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

[Dvorak]

'95.
Yep. I do remember.

My bible back then.

[edit] Still in my bookshelf.

[gumboot]

Because my eyeballs have lenses attached at a point other than their centres, my field of view changes as I swivel them about. It's rare that I'm in a position where I can see the difference that makes (don't know how much is brain compensation and how much is insignificant effect).

Effectively it means that I can see slightly further around obstacles at my sides by looking straight ahead and using my peripheral vision. If I look directly at the same object then my point of view changes by a few millimetres and I can no longer see it.

Does this totally screw up all the optics in the small-scale world of VR goggles? It seems like it's something you'd have a horrible time compensating for as you look through different parts of the lenses as you look around.

Or is the human brain on top of all that? Certainly I don't notice the effect of my glasses as I look around the room; but they're pretty simple lenses and they don't cover my peripheral vision.

[pogo]

I don't see how adjusting for the position of the lens within your eye will make much difference. You can only see detail straight ahead, stuff in your peripheral vision is too ill defined to be of much use. Our eyes are set up to ignore detail to the sides but detect movement, after which we can move our eyes, head or both to look directly at whatever has caught our eye, which of course the Rift supports.

Besides I don't think it will be doing eye tracking any time soon. Apart from the above the other reason suggested, that you only need to render in detail where the user is looking, seems an unnecessary optimisation. With GPUs getting better and cheaper all the time a decent gaming PC (and anyone buying these should have a decent gaming PC) will be able to drive whatever resolution's in these.

Head position tracking's another matter. Reportedly the next version will have this, and it will work best if tracks the position of the eyes, not the centre of the head or the screens, but I don't know if it will be that precise.

[merman]

I have this feeling that immersive VR is going to be like 3D cinema - popping up every decade, promising new and better tech before fading into the background again...

[gumboot]

I don't see how adjusting for the position of the lens within your eye will make much difference. You can only see detail straight ahead, stuff in your peripheral vision is too ill defined to be of much use. Our eyes are set up to ignore detail to the sides but detect movement, after which we can move our eyes, head or both to look directly at whatever has caught our eye, which of course the Rift supports.

The closer things are the bigger the difference it makes. I've already established that it changes the origin of your view by a few millimetres (I don't know how many; perhaps the whole radius of an eyeball), and what I wonder about is if all of these tight geometric calculations aren't put off by the change.

[pogo]

The goggles may be close but they're simulating something far away, so there's no need to modify what you see – the effect would be negligible.

The only question is whether, because the screens are so close, you'll notice the scene 'shifting' because it's drawn so close to your eyeballs. I suspect not for most people as it's not something they look for. Your won't be keeping your head fixed but will be turning and moving it which will affect what you see far more significantly. It also won't be that close as it has to allow for both people's differing degrees of nearsightedness/farsightedness and glasses - I'm near sighted so have no problem with things being near but also have astigmatism so need glasses no matter how close things are.

[gilesgoat]

Guys I said it, "you don't want to listen" but I'll say it AGAIN : PEOPLE WITH LENSES HAVE PROBLEMS WITH VR !

Oculus WILL give you problems with lenses and even without lenses if you don't learn to "move your head DON'T move your eyes" while looking you WILL have problems, it takes a certain time to get used to find the best from it.

Unless you have a 20/20 perfect vision and you don't wear anything I believe that *ANY* ( current tech ) VR system "will give you some sort of problem".

Read page 4 or so of the OR SDK manual and you'll see they mention something about, it specifically says something you are going to loose if you wear glasses and VR.

Said that, it still works .. "acceptably" BUT it's *NOT* perfect !

Fundamentally :

- you have to learn to keep your eyes fixed/focused in a position

- you have to learn to cope with "move your head not your eyes"

- you have to learn to cope/get used to "even in the best conditions" it's NOT a "natural vision"

- you'll get tired after some time/experience anyway, this is NOT kind of stuff you can wear 10 hours without a problem

Hopefully as technology will improve and such bit at the time problems will be solved and things will get better.

BUT .. to state NOW that "no there are no problems" or "the problems are minimal" it's a LIE !

"Yes there are problems".

If you don't believe me, take an OR and put in on, with glasses and use it.

And don't say I did not warn you.

[pogo]

- you have to learn to cope with "move your head not your eyes"

This makes some sense. It makes no sense to cover the whole width and height of where your eyes can look by rotating in their sockets. Maybe half of that is too far to the left or right to be visible to both eyes, so when looking that way one screen would be inactive. Also you have to deal with curvature and distortion.

So instead it will be more like goggles in a old style diving suit, or the eyeholes in a suit of armour. You have little to no peripheral vision and have to turn your head to see things. This works well for a game as it knows what you're looking at, so can update the view and perhaps trigger events such as AI char's returning your glance - this could be a useful gameplay mechanic, to let you interact with characters without a mouse/keyboard for example.

This means also that the eye can rotate much less, and so problems due to its displacement should be much less severe.


They need to fix any problems with using them with glasses, which are required by a large fraction of the population. You can use contact lenses or have your eyes modified by lasers but these are expensive and used by few, especially outside the developed world. Or if they don't fix such problems someone else will, such as Sony or yet another company.

[Darkflame]

To be honest, I'd just like to have an immersive (ie, fills my field of view) stereo effect without any head tracking. That would be a great start imo, and head tracking could be worked on while people play with that.

Here's why I don't agree:


I think stereo is more of a gimmick that falls out of being much, much easier to implement with recording devices. For rendered content on devices more expensive than cellophane and cardboard you really need to re-evaluate your priorities.

Id like to say I downloaded and played with Johnny Lee's exact demo there.

ABSOLUTELY headtracking is a BIG DEAL. You instantly got a much stronger 3d effect with even subtle motions of your head with it.
And its instinctual too - you lean forward back/left right a little naturally "too look around".

However, I also tried it with stereoscopic shutter glass's at the same time, and the effect was naturally best when combined.

Basically, I think anything off will tell your brain its 2D.


Anyway...Microsoft failed to take advantage of Lee while they had him, he's now at Google working on the awesome looking project Tango;



AR>VR

[Tanatoes]

You know, I check in o n Rift developments once in a while and watch dev videos and stuff - but to date I haven't seen a Kinect sensor bar strapped to one. That would be my expected next development: giving the Rift info about your surrounding environment and feeding it back into the system. (Primarily to read hand movements and use them Minority Report style to manipulate images inside the Rift.)

Of course you'd have to develop a much lighter Kinect bar - neck strain is already going to be an issue with extended Rift play I think.

Somebody out there must be developing patents on the motions of AR/VR right? Like Apple owns the "pinch to zoom and rotate."

[pogo]

I think the other way round, with the camera fixed, makes more sense. Having a sensor attached to the device is a much more complex proposition as it's got to on the fly either do quite complex calculations or pass the information back to the PC/console to do those calculations. You also have a problem with placing it so it can see your hands - or you ask the user to always hold their hands ahead of them which could get very tiring.

A camera on the PC/console makes much more sense. It's Sony's approach, and it lets the camera track your position (perhaps reducing the need for position detectors in the headset) as well as your hands (via Playstation Move controllers which are PS4 compatible, or the current DualShock controller with its lightbar).

This is with current technology. Maybe in a few years when you can fit the CPU, GPU storage of a console inside a headset, so it's a completely autonomous unit, doesn't need a PC so needs its own camera(s). Even then though they might work better if they're separate from the headset and talk to it wirelessly.

Rift doesn't seem be doing anything like this. Maybe because their device is cross-platform and there's no single camera that will work with all platforms, maybe due to the extra expense, or even both. Or it could be something they decide to do when it eventually goes on sale to end users.

[Darkflame]

For headtracking it makes more sense to be static, for handtracking (etc) you want it on the device.

Meta/Spaceglass has a leapmotion-esq sensor design;
https://www.spaceglasses.com/

Theres also camera only (no-depth) based sensing tech for gesture recognition on Glass. Cant remember who makes it though, pretty recent.

I think ideally, Tango is the way to go - once you mapped the environment tracking your position and orientation within it should be trivial.

Somebody out there must be developing patents on the motions of AR/VR right? Like Apple owns the "pinch to zoom and rotate."

God I hope not.
Two-point zoom/and rotate (which is what pinch to zoom and rotate basically is), was already well established in stuff like Autocad. Hell, its basic geometry. Letting this patent "creep" to define something as different merely because its done with your hands or body could get ridiculous pretty fast.

Not just Apple, of course, this lot already exists;
http://io9.com/5808604/10-physical-gest ... n-patented

I think I heard dance moves cant be patented, maybe that can be expanded to "human body motions" or something.

[pogo]

I don't think patents will be a big concern. Almost uniquely among "new tech" VR has been around for a very long time, for decades. It's appeared occasionally as a commercial product but it's been in academia for all this time. People doing research and inventing things. Maybe patenting them, but long enough ago in many cases that patents will have expired, or not but establishing prior art.

So chances are many major innovations in VR have already been discovered. We don't know about them as they're not on sale yet, so to us it looks like a new technology with problems which require technological solutions, so patents. But its likely we'll see these innovations, in the hardware and software that runs on it.

[Darkflame]

Well, with any luck it will be like web-browsers (which have a copying-free-for-all, and we have been all the better for it).

[Baggypants]

I am the only person who keeps reading AR as "Actual Reality" and then immediately wonder why you need a screen and a camera to see what your already seeing?