the Oculus Rift, have a physical IPD adjustent. I don't see why the ML headset couldn't have this feature in the consumer version as well, though.
The current/near future of augmented reality is mainly about preparing for the future of AR, and the future of AR is full-FOV devices that you wear all the time to seamlessly integrate data with the world around you.
As it is, a tiny FOV drastically changes the experience and the possibilities of the equipment. It could also mask issues that are only apparent at higher FOVs (like lagging overlays causing motion sickness). They've managed to make one of the smaller and sleeker AR devices, which will be important for mass adoption, but based on the current state of the tech it's probably putting the cart before the horse.
At this point I'd rather work on a tethered and bulky system with a massive FOV like Leap Motion's Project North Star, as it more closely matches the capabilities (but not form factor) of what we should expect from the first mass market AR devices.
When you're trying to interact with objects placed around you, and you can't seem them out of the corner of your eye, it's weird. You're looking at your desk, and you can see the AR objects placed right in front of you, but slightly to the left and right, the desk is empty. Swivel your head, and you can see some more objects, but now the entire left side of your desk is empty.
It's a pretty big problem, it feels like you're looking at the world through a straw that you have to scan back and forth.
Hookers and blow, same as always. The good news is that the huge investors who were hoodwinked here will be highly incentivized to avoid the embarrassment of shareholder lawsuits. And unlike Theranos, they at least have some kind of product that does something vaguely related to their claims.
Still, you'd think that at this point the CEO/founder black mock turtleneck should abort any pitch meeting instantly.
Leap Motion (which is not Magic Leap) recently showed a design with a much larger FOV, using so-called "bird bath" optics (which Meta 2 is also based on). It's not anything you'd want to actually wear in everyday life, but it should be an interesting platform for prototyping AR software. http://blog.leapmotion.com/northstar/
It would buy a lot more if it wouldn't only be in the same setting, but I think pretty much all current day AR headsets are too dorky to wear on the subway anyway. But it would still help a lot with office work, being able to continue working and doing small chores is still a big pro that VR can't fulfill.
Everything has tradeoffs and Hololens, MagicLeap and Meta have made different choices resulting in different ones.
The Meta2 has quite a large field of view (though a very different design compared to the Meta1, Hololens or ML). It is conceptually similar to taping a smartphone to the underside of a baseball cap lid and hanging plastic googles off the lens to reflect the light back into your eyes. Also, you need to connect a bunch of cables back to a gaming level PC to make everything work.
Hololens is very light, but is using tiny projectors to reflect back into your eye (also entirely self contained in the headset!).
MagicLeap has a different display tech, along with a bunch of other refinements and additions (eye tracking for foveated rendering, occlusion, etc.) also requires a smartphone sized battery pack/processor with a cable that goes to the glasses.
And then on the "number of super promising and ambitions technologies put into a single device:
MagicLeap beats Hololens beats Meta.
ALL of these devices are still clearly being pitched as "Developer" Units, none of them are ready for even the level of (still somewhat lackluster) consumer adoption that VR has seen.
Which is not to say that they're bad or that ML has blown $2B dollars, etc. I really don't think this is Theranos situation, it's just EARLY. In cell phone terms, AR still isn't quite to the Motorola razor level of development (and everybody's expectations are at iPhone/Android level).
Just from my anecdotal experience of using the HoloLens, (but not the ML1) the FOV problem these devices have is more tied to the display size than the unobstructed view of the real world. You brain will compensate for the obstructed view and you will forget about it pretty quickly. A larger FOV is obviously better, but it isn't a deal breaking flaw. The display being limited to an even smaller area will be something your brain will never get used to. You will have to be looking directly at a virtual object to see it. As soon as you slightly turn away the virtual object will disappear and immediately break any immersion.
It's not simply the size of the display device, it's the optics of getting the image to your eye, at a distance (virtual or real) you can focus on and clear throughout the whole FOV. A few hours strugglng with Zemax will convince you it's a Hard Problem.
Well, I think what folks hope for on HN is a certain "meritocracy" for news surrounding advanced technology. Magic Leap has arguably been trying to build buzz around their product with slick marketing and has been gaming the tech press to get unfair attention for their project and syphoning investment money away from other companies in this process. I think it's only fair, in turn, to keep the company accountable to at least a tiny sliver of their original claims.
2b is a drop in the bucket compared to other companies like Microsoft and Apple. They’re not just building a headset, but the software to run it and funding both 1st and 3rd party content.
The Model-T is a great comparison. It was a commercially successful product, not a prototype under NDA. It also tells you a lot about the state of the industry.
Ford Motor Company was founded with an investment of $28,000 in 1903, the equivalent of 34 Model-T cars - say around $2 million now (for a very expensive $50,000 car).
It was profitable in very short order, selling several models of cars. The Model-T itself was available for sale in just over 5 years after the company was founded. Ford innovated in production and product design, and made ground breaking products available for the masses.
Magic Leap raised $2.3 billion (the equivalent of 46,000 cars) 8 years ago. Total publicly available output: a few promotional videos.
The title implies (and some commenters seem to believe) that this is about an actual magic leap product, not a speculative recreation thereof. Can we get a title change to something like "Recreating a Magic Leap device" or "The view through a magic leap recreation"
I see it as a collaborative effort by the community to have at least one place on the internet that isn't driven by baitiness and spin. That seems worth doing and we're happy to help.
55 comments
[ 2.9 ms ] story [ 105 ms ] threadThere are so many hits on the site that it got throttled.
I just upgrade the hosting plan to hopefully solve the problems, but I won't take effect until 2AM PDT.
The rest is absolutely spot on though, without IPD adjustments any head mounted display is going to have real problems.
He may be right, but whether or not ML1 is a success really won't depend on whether or not stuff like the original concept of fibre optical scanning light fields will work, the ML1 will succeed or fail based on consumer experience and content availability. And I really don't think this Kremlinology-based approach could capture it anymore than people speculating on what the Apple watch or iPhones would be like based on photos of the device and patents.
This is an industry where outsiders have a very limited understanding of the serious serious problems in front of AR.
It's also an industry that is taking advantage of the very limited understanding of just about everyone not in it to hype things far beyond their capabilities.
If you care about AR in even an offhand way you should spend some time self educating and Karl Guttag's site is a good way to get started.
It's very interesting field but there are tremendous challenges in front of it and it serves no purpose to just assume the companies involved have actual magic leaps that solve the problems.
I think it is wrong in this case, however. As I understand it, the problem is "half-working" in the case of VR/AR/etc isn't something like less convenient that early adopters can simply put up with. Half-working VR has ability to sicken a person and injure the vision system.
See for example. https://www.cnn.com/2017/12/13/health/virtual-reality-vr-dan...
Broadly, almost everything in it has been either not born out (vision concerns) or less of a problem then first assumed (motion sickness) once the limitations of different users were understood.
Many people using VR find that the vestibular concerns are not an issue over time as long as the framerate of the system is high enough (90fps+). People actually do get their "sea legs" over time and many (largely gamers so far) ask for locomotion systems that were considered terrible ideas even 2 years ago.
You can't separate "consumer experience" from the technology like that. This is a case where the technology tightly constrains how good the consumer experience can possibly be. The consumer experience absolutely depends on whether fiber scanning displays or other new technologies can work, because tha changes the possible form factor and capabilities of the device.
It was so good because they figured out how to do more with limited hardware. As an example, if you wanted to see an image on the iPhone, apple stored a version with just as many pixels as the screen, the windows treo pulled the full image. The difference was that one felt snappy and one felt slow.
My point is that your comment sounds like what everyone said back then... you need more or better hardware. Maybe, maybe not.
Focusing too much on the display itself completely ignores the other aspects that inhibit adoption. An AR display that had a resolution of 100x100, could only display a green HUD, but was completely invisible inside a normal pair of sunglasses or glasses, would completely trump a super high quality fibre scanning display that was incredibly nerdly and unwieldly looking.
AR is pointless if you can't take it outdoors, if it's in-the-home only, you may as well go with VR.
Maybe you could get by with little helper things, but that's just a fancy HUD and not really AR.
E.g.: Tagging objects, directions and buildings with tracked information, following characters in the real world etc.
Long term a complete immersive solution - where virtual objects persist in the real world, will absolutely be more useful and compelling than both VR and limited FOV AR. Such a solution would literally be able to selectively replace and amend elements of reality as needed / desired. Although we can't fully predict it's uses, it seems self evident there would be many more than a poster sized slice of AR will allow.
I think however there is no "proof of existence" of any software or capability that would compel even early adopters to use a 100x100 display, even with all the other physical issues being polished to a high sheen.
Looking down at your phone or smartwatch seems far more likely.
The "maybes" that people frequently suggest require a staggering amount of unwritten software infrastructure (sometimes up to and including sentient AI).
Given that I’d wonder how much of his site you looked over before you came to your conclusion about a “singular” focus, which is demonstrably untrue. I’d also like to know which ML competitor he supposedly works for, since I couldn’t find a hint of that.
...eh? How is "study and analysis of the politics and policy of Russia" related to this?
https://en.m.wiktionary.org/wiki/Kremlinology
This wasn't ever supposed to be like Google Glass, a device that you wear all day as you walk around. You put the glasses on for a specific experience.
If I'm putting them on to play a game, or read interactive motion comics from Marvel, I'm fine looking through a smaller window. I used to stare at my GameBoy screen for hours and its only a couple inch rectangle.
With VR, even a narrow FOV doesn't have this problem: As you rotate your head and a virtual item moves out of the FOV, it's simply clipped into blackness... no visual dissonance as you have with AR item vs. real world it lives within.
Edit to address the article: I think the fixed IPD is the real deal-breaker here. People have greatly varied IPDs. Some VR HMDs, such as the Oculus Rift, have a physical IPD adjustent. I don't see why the ML headset couldn't have this feature in the consumer version as well, though.
As it is, a tiny FOV drastically changes the experience and the possibilities of the equipment. It could also mask issues that are only apparent at higher FOVs (like lagging overlays causing motion sickness). They've managed to make one of the smaller and sleeker AR devices, which will be important for mass adoption, but based on the current state of the tech it's probably putting the cart before the horse.
At this point I'd rather work on a tethered and bulky system with a massive FOV like Leap Motion's Project North Star, as it more closely matches the capabilities (but not form factor) of what we should expect from the first mass market AR devices.
It's a pretty big problem, it feels like you're looking at the world through a straw that you have to scan back and forth.
Where have all the billions gone? did they spend it already?
Hookers and blow, same as always. The good news is that the huge investors who were hoodwinked here will be highly incentivized to avoid the embarrassment of shareholder lawsuits. And unlike Theranos, they at least have some kind of product that does something vaguely related to their claims.
Still, you'd think that at this point the CEO/founder black mock turtleneck should abort any pitch meeting instantly.
The Meta2 has quite a large field of view (though a very different design compared to the Meta1, Hololens or ML). It is conceptually similar to taping a smartphone to the underside of a baseball cap lid and hanging plastic googles off the lens to reflect the light back into your eyes. Also, you need to connect a bunch of cables back to a gaming level PC to make everything work.
Hololens is very light, but is using tiny projectors to reflect back into your eye (also entirely self contained in the headset!).
MagicLeap has a different display tech, along with a bunch of other refinements and additions (eye tracking for foveated rendering, occlusion, etc.) also requires a smartphone sized battery pack/processor with a cable that goes to the glasses.
So you have on the physical UX front:
Hololens (headset) beats MagicLeap (headset + pack) beats Meta (headset + gaming PC).
And then on the FOV front:
Meta beats Hololens beats MagicLeap
And then on the "number of super promising and ambitions technologies put into a single device:
MagicLeap beats Hololens beats Meta.
ALL of these devices are still clearly being pitched as "Developer" Units, none of them are ready for even the level of (still somewhat lackluster) consumer adoption that VR has seen.
Which is not to say that they're bad or that ML has blown $2B dollars, etc. I really don't think this is Theranos situation, it's just EARLY. In cell phone terms, AR still isn't quite to the Motorola razor level of development (and everybody's expectations are at iPhone/Android level).
Seriously, does no one realize product development is a process?
Self-aggrandizing voices like this really turn me off. "Sorry the world didn't do better by you! We'll all really try harder next time, dad."
Amazon raised total of $108 million before its IPO.
Google raised less than $50 million before its IPO.
Facebook raised $2.3 billion before the IPO, and most of it was after it became a world-wide phenomena.
Ford Motor Company was founded with an investment of $28,000 in 1903, the equivalent of 34 Model-T cars - say around $2 million now (for a very expensive $50,000 car).
It was profitable in very short order, selling several models of cars. The Model-T itself was available for sale in just over 5 years after the company was founded. Ford innovated in production and product design, and made ground breaking products available for the masses.
Magic Leap raised $2.3 billion (the equivalent of 46,000 cars) 8 years ago. Total publicly available output: a few promotional videos.