Fool me, this is real good shit ! [the VR thread]

[error]

They're saying that it's the same tech kinect which can generate a depth buffer of what you're looking at really quickly.

It cant be that though, Kinect is a fixed device looking at moving things. It can get information by projecting IR dots and looking at how they change.

If a HMD projected dots both the dot projector and the camera move in the same way - you gain no information.

It doesn't need an object to move to scan it's distance. It compares the captured image with the image it's projecting. The camera is offset from the projector, so there is distortion in the captured image which gives the depth information.

[Darkflame]

I didnt mean to imply that it did - I meant that what its measuring the position is relatively to itself which is in a known position. In this case your trying to measure where you are relatively to yourself, which doesn't work.
If you already knew the "expected pattern as it would appear on the wall" then, yes, the Kinect would work. But in this case both the patterns reflected result in not known, and where you are is not known.

You can use the old bomber technique to get your distance if you cast two lights from two separate angles and look at the distance between their lights on the target - but even that wont give you your rotation.

Short of recreating the whole environment by moving around significantly like optical techniques do, I cant see how it can be done as simply as with the 360-era Kinect. Its no longer "just trig" as you have lost a known variable.

[Little stoned kiwi]

Microsoft's HoloLens uses unreleased Intel Atom chip
Which doesn't really give anything away, the CPU isn't going to be doing any of the heavy lifting.

The HoloLens should probably have a thread of it's own instead of stealing the Rift's thunder. It might even warrant a mod moving the last few pages into one, since Yak, Giles, and a few other yakkers are developing/using the Rift?

[gravy]

I mean I don't even...


Hello Mr Pot, please meet Mr Kettle

[error]

I didnt mean to imply that it did - I meant that what its measuring the position is relatively to itself which is in a known position. In this case your trying to measure where you are relatively to yourself, which doesn't work.
If you already knew the "expected pattern as it would appear on the wall" then, yes, the Kinect would work. But in this case both the patterns reflected result in not known, and where you are is not known.

You can use the old bomber technique to get your distance if you cast two lights from two separate angles and look at the distance between their lights on the target - but even that wont give you your rotation.

Short of recreating the whole environment by moving around significantly like optical techniques do, I cant see how it can be done as simply as with the 360-era Kinect. Its no longer "just trig" as you have lost a known variable.

I think I must be missing what you mean? You still have 2 points on yourself, the projector and the camera, the third (well nth really, as it's many points) is the projected image.
The Kinect does know the pattern it's projecting so it can easily triangulate it's location. It also, strangely, has accelerometers so it can handle rotation. I think the only things missing are initial direction when you turn it on, which you could fix with a compass, and initial location. But a lot of things you would start up while it's on, like that crappy 3d thing that looks like a simpler version of spore and sticking a holo tv on a wall. I'm assuming it would cache your surroundings so your home, office, garden, space station, ancient abandoned mine, toilet etc would be like a desktop and you could pin things in rooms, leave and they would still be there when you come back.
Check this video, the Kinect clearly knows where it is in relation to it's surroundings.

I mean I don't even...


Hello Mr Pot, please meet Mr Kettle

So we won't be seeing Godus in the Hololens anytime soon then?
Did make me think of this though.

A decent god game would probably be a system seller for me.
I've actually made a basic populous style landscape editor in unity for the rift. It needs a lot of work to be anything more than a novelty though.

[pogo]

Hello Mr Pot, please meet Mr Kettle

He’s mostly right though. It does need a killer app, it does need to convince developers that it will sell in the millions before they commit the sort of resources needed. And he was one of the people involved in Kinect from the start, so knows how it fell flat despite MS’s massive investment in it.

Here’s one half-serious thought. This is clearly far from ready. It’s not ready in an entirely different way from Rift and Project Morpheus, both of which should be able to ship this year. The HoloLens is probably years from appearing in shops, if it’s even possible to fit all that tech into something affordable people will wear.

So why announce it now? In theory for developers but it’s of little interest until it’s much closer to final hardware and release, especially as it’s so radically different from other platforms there’s little scope for re-using stuff made for it. Could it be just they wanted to get something out there before Oculus and Sony, to raise expectations and disrupt their launches? Tech companies do this a lot, announce products early to coincide with other companies launches and so get in the way of their news coverage.

[gravy]

Makes me laugh because he was the one over hyping the kinect.

As for the other stuff? Don't fuck it up? Amazing insights from Molyneux there

[gumboot]

Don't fuck it up?

Well if it's so obvious, Mr Smarty Pants, then why isn't everybody "don't fucking it up"?

[Darkflame]

It will have developers easily.
Unless Microsoft deliberately excludes them, it will get shitloads. Much smaller platforms have got plenty of support and this has far more potential sales.
And, yeah, more then the Rift. The repairing-a-lightswitch demo moves it out of "toy" space and into "actually useful for stuff". Id guess anyone involved with product design of some sort will be looking closely. Architecture. Medical.
If Autodesk and Photoshop doesn't start adapting their 3D software for its use, then they are just handing more market share to Blender and Krita. (and, really, with Adobe going subscription only, many people will be happy to leave). Its going to be seen as an opportunity.
Unless Microsoft really makes it nasty for developers some how, they will come.

Anyway, I read about 7 hands on impressions now;

Criticisms;

1. Real device is bulky.
2. Needs power cable
3. Bit uncomfortable for larger heads.
4. Field of view isnt complete.
5. Voice control sucks is a stupid sucky way to control some things.
6. Sometimes things seem too transparent. (I assume if the room is too brightly lit - you cant project black, after all)
7. No one played with the 3d creation app.

However, almost everything else is universally praised.
The most significant thing imho is this though;

“Tracking -- which is to say, "how the headset interprets where your head is in relation to the world around you" -- felt the most fully-baked of any of the headset's sensors. Though the prototype was a bit finicky to move very quickly in, I had no issue turning around quickly or kneeling, or any other movements I tried.”-Ben Gilbert (Engadget)

Cracking the tracking problem to that level is probably the hardest challange for AR. Well, maybe true focus is the hardest, but tracking as traditionally been the big thing holding back a lot of serious applications.

Things being too bulky? Not slick? Pft.
Tech gets smaller and slicker. Comfort is important here, but this thing doesn't need to be fashionable. Its a home use thing. Plenty of people happily run Frankenstein monster looking PCs. People with everything in their house looking stylish - those people are the abnormal ones

The functional problem (at least for the actual AR) seems solved.

Nothing else is nearly as important. Even if voice control sucks. Even if gestures fail. Sure, the "package" microsoft is selling falls apart. But if its a windows platform? People will use a pointer device. Or a wireless keyboard.
Or someone else will come up with something.
AR headset with the OS designed for it and near perfect tracking will sale. Using 3D space and interaction rather then 2D screens just makes too much sense control wise for so many uses.

Oh, and eventually, this products descendants - not the Rift- could sale a billion.

===

error -

It also, strangely, has accelerometers so it can handle rotation.

Ah, thats part of the missing information. Thats...weird...Microsoft must have considered some sort mobile use for Kinect fairly early then.

Heres why it wouldnt work without it though;

Imagine theres a wall in front of you. Its flat with no details. You project a square onto the wall. The projector has a camera next to it.
The room is black. All you see though the camera is a square
You turn your head.
The square does not change at all - because the camera and the projector are tied together. Relative to eachother they have not changed.
Theres no way you can tell your information from those facts alone.

The example you posted is awesome - but the software there is doing a lot more work then a normal Xbox does. (at least I dont believe it does)
A normal Xbox to my knowledge needs to do no environment mapping - it simply looks at what points change and knows thats because someone is moving. Its making the assumption
its static itself.
With a Hololens setup, unless theres a secret "no one else in the room" requirement, it cant make that assumption - a group of moving points could be either itself moving, or something moving towards it.
Relatively theres no difference. So it needs to build up a picture of the environment by moving about, identifying common points, and then using those as reference points - slowly disgarding the ones that dont make sense and are either moving subjects or just noise.

[pogo]

Unless Microsoft deliberately excludes them, it will get shitloads. Much smaller platforms have got plenty of support and this has far more potential sales.

Like Windows phone ?

Windows phone isn’t really a small platform. It’s just a lot smaller than Android and iOS. So it attracts far fewer developers. If you are interested in a particular app or just want a wide selection either Android or iOS serve you much better. Which just makes them more popular and so even more attractive to developers. Not just small developers but even big developers ignore Windows Phone as just not worth their while.

MS have particular challenges with this:
* It seems years from release, if ever. What was demoed behind closed doors, and described in various reports, sounds like it will be nightmare to miniaturise and make into an attractive product. The final product may work very differently from the concept video and photos, as happened with Kinect.
* It’s a brand new product category; AR rather than the VR of Rift/Morpheus, not a peripheral but a complete platform. That makes it interesting and sexy but also makes it especially challenging for developers.
* Windows. As with Surface this seems as much about finding new platforms for Windows as making a good device. But we know how Surface and Windows Phone are doing, and extending Windows to yet another totally different platform just fragments and confuses things more.

[Darkflame]

I didnt mean it will have as wide a selection of adds as other things, certainly not, merely wide enough.

Comparatively, iOS even would have a poor app selection compared to WindowsX86 PC. After a certain point it stops mattering.

Provided anyone can develop without big fees or approval process's - ie, like Windows - apps will come.

Currently existing AR platforms on phones work via another bit of software, and even despite that subset of a subset theres still quite a large selection.
Perhaps it will a bit like odd PC devices. The Leapmotion, or my LIFX Bulbs. Both have a quite large app selection now. (despite Lifxs API being rather terrible for a long time)

Hell, Leapmotion has things requiring both a Leapmotion AND a Rift. That cant be a very big market surely?

Now, your not getting any of the big company's involved without big sales. Sure.
But you dont need that. The open-source and tinkering market is enough these days to give you some very impressive stuff. As I said, if no one else does a 3D App expect Blender ported. If photoshop doesnt come expect Krita. The big firms have got pretty lazy anyway. Krita leapfrogged over anything Adobe put out in the last decade with ease.

As long as its easy and accessible for developers, I think thats more important then gaurentied sales.

It seems years from release, if ever. What was demoed behind closed doors, and described in various reports, sounds like it will be nightmare to miniaturise and make into an attractive product. The final product may work very differently from the concept video and photos, as happened with Kinect.

I'm sure it will look different, but what makes you think it would be a nightmare to shrink down? The rig up sounds like it had no attempt at it - exposed circuits, big wires. I dont think they are pushing up against any limits here so much as not yet trying.
Likewise, I dont think they even have to make it look slick. Comfortable yes, but for home use fashion doesnt matter. Not everything needs to be a fashion item.

Power might be a bigger issue - but, again, for home use makes short charges less bother too.

It’s a brand new product category; AR rather than the VR of Rift/Morpheus, not a peripheral but a complete platform. That makes it interesting and sexy but also makes it especially challenging for developers.

Oh, absolutely. There is a lot to learn in this space.
But I think its the correct sort of challange.

Windows. As with Surface this seems as much about finding new platforms for Windows as making a good device. But we know how Surface and Windows Phone are doing, and extending Windows to yet another totally different platform just fragments and confuses things more.

While of course its for buisness reasons, this is a very needed thing for AR. If your going to evolve windows this is a good way to do it because theres a need that you can fill.

So far there has been a large number of "AR Apps" on phones and tablets, but all separate. Some of them call themselves browsers (Layar,Wikitude, 3 or 4 more), but all use bespoke protocols and demand it all routed though their company's servers. Worse, because each app tries to do everything, they are all pretty terrible at doing anything well. And you can only use one at a time.

Moving the AR tracking/overlay to the OS level changes things quite fundamentally. The apps wouldnt have to worry about the hard problems of tracking/overlay/occlusion and instead can devote much more resources to making decent apps. It allows things to evolve separately - driver updates could improve aspects of the tracking/3d registration without app makers needing to expend that effort.

Additionally it lets mulitple apps run at once - the OS can worry about how they render in the view together, the apps don't need to be aware of eachother.

Also, while I am less sure here, I am pretty sure handling the graphic and positioning stuff at the OS level allows vastly better optimistion - maybe even hardware acceleration specially for this purpose.

As for fragmentation;

I dont think its an issue. Windows has always couped with different monitor sizes just fine, and apps in 3D Space are, in the most basic form, just 2D windows with any sized monitor you like. It should be fairly easy for Microsoft to let any windows app be pinned to a wall. Most apps support different font sizes and skinable buttons too, so it wouldn't take much work to make things less precise if people insist on it all being finger based.

3D Apps surrounding you is, meanwhile, a complete new category - so theres nothing to port to start with. It has to be new and (realisticly) only for this platform.

[pogo]

The fragmentation point is it defeats MS’s reason for porting Windows across all these platforms, and so perhaps even makes the device pointless.

I mean Windows 8 was common to PC and Surface tablets and what did it get MS? On tablets it’s a bloated mess, stuffed with desktop crap that limits performance and means smallest available capacity is 32GB, pushing it up into a price bracket where it can’t compete with the far superior iPad. And we all know what happened on desktop:

Windows 8 - Are MS heading for a desktop fall?

And what did it get them? Desktop vendors largely ignored the porting opportunities because of the small market share of both Surface and Windows 8 – for most classic Windows is where it’s at. Now they’ve killed the RT version of Surface and there’s no need to target those APIs, just use classic Windows APIs and your app will run on everything from XP to Windows 8 today and no doubt Win 10 when it’s released.

HoloLens is even worse in the sense it’s even further from both desktop and tablet platforms than they are from each other. It doesn’t matter how many APIs there are in common there are very few opportunities for porting Windows apps to it. If MS sell millions of them then developers will be interested, perhaps porting apps made for Rift to it (as Rift seems certain to ship long before), or even making brand new apps as it’s a new platform with unique features.

[Darkflame]

HoloLens is even worse in the sense it’s even further from both desktop and tablet platforms than they are from each other. It doesn’t matter how many APIs there are in common there are very few opportunities for porting Windows apps to it..

I disagree.
Purhapes you should stop looking at AR as a separate type of pc and start seeing it merely as a different output device. As thats all AR is - a different way to display information. All Windows apps are the moment are essentially just 2D Things displaying information in windows.
A "window" can be stuck in 3D Space fairly easily, hell, windows already does this if you hit the windows icon and hit tab.

With HoloLens, theres nothing to stop Microsoft allowing windows applications - lets say VideoLan - being able to be pinned to surfaces simply at the OS level.
Maybe a painting app pinned to the table while you watch the video on the wall. There should be nothing technically hard to do that at the OS level.

The OS would need to emulate the finger tip as a pointer with some threshold for hover vs click, but thats trivial. (and done elsewhere, like Leapmotion emulates a mouse). Worth noting, that tablets cant emulate hover.

Really Holo is merely a different output form - depending what the app is for the "hard part" might be quite disconnected from the output.

--

Some apps certainly need interface considerations - especially if you want to support Microsoft's vision of everything gesture and point based.

But as long as this thing is running on Windows10 standard architecture. Which I assume is x64, theres no reason at all to think its anywhere near like porting to ARM based stuff.
Making complex apps fully gesture based would be hard, certainly, but thats a similiar problem to tablets really now. Arguably easier because you got more options.

Expect lots of "flat apps ports" first with enlarged interfaces here and there.
Probably true 3d stuff will trickle in at a rate proportional to how easy it is to tell windows10 to display models at a point in space. And, critically, how easy and fast it is to keep them updated.
The wholesetup could be as "simple" as a DirectX like scene being created where the scenes camera is just the real eye view and the background transparent.

Most major 3D Apps have alternative render methods (OpenGL,DirectX and Software are all options in Max) so I would imagine getting their scenes into the HoloLens scene would mean making an alternative adaptation of
one of them.


--

Incidentally, regarding the look of the thing and whats possible;

http://blog.laptopmag.com/meta-pro-glasses-hands-on
(Also remember, this is what was achieved a year ago now, and that is the form factor that group managed to pull off - Microsoft isnt being that unrealistic with what can be done in terms of shrinking)

[pogo]

Vice Uses Virtual Reality to Immerse Viewers in News

A young woman stands in the middle of a crowd of protesters marching up a New York City street, shouting that she is fed up with police brutality, fed up with people saying that black lives don’t matter, fed up with people telling her not to be angry.

“My people don’t deserve this,” she yells, inches away from your face. “We have right to protest. We have right to be angry.”

You look to the left, and see protesters holding a sign reading, “White Supremacy Is Deadly.” Spin to the right, and you see a throng of observers, their expressions stern as the protest continues.

These vivid images are part of an eight-minute virtual-reality experience that catapults audiences into the center of the Millions March protest in New York in December. Created by two experienced directors, Chris Milk and Spike Jonze, in partnership with Vice News, the project is a virtual-reality journalism broadcast. It will make its debut on Friday at the Sundance Film Festival and in a new virtual reality mobile app called Vrse, which is available on the Vice News site.

Very interesting, and is one way VR could move beyond the realm of just gaming. The trick will be to make it more immediate, even to the point of being able to experience live VR broadcasts of news events. I‘m sure that will happen, it could probably be done today with the right kit, a bit of programming and lots of bandwidth.

[gilesgoat]

Anyone heard of this thing ? -> http://nweb.vrunion.com/