Microsoft HoloLens [Virtual Reality, Augmented Reality, Holograms]

Wrong thread.

 

But this can already be done with AR can't it?

Sorry, late to the party and probably been covered - really disapointed to learn the FoV restrictions, clearly not going to be used as per the E3 demo, I can't help but start to think this is like Kinect all over again - potential but over-promising what it can deliver.

 

Thanks for the link, I was looking for this.

It looks like holographic waveguides have an absolute limitation of about 36 degrees with normal 1.7 glass, but can go up to 47 degrees with 1.85 glass so maybe that's what they used for their first prototype and dropped it afterwards because it's crazy expensive. The first proto was said to have felt wider from most accounts.

 

Would sapphire help with that?

 

Not really, sapphire is 1.76 index, and it's expensive, the major problem to go even a little beyond 36 degrees seems to be cost. The example in the document was a 1.85 glass which is probable a hundred dollar per eye. It is also the maximum possible. Well nikon makes something at 1.9 but that would add maybe two degrees... for even more money.

Anything above 1.7 is stupidly expensive, they need a different technology to reach significantly beyond 36 degrees.

Nokia had a thing called Exit Pupil Expander... maybe MS got that technology as part of their aquisition? Don't know if that can break the limitation, or if they are already using it to go from 20 degrees up to 35.

 

https://twit.tv/shows/windows-weekly/episodes/418?autostart=false

IF you listen to Paul the original demo's featured a wider FOV. So its either the hardware can't power it or the lenses are to expensive. I think they would have known about the cost of the lenses before they built the original protypes.

 

Prototypes are for exploring options. They would likely have known about the cost, but they'd still have explored the option. Not least because they're targeting industry and business where cost isn't as much of an issue. 36 to 47 degrees is probably worth an extra few hundred bucks to business if it makes a difference in use. I guess the window is so obviously a window that there's not a massive difference in 11 more degrees FOV - the base experience it still the same with the content being very centered.

 

It makes as much sense for Microsoft to build internal hardware that can't be replicated in a commercial unit as it does for Microsoft to do fake stage demos that won't be matched by any internal hardware they'll have for the next 5 years, and yet they do it anyway. All of the articles I've read from the January press function that mention the FOV use words like "tiny" to describe it, so I don't think we're talking about a dramatic drop in FOV, perhaps 10deg or so, which would still put it within reasonable limits of the waveguide tech.

My guess as to what happened is this: their CEO made a strategic decision to shift what was a blue skies R&D Xbox project and fast track it to use as a high visibility tie-in for the Windows 10 launch and a much needed halo tech for the Windows brand. Because it does technically do what they say it does, it gives them just enough license to get away with showing their stage demos as they are, and considering that no meaningful content would have been developed for it during the Windows 10 time frame anyways, whether the FOV is 30, 40, or 50deg doesn't seem like a huge deal. If it had stayed an XBox peripheral they probably wouldn't launch it for another few years, perhaps even just waiting until the next console cycle so they can hit a reasonable price point. As a stand-alone Windows 10 niche tech however, they can launch it as early as they want, charge as much as they want and have it be as useless as they want, and still get the benefit they're after.

 

Except if they release a useless product, they'll reduce interest in sequels when they're trying to AR for real. Windows 10 had enough coverage that it didn't need magic glasses to catapult it to attention. Just giving it away free is all that's needed to hit the headlines in a big way.

 

Isn't this exactly what happened with kinect? The sizzle demo was showing things impossible with prime sense. Kinect 2 is a completely different technology and provide much more usable data... but gamers don't want it because of the bad kinect1 experience.

Immersive AR, will need 100 degrees and physical alpha blending. The tech is not ready but it could be in a few years. While Hololens v2 would be competing with the same tech as their competitors, their software and the rest of the hardware would be already developped and MS would have a head start.

The problem of not having proper alpha blending with reality (it can only add light) is much more important than the FOV. It's easy to see with the following demo which is in a normal environment, not a dark controlled room, they put a camera behind the actual headset. (this is SEER, which is huge and clunky, but 100 degrees)

 

I think Kinect worked. It was just limited, and burnt out the 'jump about' market in one iteration. Kinect 2 needs something more substantial to justify it. Hololens and AR are a virgin market. Just as Sony says about VR, not wanting to provide a crap experience and scare consumers off the idea, AR needs to be done right when the masses get their hands on it.

If you design the AR experience around the bleed-through, it shouldn't be an issue. It will greatly impact visualisation experiences though. One option is to have a selectively opaque visor to keep environment brightness at a maximum threshold for the intend VR.

 

Regarding some of the comments above that you could do AR in VR with a stereo camera feed - aside from the tracking requrement mentioned, it just doesn't work. The cameras sit outside of the headset so are not in the same position as your eyes. *

A way round it would be some sort of reprojection of the image and then overlaying the AR elements. That's presumably computationally expensive to do at 90fps. I assume Oculus or whoever have experiments running though.

It seems more likely AR will merge into VR tech with this sort of thing than there being an AR display breakthrough that allows a FOV wide enough for VR.

I think hololens and magic-whatsit will have a fair few generations before this happens.

*sorry if I missed someone say this!

 

I wonder about CastAR, they did a kickstarter for AR that could also be used as VR...

 

The res is 720p on the cast AR isn't it?

They can target a high fov than used in the AR view with their VR clip on. It would be a trade off with resolution though.

 

Kinect 2 is basically let down by hardcore gamers and developers alike, and a tiny little bit by space requirements. I'm personally hoping it will still become a success eventually. They should have served the Windows market much sooner and better so indies could develop for it easily. It should also be really good for .... VR.

Hololens is bound to be a success. It's a huge deal that this is an all in one device - you can just get one and use it anywhere for almost anything. Of course its FOV is still limited, and that's quite a big thing, but on the other hand it also really works. None of what was shown has been faked as far as I can tell, and I've heard from guys like the Bombcast crew who've actually tried everything that was also shown on stage in person and the whole thing is in there, pinch to move the minecraft world included. They also confirm that the FOV is indeed really small, but other than that it is really amazing technology.

 

microsoft shown the limited AR field of hololens

 

The FOV shown in that video is still like 50% larger than it actually is.

 

Ah ?
Used an Hololens ?

 

No, but there's a researcher in this field that's measured the FOV, so we have a pretty confident estimate of what it actually is (30deg horiztonal). According to my quick napkin trig, a 30deg horizontal is what you get from sitting 39 inches away from a 24" 16:9 monitor. If you take that seating position in front of your monitor and try to replicate what's seen in this video (a hand in view, bent comfortably at the elbow) you'll see their hand occupies a noticeably smaller portion of the viewport than it ought to. Trying to replicate what's shown in that video makes it look closer to ~50 deg. That might come across as nitpicking, but it does strike at the heart of my issue with the practicality of the device. Most of the obvious, low-hanging fruit applications as a consumer device (imo) involve arm's reach, seated table-top or floating window style applications and a 30deg viewport is too small to comfortably contain even a standard sized chess board or desktop monitor at that sort of distance. If there were strong prospects on the horizon for wider waveguide displays then I would happily accept this iteration of Hololens as a developer unit and spend time on applications that might be viable in a couple years on a future iteration, but there's no indication that this is the case.