DK2 Development kit

I've got to say, there is one person on here, that when he pops up on a thread I just think oh no...

If this stayed on topic it would be interesting.

I must say I'm expecting a DK2. We have two DK1s and we have gone as far as we can with them. Our project is parked until an updated device comes out.

We really are dependant on the final resolution and head tracking. We are not a games company.

So if there is a DK2 with these features in, we will buy them and continue our R&D. And we saw that powerpoint slide a long time ago and still expect a further release before the consumer version. If we are wrong then fine, but we can't go further until the next release comes out.

"andyr123" wrote:
I've got to say, there is one person on here, that when he pops up on a thread I just think oh no...

If this stayed on topic it would be interesting.

I must say I'm expecting a DK2. We have two DK1s and we have gone as far as we can with them. Our project is parked until an updated device comes out.

We really are dependant on the final resolution and head tracking. We are not a games company.

So if there is a DK2 with these features in, we will buy them and continue our R&D. And we saw that powerpoint slide a long time ago and still expect a further release before the consumer version. If we are wrong then fine, but we can't go further until the next release comes out.

A 1080p Rift is only twice as many pixels as a 720p Rift (or perhaps a bit more useful area depending on screen size and optics). You can use a head-mounted hydra for position feedback. No need to park your project waiting for a significant new iteration of affordable Rift hardware, unless you have a non-technical (personal or political) reason to do so...

Just continue your project with industry-standard modular software design allowing variations in resolution and head-tracking hardware...

Or not... Your choice...

If your project cannot tolerate the contraints of the Rift DK1, it may not be satisfactory with even the consumer Rift. Less screendoor effect, more pixels, positional head tracking, sleeker design, but technically not THAT much different from what we already have with a Rift DK1 and a Razer Hydra. How do you know that the final design of the consumer Rift will be good enough for your needs, if the currect hardware is not? Some of us will design around such limitations, and you can too... Or use a different product that does suit your needs.

"geekmaster" wrote:
If your project cannot tolerate the contraints of the Rift DK1, it may not be satisfactory with even the consumer Rift. Less screendoor effect, more pixels, positional head tracking, sleeker design, but technically not THAT much different from what we already have with a Rift DK1 and a Razer Hydra. How do you know that the final design of the consumer Rift will be good enough for your needs, if the currect hardware is not? Some of us will design around such limitations, and you can too... Or use a different product that does suit your needs.

Agreed, It would also help a bit if he at least described the project and why it's so dependent on resolution.
Is it because he needs to resolve details from far away? or have small text be readable for some reason?
If we at least knew that, maybe someone could give him some advice.

It's like geekmaster said: the resolution will get better, but it's not going to change the fact that it will still seem low resolution to anyone who requires high pixel density.
With the rift, the actual pixel density your eyes get will be much, much lower than what people are used to on their monitors, at least until we get 4K displays; at which point pixel density should at least be somewhat comparable to what we're used to with our monitors.

But like i said before, maybe if you describe why resolution and positional tracking are so important we could point you in the right direction...

Hi, yes small text is one is of the issues. We are a financial trading/modelling firm.

At the moment its an R&D project , to see if our idea is possible. There are other issues also to overcome before we can invest fully in the project.

FOV is less important, but position and resolution are. I take your point about hydra, but we need firmer details on what the end product will look like. A dev kit 2 would help a lot as we could move forward some more.

I think you can see some of the issue here. The demos coming out are not offering anything new, I think they have stalled. I'm not saying games companies cannot progress, but there are still some big details to reveal.

"andyr123" wrote:

FOV is less important, but position and resolution are. I take your point about hydra, but we need firmer details on what the end product will look like. A dev kit 2 would help a lot as we could move forward some more.

I think you can see some of the issue here. The demos coming out are not offering anything new, I think they have stalled. I'm not saying games companies cannot progress, but there are still some big details to reveal.

I don't see Oculus changing their dev-kit pipeline anytime soon now that they are finally meeting shipping estimates. Plus the HD oculus making the rounds probably looks nothing like what will be released anyways.. Which I would assume might have a higher resolution display than 1920, stereo cameras, position tracking, and all sorts of surprises.
I think the HD oculus really represents a great future carrot for the press/conferences while allowing oculus to keep the actual future specs close to the chest. Gotta play the hype game a bit- it wouldn't work so well if everyone had access to the HD unit now.

Maybe when the stereo-cameras/position tracking gets locked down we might see a second iteration. (or an add-on for the current devkit owners)-

If you need better legibility for current R&D right now, check out the new supersampling/MSAA features available with the newest SDK. It improves text legibility by 2-4x

viewtopic.php?f=37&t=3045

"antigravity" wrote:
Maybe when the stereo-cameras/position tracking gets locked down we might see a second iteration. (or an add-on for the current devkit owners)-

I wouldn't hold my breath for stereo cameras. Apparently it has been attempted, but the problem is that rendering real world imagery into the Rift has an even lower latency tolerance than rendering generated imagery, and imagery generated from cameras that don't have their viewpoint where the brain expects it to be (relative to the base of the neck where rotations occur) can exacerbate the problem.

I've already tried a connected webcam (though not in stereo) and I personally don't have a problem with it, even with a wide field of vision, though I'll admit it's pretty difficult to walk with the latency (it's very much like being drunk).

"jherico" wrote:

"antigravity" wrote:
Maybe when the stereo-cameras/position tracking gets locked down we might see a second iteration. (or an add-on for the current devkit owners)-

I wouldn't hold my breath for stereo cameras. Apparently it has been attempted, but the problem is that rendering real world imagery into the Rift has an even lower latency tolerance than rendering generated imagery, and imagery generated from cameras that don't have their viewpoint where the brain expects it to be (relative to the base of the neck where rotations occur) can exacerbate the problem.

I've already tried a connected webcam (though not in stereo) and I personally don't have a problem with it, even with a wide field of vision, though I'll admit it's pretty difficult to walk with the latency (it's very much like being drunk).

It was the image below that let me think it was on their radar- ironic since this thread is about debunking prediction powerpoints. lol - so yeah maybe it won't make the final cut.

But still, while I get that there may be latency issues for walking around and flipping your head side-to-side.. How about a game of battle chess on your desk?? There are plenty of experiences you could generate that would blow other AR experiences out of the water that don't require a perfect sync.. Personally I'll never forget making a dragon crawl out of my wife's belly with the Nintendo 3DS.

"jherico" wrote:

"antigravity" wrote:
Maybe when the stereo-cameras/position tracking gets locked down we might see a second iteration. (or an add-on for the current devkit owners)-

I wouldn't hold my breath for stereo cameras. Apparently it has been attempted, but the problem is that rendering real world imagery into the Rift has an even lower latency tolerance than rendering generated imagery, and imagery generated from cameras that don't have their viewpoint where the brain expects it to be (relative to the base of the neck where rotations occur) can exacerbate the problem.

I've already tried a connected webcam (though not in stereo) and I personally don't have a problem with it, even with a wide field of vision, though I'll admit it's pretty difficult to walk with the latency (it's very much like being drunk).

I believed that stereo camera weren't to restitute reality, but to make a positional tracking by analyzing the changes of the images captured

"antigravity" wrote:
It was the image below that let me think it was on their radar- ironic since this thread is about debunking prediction powerpoints. lol - so yeah maybe it won't make the final cut.

My impression is that they tried it and it didn't work out for the reasons I mentioned. Perhaps the issues can be overcome eventually, but I suspect that slide is from before testing or early in the testing. On the other hand it's possible they're just not saying they've solved the problem because of competition.

"antigravity" wrote:
But still, while I get that there may be latency issues for walking around and flipping your head side-to-side.. How about a game of battle chess on your desk?? There are plenty of experiences you could generate that would blow other AR experiences out of the water that don't require a perfect sync.. Personally I'll never forget making a dragon crawl out of my wife's belly with the Nintendo 3DS.

Sure, there's less motion to interacting with a nearby game board than walking around through a convention hall, but that's no guarantee that the problem of mismatch is less. In fact, it's entirely likely to be more. If you're walking down a hallway your brain is probably putting fewer constraints on how it expects the world to move than if you're using fine motor control to make tiny head movements.

I think the solution to that is not to try to render the real-world imagery into the display, but rather to construct a 3D scene either via registration marks like QR codes or a depth sensing camera and then render that fake scene (perhaps with textures and or colors culled from the real imagery) so that it's exactly at the correct depth and has exactly the correct motion when you move your head.

"Tgaud" wrote:
I believed that stereo camera weren't to restitute reality, but to make a positional tracking by analyzing the changes of the images captured

Doing that, particularly in realtime, with arbitrary real world imagery, is an extremely hard problem in computer vision.

Consider: the original Kinect uses distortions in a projected known pattern of light it projects to construct 3D features. The XBox One version uses a time of flight camera to do the same. If these devices could use dual color cameras to reconstruct 3D features they would, because it would be much cheaper in terms of hardware.