Hacker News Reader

I saw the magic leap demo in person at their Florida office. It was quite something.

Imagine minecraft, but in real life. They had blocks you could put on walls, dinosaurs roaming around on the ground, knights fighting the dinosaurs, and all of it was controllable.

It was in a small-ish room, roughly ... 15x15 feet? a few meters by a few meters.

It had couches in the room, and pictures on the walls. It didn't look special. But in retrospect the room may have been part of the demo in some way.

(I went through their interview process, and one of the benefits was getting to see the ML in action. Supposedly they also had an "AI assistant" demo or something like that – Cortana? – but it wasn't available on that day.)

If I were an investor, I would probably invest based on the strength of that demo. It was enough to make you question the reason we're all staring at laptop screens. The device was comfortable, and I could imagine myself sitting at a desk typing into thin air (because goggles) rather than typing into a computer screen.

Of course, it looks like I would have lost my money if I were an investor. But how could we know it would play out this way? All they had to do was build a strong developer ecosystem. The lame demo-style apps we see are a direct result of inconvenient APIs and SDKs.

In fact, they were actively hostile to developers. I remember getting a C&D just for publishing their SDK's manual on a personal website. No idea how they even found the link.

The premise is real – in the same way the Vive was in many ways superior to Oculus, I think the next "Magic Leap" will be superior and more affordable than what we see here. If you are looking for an investment opportunity, the AR scene is still a strong bet over the next decade or so.

(If that seems unlikely, think about how many major advances worked out after seeming so unlikely: deep learning in AI; consumer-grade VR; voice controlled devices; the list goes on and on.)

It's not just software though, the viewing hardware they eventually shipped is extremely similar to the hololens, but 2 years later and with a slightly larger viewport. And worse hand tracking, from my experience. Cheaper though.

What you and other early-people seem to describe appears to be something else entirely, in which case yea - original plan fell through completely and they pivoted to their current thing. But was it actually different?

>But how could we know it would play out this way?

For anybody who hadn't seen the demo, the company always looked like typical SV smoke selling pitch. "This is the best thing ever", "It will change the world", "We have great stuff but we can't show them in public because reasons".

The whole "demo in a closed, secret room and then you can't tell anybody about it" reminds me too much of the carnival fair fortuneteller experience. You get shoved in a mystery room, get shown a bunch of smoke and mirrors, and then you're out before you can't think too much about what happened.

Actually putting on a headset and getting the experience is not equivalent to smoke and mirrors, even if it is optimized for that one room

Not everyone. The potential is absolutely there, and even the devices that are available today would be groundbreaking, if only someone could figure out something to do with them.

As for that, the ecosystem is far too closed. ML and Microsoft seem to be approaching the technology Apple-style, with expensive hardware and by-the-numbers (if that) developer support, which would allow them to control the platform - and profits - once the we reach some sort of inflection point where either a unpredicted killer app emerges or the tech matures enough for the more obvious applications to be viable.

They're not going to get there, though; someone is going to (or perhaps already has) released a cheaper alternative with a low barrier-to-entry creation platform, and that will allow the number of people who have access to both a good-enough device and a good-enough skillset to reach a critical mass. They'll create for that device and platform and everyone else will be playing catch-up. It's the early PC and BASIC all over again.

Do you know what would be really interesting? If Sony's PS5 were to launch with an affordable VR headset and an updated version of Media Molecule's Dreams.

I think the thing you missed from the 'all they had to do' list, was make a product that a large audience could afford.

I think Zuckerberg said at the last FB conference their aim with the Quest was to get 10,000,000 sold because that's the tipping point to a self sustaining app ecosystem.

Software is worth making, so hardware is worth buying, so software is worth making... etc

They had zero chance of achieving this at their price point.

Interesting experience! Do you know if what you saw was the same as the product that shipped (magic leap one)? If not, what were the differences?

Could you talk more about what made the demo so great?

> creative nomad jukebox

Holy crap, the memories. I had the creative nomad jukebox and for years convinced myself if fit in the pockets of my jeans...it did...but it didn't. The folly of youth!?

In it's defense, mine still works.

Wow! I think mine broke about 4 times. Kept taking it back and getting a replacement, teenage me must have cost them a fortune.

I had one of these too. It was so _nearly_ good. So so close. That tiny track pad was just too sensitive, or was in insensitive, or both?

Actually, back in those days I was wearing Jnco jeans and you could fit a laptop in the pockets if you wanted to. I made a throwaway username for this comment because I don’t want people to know I wore Jnco jeans.

When you can control every element: lighting, view angle, distance, background, etc, you can hide a ton of fatal flaws.

I got to demo their consumer product and it really is pretty cool. Perfect? Far from it. But it's good enough that you put it on and say "wow" for the next 15 minutes.

In the demo I saw, you get immersed into a coral reef and walk around. It's very cool, but I'm not gonna buy a unit just for that. So you need lots of content before it makes sense to buy one of these things, and then you have a chicken and egg problem. Who is going to spend massive amounts of money to create content when there isn't already a big audience for it?

Have you tried one of the other VR systems? If so, could you comment on how the experience compared?

I'm not the person you're asking, but imagine looking through dark sunglasses through a little window at a faded image sitting on the table in front of you.

For me the illusion of it sitting on the table didn't even feel like it was really on the table because of how dark the glasses were and how faded the image was.

VR was like being in a different place with a real sense of perspective and your hands in VR felt like a part of you. Sure it was low resolution, moving was strange, and the sides were letter boxed a bit but it was an impressive thing.

I think Apple are shipping their AR hardware, mostly anyway. The iPhone and iPad are it, modulo enhancements like LIDAR. I really don’t see Apple bringing out a headset. I’m not even sure it’s a technology problem. People just don’t want to walk around with cameras and LIDAR and goofy goggles on their faces, not at Apple scale anyway, possibly ever. It’s a fundamentally flawed concept. Specialist applications sure. Mainstream, one in every home? I don’t see it.

I think they're definitely working on something: https://www.macrumors.com/roundup/apple-glasses/

No idea when it will be a viable product in a nice enough hardware package, they're working on the ecosystem and platform in the mean time via iOS/iPadOS, but AR via those devices is a lot less compelling.

Doing this in a real way would be big if the hardware is possible, but it may be a ways out.

Interesting link, thanks. It looks like all their acquisitions and hires, at least the ones linked there, are on the capture, interface and authoring tools side. The only actual mention of a headset were of glasses with a camera, presumably for environment scanning, but using a phone as the display. I really could be wrong, but every time I start writing something where I hedge my bets it just doesn’t feel right. My gut persistently says no, VR/AR headsets are niche tech.

They are absolutely building a headset.

They acquired Akonia Holographics in 2018 who specialise in holographics for an AR headset. And they already have many patents:

https://www.patentlyapple.com/patently-apple/2019/11/apple-w...

The link up thread shows patents going back 10 years on headset ideas but still no product, they could work on it another 10 years and it still go nowhere. They look at all sorts of things and patented all sorts of crazy ideas for much longer that but also went absolutely nowhere.

The only real indication is Akonia, that is interesting, but they have 200 patents on display technology. Any of that stuff could be useful in other applications. So I’ll stick to my guns.

> Supposedly the magic leap demo was actually cool and used different technology than the eventually crappy hardware they ended up sort of shipping.

i think that technology was CGI:

https://hothardware.com/news/magic-leap-admits-outrageous-au...

That's not even a demo of the actual physical product though, that was just a video that was posted online that purported to show what the experience would look like (but actually was not). It's not like you would've seen that had you actually been looking through the glasses.

That was the concept video which was different from the demo, but you are correct and they did a horrible job conveying that that was a concept and not the actual product.

Did they? They raised a billions of dollars to try to basically try to drag reality to this demo. It feels like a bigger version of a regular venture story.

"Gimme a ton of money to run this experiment. If I'm right you'll be rich"

That would have been fine if they said that, but they pretended the concept videos were real.

Even this post is another example of the continued dishonesty I'd expect from them. This pivot is obviously not about COVID-19.

I'm not talking about their fake demo ads they put everywhere, I think they actually had an impressive in person demo using some different technology they couldn't miniaturize.

WHy wouldn't they push the narative "Look at this - now we'll make it smaller" vs. "it's awesome - trust us and wait"?

Even snake oil salesmen have a demo if they're good; these guys suck at being phonies.

I'm not a marketing person, but constantly alluding to something amazing without revealing details is a hack that stirs up a lot of curiosity and people discussing what it could be.

If you reveal the thing then that dies down (or worse knowledgeable people know that what you're dong isn't possible), but if you keep it secret while giving content-free little hints about it you can keep it going longer (and maybe raise more money by letting people in on the secret?).

I have a strong dislike for this kind of thing, but that doesn't mean it's not effective.

They had a few-hundred-pound cart-bound prototype called The Beast that was supposedly mind blowing to use, and that's what convinced a lot of engineers to drop everything and move to Florida to work on it. I agree pushing that technical narrative would have sounded much better.

They got to series E funding and raised over a billion dollars. If they were phonies, they were phenomenal ones.

Tharanos raised almost a billion, and look how that ended up

What's really crazy about that video--while it's technically possible to make something mostly like that with current hardware[0]-- is that, if you have any experience with AR at all, you know that most of those UIs would be terrible to use.

[0] The FOV is accurate, given we're looking through a narrow camera lens, but gives the wrong impression that it fills what the user could see because it fills the video frame. The graphics wouldn't be "solid", they'd be transparent, but a pre-setup room can definitely do occlusion effects with foreground furniture. The physical gun controllers could be done, though nobody would fork out the money for it. And all the hand gestures and UI pinning stuff could be done, though the software support on Magic Leap does not help you in the least.

Yeah - even the fake demo use case isn't that compelling to me.

This is the kind of thing that I think the real AR value will be from: https://twitter.com/st8rmi/status/1249950879807045633?s=21

Basically a meta-layer for the real world that you can interact with outside of a screen. This would let you do things like interact with a lightswitch from across the room by looking at it, get metadata about most object states by looking at it, anchor big displays to white walls, etc.

I think there's huge potential for this kind of interface, but I suspect the hardware isn't possible yet.

The hardware can do this, it's just that you can't get any funding for anything interesting. You're basically stuck with hobby apps and marketing demos developed via consultoware. The hobbiests can't afford the tech or the lack of reach and the consultoware shops have exactly zero imagination (I know, I worked at one).

I personally define VR vs AR as "who provides the context in which we are working? The app (VR), or the user (AR)". A lot of extant "AR" apps don't do anything particularly interesting with your surrounding environment.

If your AR app needs me to clear out a space in my livingroom to give you room to drop some 3D models that maybe bounce off my walls, you've not actually made an AR app, you've just made a crappy VR app instead. Facebook could release an update to the Quest any day now that auto-scans your room to set the boundary and then you'd have exactly the same experience in an occluded headset, but with twice the FOV and better input.

This is the thing that gets me. There are plenty of people who would be willing to design around existing constraints, just because they think the tech is cool and they want to see what they can do with it.

But the cost of entry is mid-4-figures, between the hardware itself and the required development equipment. It's obvious that the people making the hardware and platform aren't a wide-enough cross-section of the public to create what business interests or consumers don't know that they want yet. They're shooting themselves in the foot, trying to maintain control over the platform.

FB actually already working on this for their VR/AR:

https://www.youtube.com/watch?v=NNppUJzsvXQ&ab_channel=Oculu...

https://www.youtube.com/watch?v=BbRvARuCoRU&ab_channel=Oculu...

My guess at the killer app for AR is airplane maintenance.

Imagine a physical checklist where areas get highlighted, arrows to direct you to the next step, and a little red icon that goes green when you're done.

I think this could shave real time (maybe a third?) off airframe downtime while keeping the very high accuracy requirement. That would save actual money.

This already exists

Cool!

Is there anything I could read about that? I'm taking your word for it, I would just find that interesting.

Or do you mean "this already exists but isn't in AR goggles"? Because I can see where my phrasing was unclear, I mean this mounted in an AR HUD.

https://upskill.io/resources/blog/aircraft-maintenance-breez...

https://defensesystems.com/articles/2018/06/21/af-augmented-...

https://atheerair.com/ar-mro/

I personally also worked on similar systems for industrial factory equipment repair.

As funny as this sounds I must agree. If they had something cool & innovative they could have shown the world and been upfront about the challenges to make it smaller. That could in turn bring new talent to help them. Instead we now see a drowning company.

In some way this reminds me of Theranos.

Reminds me of this Magic Leap - Expectations vs Reality https://www.youtube.com/watch?v=UZ7-F_vWUVE

> than when I had played with VR hardware for the first time

AR is 1-2 orders of magnitude more demanding than VR, so if we get to acceptable screen-door-effect-less VR on 30TFlops hardware, we might need like 1 Petaflop for the same with AR. That won't fit into a pocket anytime soon, but we can build such experiences on beefy demo rigs.