Here's a clip of the intro. Decide for yourself: https://clips.twitch.tv/BreakableTentativePassionfruitPipeHy...
The twitch clip has a home shopping esque feel to it. That seems ...appropriate. All this from a company with a ~$6B valuation. Imo, that was the real magic leap.
I predict that in a few years, AR vendors will focus on AR for industrial purposes, and then it will be useful.
Off topic: When I was a kid we once augmented it with bottle rockets. A little stupid, but really fun.
Part of me genuinely hopes I'm wrong and wishes we'll have cool AR soon... but right now it just feels like nothing good will come out of the work at Magic Leap.
The most worrisome aspect of the presentation for me was that it's not designed to work outdoors. They said explicitly that the initial product is aimed at indoor use. Which makes me wonder what it has that my Vive doesn't, since the field of view is apparently limited. (One of their presentations in July is titled "FOV mitigation.")
Still, Magic Leap has proven time and again that they have a plan, and mixed reality is coming one way or the other. ML won't be the last player in the space.
As happened with the "hoverboards" that are basically just cheap clones of the SegWay concept.
I don't have sales numbers but there are any number of articles suggesting that they came and went. [1] Certainly, one doesn't hear or see much about them these days.
Outdoor environments get a huge amount of NIR light which can swamp cameras without an IR cut filter. It's possible that the Leap is using infrared imagery for motion/depth tracking. Similarly devices which project structured light for 3D reconstruction usually use NIR projection. These devices fail in sunlight.
The other issue is depth. Depth sensing on a device like the Leap is going to be done using either time of flight, structured light or stereo [1]. Time of flight is unlikely due to the amount of power you need to dissipate on the LEDs (see the Kinect V2) - which are mounted next to your face. This approach would also preclude outdoor use, see above. If they're using stereo alone, then the depth accuracy will only be good at close range. For a typical baseline of say 20 cm, you're going to get reasonable results up to 5 metres [2]. More robustness if you use structured light to assist.
[1] There appear to be multiple cameras, so why would you not exploit the geometry?
[2] Back of the envelope calculation: stereo error at 10m using a 20cm baseline, 2.2um pixels, 5mm focal length and 0.5px accuracy is about 5cm. Accuracy at 5m is about 3cm.
Loads of processing power required... solvable.
Now the real problem is the display of sufficient brightness and colours. Some sort of future translucent LED screen could do it, maybe. DLP can do it at big cost and with bulky projectors. Current picoprojectors are not very bright...
And then three is a focal length problem to solve to have sharp image and good depth range while not messing with accommodation.
Interesting that she called the controller a "totem". Is that a common term? A Montreal VR startup called Vrvana was developing a headset called Totem.
For better or worse, Magic Leap PR is apt to invent Words.