Minecraft Earth and Dennou Coil

[efaardvark]

If you haven't already, go watch Dennou Coil, aka Cyber Coil.  It is kind of a fun sci fi about a bunch of kids and the fun they get into while wearing augmented reality glasses.

In the real world we're not quite there, yet.  Augmented reality in the real world is bounded by things like Pokemon Go and StarChart on smartphones, and Microsoft's HoloLens is still pretty bulky and expensive.  Nowhere near the nearly-invisible glasses they wear in Dennou Coil.  If you've never seen MSFT's 2015 E3 Hololens minecraft demo you should.  Microsoft obviously wants us to go there.  They've just announced their Minecraft Earth app (below), which to me looks like a portal, both in a marketing sense and a gaming sense.  Clearly - to me anyway - the plan is to get people using the app to build content and attach it to the real world so that when things like Hololens get cheap enough there will already be a rich environment to work with.

[mal type=anime id=2164]

and

Minecraft Earth,

 

 

 

[leinwandname]

Global, augmented reality with "simple" glasses (and, perhaps, in the Dennou Could universe in the future with contacts) certainly would be interesting, furthermore I think Dennou Coil made these cyber battles really interesting (I was recommended this show after I felt let down by "Summer Wars" and its depiction of "battles" in a bit more advanced internet). Finding a way to project a new texture over literally any object IRL, is probably a bit far in the future - having 3d creatures roam the streets actually seems more feasible after Pokemon Go- but it'd probably be pretty awesome. Although the safety risk would be monumental.

[Geano]

I think its certainly a interesting concept to be sure. It is along the line of what Microsoft wants to focus on considering they are working a lot with VR/AR development. So something along this is not to surprising. I do think like something like PokemonGO it will need some time before it is really good though time is on its side.

[efaardvark]

59 minutes ago, leinwandname said:

Global, augmented reality with "simple" glasses (and, perhaps, in the Dennou Could universe in the future with contacts) certainly would be interesting, furthermore I think Dennou Coil made these cyber battles really interesting (I was recommended this show after I felt let down by "Summer Wars" and its depiction of "battles" in a bit more advanced internet). Finding a way to project a new texture over literally any object IRL, is probably a bit far in the future - having 3d creatures roam the streets actually seems more feasible after Pokemon Go- but it'd probably be pretty awesome. Although the safety risk would be monumental.

Stuff like Dennou Coil is why I like "hard" SF.  It gives you a glimpse into where we are, or at least could be, going.  Sometimes it is a warning.  Sometimes it is a wonderful vision.  It is always interesting, IMHO.  Unfortunately "hard" SF is hard to get right.  You re constrained by the known, and very limited in what you can extrapolate and make into a story.  Stuff like Star Trek and Summer Wars is more entertaining for most, gets a wider audience, and is ultimately more profitable.  STEM education being what it is these days quite a lot of people can't tell the difference anyway.  Not that I'm knocking the "soft" stuff.  I can appreciate that too.  I enjoyed Summer Wars.  But to really get my brain engaged I need the hard stuff.

Current best-tech for AR/VR is anything but simple or cheap.  The hololens is $3000 last I checked, and is bulky and balky.  The stuff from Magic Leap is cutting edge and cheaper, but still not anything like the almost-not-there I/O glasses/devices in Dennou Coil.  That said, once a prototype of a device like this is built, no matter how bulky and impractical, if there's enough perceived demand then the electronics of things like the cameras that do the "inside out" positioning can be merged and miniaturized.  The software that runs on a general-purpose CPU to do those "background" tasks can be hard-coded for a dedicated, massively parallel processing chip that does that one thing far faster and cheaper and use less power besides. 

We've seen this effect already in 3D gaming with GPUs, which themselves evolved from FPGAs.  GPUs could be used in machine learning, but now nvidia and Intel are making even more specialized chips for that.  Tesla's also done their custom "FSD" - Full Self Driving - chip for their cars which consolidated, "simplified" (from an outside perspective), and made more reliable and cheaper a bunch of stuff that had been done the hard way with off-the-shelf general purpose components.

The HoloLens/Magic Leap AR gear (and some consumer VR gear as well) are already mounting outward-facing cameras that the system's software uses to position and orient the headset in the user's existing, ambient environment.  In some cases it also positions associated "peripheral" devices like hand controllers in 3d space.  This is what I meant when I used the term "inside out" positioning, above.  The position and orientation information originates from the device itself.  As opposed to "outside in", which relies on separate outside sensors pre-positioned in an environment and connected to a computer to feed the headset's position and orientation data to the VR engine. 

Clearly "inside out" is what you want in an ultra mobile device.  That's pretty hard.  The outside in method is currently the cheapest, most accurate, and most easily do-able with existing off-the-shelf components.  Unfortunately it kind of ties you down by limiting you to a predefined operational area defined by the outside sensors.  There's nothing tech-wise which is impossible in the Dennou Coil world however.  Cameras can be made almost as small as you like, and their inputs fed into massively parallel, special purpose silicon built into the frames, which in turns outputs position and orientation info to the upper layers of software.  Virtual retinal displays are already a thing.. actually kind of a solution in search of a problem like lasers were back in the day.  Smarphones can already play 3D games, and connect wirelessly to periperals and the 'Net. 

Support a VRD with a phone-sized "compute brick" that uses an inside-out headset as its user interface and there you go.  Probably won't happen any time real soon, but I see no intrinsic problems with implementation.

And yes, there would no doubt be hazards with the new technology.  There's already been people running out into traffic chasing pocket monsters.   4 billion years of evolution are wasted on some people.  SMH.

 

[leinwandname]

@efaardvark

Hard SF is really the only SF I'm into, perhaps because, as you said, it gives us a glimpse at what we might have one day, in one form or another (Although I have to admit that I'm not that much of a SF fan except Cyberpunk and the 'very near future stuff' like Dennou Coil, Psycho Pass, etc. ...)

 

I didn't know about VRD, thanks for mentioning that. Definitely an interesting topic. Imaging having glasses/lenses connected to some kind of central computer that processes most of the stuff after receiving information from the glasses and then sending it back for the user. Something like what came up (not so) recently where you play a game which is on a pc somewhere else and you just send the (controller) input and receive the output on your screen. With fast wifi available pretty much everywhere, you could reduce the lag to a bare minimum (ofc, there's always going to be lag).

Since you mentioned cars, the FSD might, or probably will (since it's the point), change the rate at which self-driving cars improve. To go back to the idea above (doesn't have anything to do with the topic but still an interesting thought) - imagine having every car self driving, connected to servers, so that every car know the location of every other car in the vicinity, plus the traffic lights, pedestrians, etc. ... Could prevent accidents on a whole new level, plus plan traffic way more efficiently (for example have an AI [or several AIs working together] overlook the entire traffic in the region and managing every car's speed, direction, path, etc. .... You could steadily train it with simulations and rl exp. [like Tesla cars do] ) Having people run into traffic because they're trying to catch their 100th rattata wouldn't be a problem either, since this/these AI(s)  could see it coming from cameras on the different cars, public cameras, etc. ....

 

But if Dennou Coil taught us anything - don't fall asleep with you glasses on (or in you selfdriving car)!!