Meant to be Seen

Palmer once mentioned he has seen some newer panel technology that hasn't been released yet...how much do you wanna bet hes got his own souped up personal rift at home or in his office.

I bet there is one...and its got a 1080p OLED screen on it...

Oh I think there's zero doubt there Not just a high res panel but an extended field of view and positional tracking.

What fun it would be to visit that little work table

I don't know about positional tracking maybe hes saving that for his second secret rift...

I bet hes got a razer with it...

Extended field of view definitely..eyepieces molded from one solid piece of clear acrylic....BOOM hes got it..

Special super smooth and soft earphone foam for face...
Padded head strap.

extra thin and light hdmi cable...

Positional tracking must have highest priority. Sadly it's a difficult problem to crack, atleast for the level of accuracy we need for proper VR.

Not on his personal one yet....I'm sure hes just got a high res rift...I"m sure positional tracking is on the R and D list...

I bet he has 12000 rifts at home in his flat, refusing to send them out. Just rolling around in the big pile and trying them all on.

Fredrum wrote:...Just rolling around in the big pile and trying them all on.

LOL

You could see from that office video that there were several mobile ones that were picked apart. So they're obviously testing a lot of different screens.

marbas wrote:Positional tracking must have highest priority. Sadly it's a difficult problem to crack, atleast for the level of accuracy we need for proper VR.

Not really, it's easy, see you have 6 strings attached to your head, left-right/up-down/forward-back, these then are attached to the walls in your room. These are attached to little winches on the HMD, problem solved.

NegativeCamber wrote:Not really, it's easy, see you have 6 strings attached to your head, left-right/up-down/forward-back, these then are attached to the walls in your room. These are attached to little winches on the HMD, problem solved.

Thats it? I might detach those 6 strings form my electric guitar and build a positional tracking device instead

You actually only need 3 from the ceiling in a pyramid shape. Same principals as a polar plotter but in 3d.

marbas wrote:Positional tracking must have highest priority. Sadly it's a difficult problem to crack, at least for the level of accuracy we need for proper VR.

Where would be the best 'primer' that would take one up to the minute, in data, for understanding the scope of this problem?

Is there a specific thread where every intricacy of the issue is covered?

I'd need to see a complete problem, in all of it's intricate bits, in order to know if what I'm thinking of, will work, or not.

Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

you mean a modified use of the Razer hydra, with some variation of this? Put the base on this, directly overhead, and the hand piece attached to your head.



I'm actually quite serious about maybe having a potential solution. But I need to know the parameters of the problem, in detail. (Is there a tree, and am I barking at it? Or just fooling myself? What the hell is an I, anyway?) Which is why I'm looking for a thread or such that has all the micro details of the problem laid out. Someone who knows this stuff could probably verbally tell me in as little as a few minutes.

I'm fairly good at finding solutions in other industries, then transferring the solution across into another area of application, and even across fundamental parameters. This is what I have found. I need to know the in depth specifics of the problem with positional tracking. I have some basic understandings of the issue, but not enough.

marbas wrote:Positional tracking must have highest priority. Sadly it's a difficult problem to crack, atleast for the level of accuracy we need for proper VR.

I proposed that topic (difficult to get proper positional tracking as VR needs it) in reddit but was overwhelmed by bad criticts about my "pessimism", WISHFULL THINKING EVERYWHERE

I hope they get the perfect positional tracking of course!

@KBK

The way I see it, there are only two good ways to do positional tracking (of the HMD only):
1. Optical - This involves camera(s) tracking movement in the scene. It's fast. It can be as simple as tracking bright lights in infrared, or as difficult as extrapolating 3D motion from optical flow. Robust/Driftless/Feasible solutions use the former technique. However, when performing only bright light (blob) tracking, line-of-sight issues tend to arise, as well as problems with converting the screen-space position of the blobs into 3D geometry. Applications like TrackIR can track IR blobs (aka fiducials) very quickly. And as long as the user stays within a certain range of motion (where there is a unique solution to the blobs being tracked by the camera), both rotation and position can be acquired with extreme speed and accuracy. But TrackIR needs at least 3 blobs to be visible at all times, and they can not all be on a plane perpendicular to the camera (otherwise the equations mess up) or switch planes. It's a difficult problem because here, because both rotation and translation must be found to just find translation. I believe this problem can be simplified to tracking only two fiducials by taking advantage of the orientation data from the rift's IMUs. One might also find a blob's distance from the camera using parallax by comparing the speed of the blob being tracked with the speed reported by the accelerometers in the IMU.

2. Magnetic - This involves a set of coils (one for each axis) emitting a magnetic field to a similar set of coils (receiving), and measuring the analog output of each of the coils in the receiver to determine position and rotation of the receiver. This lets you get "driftless" 3D tracking without any line-of-sight issues. However, the magnetic system:

• has less range
• has a slower refresh rate (usually)
• requires bulkier equipment (ADCs) on the object being tracked
• is susceptible to interference by ferrous objects which change the magnetic field

Assorted funky techniques like the MadCatz GameTrack also exist.

Fredz Mega Wiki also describes the Positional Tracking problem (with potential solutions) very well in the context of VR.

Bishop51 wrote:Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

I am by no means an expert on this matter. But It sounds reasonable like you say that the pos tracking problem is easier to solve for a limited volume, if that range is good enough for immersion. When pos tracking finally gets integrated for some future vr-hmd, I wish for the range to accurately work for a standing position as for seated with no FOV limits like the TrackIR system is limited by.

But Im still taking Michael Abrash word as a reliable fact that the pos tracking problem is pretty darn hard to solve.

marbas wrote:But Im still taking Michael Abrash word as a reliable fact that the pos tracking problem is pretty darn hard to solve.

I come here from time to time now and I've read about this speech he gave about the state of VR. Does anyone happen to have a video? Thanks.

About Palmer - he has a 270° HFOV prototype. Good enough for me

i wonder if i ever will be able to play a vrmmporg so many problems to solve first "sigh"

Bishop51 wrote:Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

Being stuck in front of a desk without being able to at least rotate on a swivel chair is very poor VR in my opinion. I sincerely hope the near future of VR is free of stupid keyboards, mice and analog sticks.

MrGreen wrote:

Bishop51 wrote:Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

Being stuck in front of a desk without being able to at least rotate on a swivel chair is very poor VR in my opinion. I sincerely hope the near future of VR is free of stupid keyboards, mice and analog sticks.

i totaly agree

Randomoneh wrote:I come here from time to time now and I've read about this speech he gave about the state of VR. Does anyone happen to have a video? Thanks.

I was referring to his statement about pos tracking on the Virtual Insanity Panel at QuakeCON 2012
http://www.youtube.com/watch?v=8gaqQdyfAz8

Kirito wrote:

MrGreen wrote:

Bishop51 wrote:Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

Being stuck in front of a desk without being able to at least rotate on a swivel chair is very poor VR in my opinion. I sincerely hope the near future of VR is free of stupid keyboards, mice and analog sticks.

i totaly agree

I agree aswell, but I also agree with Bishop51. A limited volume positional tracking is alot better than no tracking at all.

Randomoneh wrote:About Palmer - he has a 270° HFOV prototype. Good enough for me

270? Do you have a link to where he mentioned this?

NegativeCamber wrote:Not really, it's easy, see you have 6 strings attached to your head, left-right/up-down/forward-back, these then are attached to the walls in your room. These are attached to little winches on the HMD, problem solved.

Actually you only need ONE string. I have a GameTrak (that I bought for $2), which does it with one string. Unfortunately, it's the PS2 version, and I haven't got a clue what protocol it uses (damn PS2 emulators that don't include USB support).

I'm sure Palmer does have some super Rifts. I wonder what he uses for software for them. They've probably been too busy on the dev kit recently to do much with their super rifts until now though.

zalo wrote:@KBK

The way I see it, there are only two good ways to do positional tracking (of the HMD only):
1. Optical - This involves camera(s) tracking movement in the scene. It's fast. It can be as simple as tracking bright lights in infrared, or as difficult as extrapolating 3D motion from optical flow. Robust/Driftless/Feasible solutions use the former technique. However, when performing only bright light (blob) tracking, line-of-sight issues tend to arise, as well as problems with converting the screen-space position of the blobs into 3D geometry. Applications like TrackIR can track IR blobs (aka fiducials) very quickly. And as long as the user stays within a certain range of motion (where there is a unique solution to the blobs being tracked by the camera), both rotation and position can be acquired with extreme speed and accuracy. But TrackIR needs at least 3 blobs to be visible at all times, and they can not all be on a plane perpendicular to the camera (otherwise the equations mess up) or switch planes. It's a difficult problem because here, because both rotation and translation must be found to just find translation. I believe this problem can be simplified to tracking only two fiducials by taking advantage of the orientation data from the rift's IMUs. One might also find a blob's distance from the camera using parallax by comparing the speed of the blob being tracked with the speed reported by the accelerometers in the IMU.

2. Magnetic - This involves a set of coils (one for each axis) emitting a magnetic field to a similar set of coils (receiving), and measuring the analog output of each of the coils in the receiver to determine position and rotation of the receiver. This lets you get "driftless" 3D tracking without any line-of-sight issues. However, the magnetic system:

• has less range
• has a slower refresh rate (usually)
• requires bulkier equipment (ADCs) on the object being tracked
• is susceptible to interference by ferrous objects which change the magnetic field

Assorted funky techniques like the MadCatz GameTrack also exist.

Fredz Mega Wiki also describes the Positional Tracking problem (with potential solutions) very well in the context of VR.

That's sorta getting there. I found a solution that was just introduced in another arena of development, to deal with noise floor drift. It may be possible to introduce it to this arena. The entire package of this proposed solution must be translated.


It is my understanding, however vague, that it is partially a drift issue. Drift in the components of the IMU. Specifically the magnometer? From my bits of reading (simply coming across it on the forum, not searching it out), that higher sensitivity can almost make it worse?

Linkage1992 wrote:

Randomoneh wrote:About Palmer - he has a 270° HFOV prototype. Good enough for me

270? Do you have a link to where he mentioned this?

I think it was was pre-rift prototype. Threads are all here on MTBS, you can just search for them. If I remember correctly the 270° prototype was really big and bulky, so it isn't really something you could wear on your head; a HMD without headtracking loses a lot of the magic. There's a reason we're getting the current Rift and not that

Bishop51 wrote:Positional tracking isn't really the tough issue everyone assumes it to be so long as you think of it from a limited volume, front-facing, positional tracking standpoint only. If we're talking very small spaces with a limited tracking FOV and range of motion (like facing a user seated at a desk or standing in front of a couch), then the puzzle becomes much easier to resolve. Its only when people start conceptualizing positional tracking that accounts for full body rotation and huge capture volumes that it becomes complex.

For my part I would be happy with a system that tracked position much like TrackIR. Or better yet, you could get away with some limited supplementary positional tracking with a magnetic base station like the Hydra built into the Oculus breakout box. The range doesn't need to be huge, just a small volume of space with enough room to crouch, lean and peer.

I know lots of people want a perfect Matrix-esque experience with free roaming, perfect body representation, etc. While I think all those things are awesome, and people should definitely be doing research to further this, in reality it's just not practical when you're playing in your living room where you don't have the space (or safety) and also don't want some cumbersome setup of various gear you have to wear, or external equipment that you have to properly set up. These kinds of things would be better left for VRcades where all the drawbacks can be justified, and a proper experience can be provided.

I'm completely with Bishop51 here about a much more limited and practical solution to address the issues holding back a good mass consumer home experience. I too believe that it can be had now (or soon) relatively cheaply. I started working on a thread which I haven't got around to finishing to discuss a bare minimum approach to positional tracking while addressing some aspects that I believe people have misconceptions about that don't apply in a constrained environment.

2EyeGuy wrote:

NegativeCamber wrote:Not really, it's easy, see you have 6 strings attached to your head, left-right/up-down/forward-back, these then are attached to the walls in your room. These are attached to little winches on the HMD, problem solved.

Actually you only need ONE string. I have a GameTrak (that I bought for $2), which does it with one string. Unfortunately, it's the PS2 version, and I haven't got a clue what protocol it uses (damn PS2 emulators that don't include USB support).

I'm sure Palmer does have some super Rifts. I wonder what he uses for software for them. They've probably been too busy on the dev kit recently to do much with their super rifts until now though.

Haha, even though I was kidding that GameTrack thing looks pretty cool!

The distance of the tracked element from the mechanism is determined through components which measure the rotation of the spool drum for the retracting cable reel, and calculating how far the cable is extended. Through the ball joint and guide arm, the mechanism functions in a similar fashion as a gamepad analog stick[citation needed] to determine the angular direction from the mechanism to the track element.[2] From the distance and angle data, a three-dimensional position for the element is resolved. The predetermined spacing and orientation of the mechanisms on the base unit allows the coordinate data gathered by the two mechanisms to be converted into positions in a unified space. According to In2Games, the mechanisms can determine position "to an accuracy of 1 millimetre anywhere within a 3m cube around the unit, with no processor overhead or time delay.

Only thing is, having a string attached to your face might not be the best solution.

Perhaps he has already one of those. I just read today in a german atricle that Japan Display and tablet manufacturers want to build in Retina displays into tablets already 2014.
Anyone knows what the average density of such an display is? I assumed 2.0 g/cm² because it mainly consits of relatively light silicon?, LCs and plastic. For an 7 inch, 1,2 mm thick display i calculated an astonishing weight of just 25 gramm!
For those who dont believe it: 2000 gramm for a 1mm thick 1 square meter panel.
a 7 inch panel has ca. 120 cm², a square meter 10,000. So the panel is 1 percent of it. So 1%
of 2000 gramm are 20 gramm. The panel measures 1.25mm. So approx. 25 gramm. WTF!
[youtube-hd]https://www.youtube.com/watch?v=2yGsWT9wfck[/youtube-hd]
when assuming a average density of 4 g/cm,² what is very dense, it would be a ultra light display weighing just 50 gramm.
Oculus need to get their hands onto one of these displays.


EDIT: Heres a link to a 10.1 inch panel which has double the area and is twice as thick, but only weighs 130 gramm.
http://www.j-display.com/english/news/2 ... 21029.html
So my calculations might not be that wrong.

....He's got a little two bedroom apartment...and the second bedroom is his VR room...
He refers to it as the "VR Dungeon"

He probably has no lights in this room....except one... a old flickering black light....dark shag carpet...the windows all covered up so not even moonlight gets through...

and there in the corner sits his desk...on top is an i7 gaming computer with 16 gigs of ram... twin Nvidia 690 video cards in SLI configuration..revo drive on his computer hard drive... next to the large 27 inch monitor sits a locked closes aluminum case labeled "SECRET RIFT"

Cookie and pizza boxes all over the floor....Signed posters of lawnmower man, strange days and the matrix scattered about the walls...

and that is where he dwells...late on a Friday night....to play Bioshock Infinite on his super secret high res 7 inch rift. While dressed in pajama pants and Homer Simpson slippers....

The game quieted by the headphones....and all that can be heard are the fans of the computer
... and a soft quiet maniacal giggle.

2EyeGuy wrote:

NegativeCamber wrote:Not really, it's easy, see you have 6 strings attached to your head, left-right/up-down/forward-back, these then are attached to the walls in your room. These are attached to little winches on the HMD, problem solved.

Actually you only need ONE string. I have a GameTrak (that I bought for $2), which does it with one string. Unfortunately, it's the PS2 version, and I haven't got a clue what protocol it uses (damn PS2 emulators that don't include USB support).

Jan has done a great guide for modding the Rev.1 / PS2 version: http://janoc.rd-h.com/archives/129

Sadly he's all out of PCBs, maybe he could do another run if there's enough interest?

I just picked-up the PC version ($1 on eBay!) and am going to give it a try with the Rift soon, would appreciate your input: http://www.mtbs3d.com/phpbb/viewtopic.php?f=138&t=15717

squibbfire wrote:twin Nvidia 690 video cards in SLI configuration

I'm actually pretty sure he mentioned in an interview that he's running 3x Titans. The value of that computer... geez

Fredrum wrote:I bet he has 12000 rifts at home in his flat, refusing to send them out. Just rolling around in the big pile and trying them all on.

bwahaha

He has loads of cool poop. In on interview he spoke about electrodes that he has that you can place behind your ears, and when correctly stimulated they can simulate the sensation of movement. If that could be fine tuned you would never need to leave your chair and you could run around all day, because after all the sensation of movement is simply your brain interpreting signals from your inner ear and your vision. Oculus already has one of those down, and I'm sure they are working on the other. Although I think simulating the sensation of moving forward in a car or something similar is one thing, the sensation of walking or running is another, altogether more complex problem; and an experience that would need to be heavily tailored to the individual, not to mention the lack of impact feedback throughout your body, your feet in particular. Im sure it will happen one day though.

I would be happy to see a 2560x1600 OLED panel with 170-180 horizontal FOV around 2016.

I'm sure he's experimenting to give us the best product.

superbike81 wrote:I would be happy to see a 2560x1600 OLED panel with 170-180 horizontal FOV around 2016.

I'm sure he's experimenting to give us the best product.

The annoying thing is we absolutely have the technology to build that panel right now, its just procuring it that's the problem, Oculus isn't big enough to commission its own spec panels.

Yeah, I think it would be possible for sure. Might double the price of the Rift but I'm ok with that. The problem for now is even my two gaming systems (i7-2600k with 7970 and i5-3570k with 680) would struggle to do stereoscopic 3D at that resolution and maintain high FPS. I think 2016 is a safe bet to where mid-range cards should be able to handle that without a problem.

superbike81 wrote:Yeah, I think it would be possible for sure. Might double the price of the Rift but I'm ok with that. The problem for now is even my two gaming systems (i7-2600k with 7970 and i5-3570k with 680) would struggle to do stereoscopic 3D at that resolution and maintain high FPS. I think 2016 is a safe bet to where mid-range cards should be able to handle that without a problem.

or a Nvidia Volta can run Brigade Engine powered games in 3d and at 1080p.