Rift, Vive, and Virtual Reality

[hughJ]

 

[manux]

Interesting novelty

Melbourne-based firm Zero Latency has raised $7 million in venture funding. The company specialises in out-of-home ‘warehouse scale’ local multiplayer VR experiences, and is beginning to open attractions around the world

http://www.roadtovr.com/zero-latency-7-million-investment-expansion-warehouse-scale-vr-tech/

 

[hughJ]

I have to imagine that with the benefit of hindsight Oculus would have opted to find an alternative tracking system to IR cameras, especially if they foresaw how well Lighthouse ended up working coupled with the draw of "roomscale" VR.

 

[manux]

This is seriously cool. Maybe time to install and try to tinker around.

Editor VR is designed to make developing both VR and non-VR content much easier by allowing creators to make games without needing to remove their headset. It uses the Vive Wands and Oculus Touch controllers for accurate, intuitive control, allowing you to access traditional settings via 3D windows and also take advantage of new ones like faster object placement. All of the features we’ve seen on-stage so far are expected to be included in the experimental release, though we’ve reached out to Unity to confirm this.

http://uploadvr.com/tomorrow-can-make-games-within-vr-unity/

 

[hughJ]

I think we probably need another iteration or two of resolution to make the UIs more usable, but for quickly blocking out environments they might be okay. Probably much the same as with asset creation in something like Medium - the 3D depth and 6dof interaction is likely much faster for early stage work, quick concept sketching, but at some point you need the precision of a mouse, high PPI monitor, and the repetitive actions mappable to keyboard shortcuts.

 

[hughJ]

https://scontent.xx.fbcdn.net/t39.2365-6/15363893_1774761836111478_5342883442994446336_n.pdf

Reading this guide for Oculus's 3-camera roomscale setup I can't help but hear the voice of Vincent Price saying "It's as easy as one, two, three."

 

[eastmen]

https://scontent.xx.fbcdn.net/t39.2365-6/15363893_1774761836111478_5342883442994446336_n.pdf

Reading this guide for Oculus's 3-camera roomscale setup I can't help but hear the voice of Vincent Price saying "It's as easy as one, two, three."

its more on the fact that a lot of boards just use one controller for usb 3 so 3 cameras on it will over load it.

I finished setting up a friend with the intac card and 3 sensors running usb 3.0 with the headset also in it all work in usb 3.0 mode fine. Set up took a few minutes after we ran the wire (he has carpeting so we ran it under the carpet by the base molding.

 

[hughJ]

I recognize the cause - I was more teasing on how obviously untenable this technology approach is to roomscale tracking. If outside-in IR camera tracking didn't seem like a dead-end before, it certainly does now. More seriously though, it makes me wonder just what sort of future the Touch controller can expect to have. Previously I was thinking that Oculus could get away with keeping Touch and carrying it over to a gen2 Rift, but now I'm thinking that they'd be a whole lot better off by scrapping all camera-based tracking and moving to a Lighthouse-derivative system. If they're already pushing the bandwidth limits for USB controllers, then it's not like they have room to increase the frame rate, resolution, or most importantly their vertical FOV of the cameras. Lighthouse on the other hand can be updated in any number of ways and likely do so while being backward compatible.

 

[eastmen]

I recognize the cause - I was more teasing on how obviously untenable this technology approach is to roomscale tracking. If outside-in IR camera tracking didn't seem like a dead-end before, it certainly does now. More seriously though, it makes me wonder just what sort of future the Touch controller can expect to have. Previously I was thinking that Oculus could get away with keeping Touch and carrying it over to a gen2 Rift, but now I'm thinking that they'd be a whole lot better off by scrapping all camera-based tracking and moving to a Lighthouse-derivative system. If they're already pushing the bandwidth limits for USB controllers, then it's not like they have room to increase the frame rate, resolution, or most importantly their vertical FOV of the cameras. Lighthouse on the other hand can be updated in any number of ways and likely do so while being backward compatible.

Why ? Just increase the sensor FOV currently its 120x70 . Also 4 Sensors work well enough in large rooms so they could simply get the price down while newer hardware will handle multiple sensors better. Or you might need less cameras if they have sensors inside the headset themselves

 

[manux]

Probably current touch is good enough for most users. I assume most people have fairly small clear area available. I see touch as the entry level solution that should carry over to next gen.

Teleport seems to be pretty decent in most games so even fairly small area+teleport is doable. I do see the value in vive lighthouse solution and see it as desirable for whomever has the luxury of large free space. Teleport should be able to bridge the gap between different tracking solutions.

I finished I expect you to die. It was very good but only 4 levels is pretty steep ask for 30$.

 

[pjbliverpool]

Yeah the camera's are a particularly annoying part of VR. My setup doesn't allow me to have them permanently in place so I have to move them very time I want to use any type of VR that requires a bit of movement. They do work very, very well once setup though. As long as you don't face away from them!

 

[manux]

Real estate sales on vr. Unfortunately works only on mobile devices. This could be the future. It would be nice to have 360 3d pictures/video to augment the realtime rendered graphics

https://hollyhaca.herokuapp.com/index.html

 

[hughJ]

Why ? Just increase the sensor FOV currently its 120x70 . Also 4 Sensors work well enough in large rooms so they could simply get the price down while newer hardware will handle multiple sensors better. Or you might need less cameras if they have sensors inside the headset themselves

Increasing the FOV means sacrificing more precision, unless of course you also increase the camera resolution to compensate, but it seems to me like we're well beyond reasonable bandwidth usage already. Each camera can't cover the full room volume and the fact that the tracking precision drops off so dramatically over distance means that the extra cameras aren't just there to avoid cases of occlusion, but to help compensate for its own shortcomings (distance tracking and the vertical FOV). Having inside-out sensors in the headset would still leave you with having to track the controllers through some other means.

 

[eastmen]

Increasing the FOV means sacrificing more precision, unless of course you also increase the camera resolution to compensate, but it seems to me like we're well beyond reasonable bandwidth usage already. Each camera can't cover the full room volume and the fact that the tracking precision drops off so dramatically over distance means that the extra cameras aren't just there to avoid cases of occlusion, but to help compensate for its own shortcomings (distance tracking and the vertical FOV). Having inside-out sensors in the headset would still leave you with having to track the controllers through some other means.

That's my point , instead of having cameras track the HMD + controllers you only have the cameras to track the controllers + the sensors inside the helmet that are inside out tracking. You should be able to eliminate the need to higher persicion since your tracking less while having more sensors to track. .
Most of the time the controllers should be in view of the sensors on the helmet , the other outside sensors will be for those few times when its not

 

[hughJ]

Most of the time the controllers should be in view of the sensors on the helmet , the other outside sensors will be for those few times when its not

I don't think that statement can be made so easily and just assumed that it will work. Tracking the hands from the HMD itself would be plagued with occlusion from your arms and shoulders to such a high degree that you would have to assume that in most instances it wouldn't work, which puts you right back to the same situation where the external cameras are required to cover the full volume of the room. On the other hand you could just use something based on Lighthouses which allows you to scale the coverage and precision in any number of ways. My issue is not just that cameras are poor right now, it's that if I wanted to improve their precision by 2x or 10x over the coming years that there's no obvious way to do that. The idea that having an HDD or some other USB peripheral suddenly access the USB controller is potentially enough to cause tracking to collapse is absurd - it seems to me a fundamental misuse of a technology that's intended to provide a shared bus across many devices simultaneously.

 

[eastmen]

I don't think that statement can be made so easily and just assumed that it will work. Tracking the hands from the HMD itself would be plagued with occlusion from your arms and shoulders to such a high degree that you would have to assume that in most instances it wouldn't work, which puts you right back to the same situation where the external cameras are required to cover the full volume of the room. On the other hand you could just use something based on Lighthouses which allows you to scale the coverage and precision in any number of ways. My issue is not just that cameras are poor right now, it's that if I wanted to improve their precision by 2x or 10x over the coming years that there's no obvious way to do that. The idea that having an HDD or some other USB peripheral suddenly access the USB controller is potentially enough to cause tracking to collapse is absurd - it seems to me a fundamental misuse of a technology that's intended to provide a shared bus across many devices simultaneously.

I'm sorry did you read what I wrote ? I wrote that the hmd sensors would work in conjuction without outside sensors. You would just need fewer outside sensors.

Also the sensors run fine in usb 2.0 mode now. The 3rd sensor ships with a usb 2.0 extender

 

[hughJ]

We're talking about tracking the head and hands 100% of the time, not 95% of the time. There's every reason to assume that trying to track the hands from the HMD would be plagued with occlusion issues, so having such a system work in conjunction with external cameras does not mean that you need fewer external cameras, therefore you're still plagued with the same USB bandwidth constraints and camera coverage shortcomings we have right now.

If the sensors "run fine" in USB 2.0 then they should never have forced the USB 3.0 spec for their architecture and gone through all the hassle of requiring customers to buy new motherboards and/or additional PCIE USB 3.0 with specific controllers on them. "Runs fine" is good enough for an "experimental" feature mode that's not the officially supported design of their system, but nothing about it seems sustainable or likely to be built upon for future 360 tracking implementations. Throwing in a monoprice USB 2.0 extender and relaxing their mandated USB 3.0 tracking in attempt to chase after the functionality that the Vive provides is a well appreciated band-aid fix, but it's still a band-aid fix, and I think it puts them behind the eight ball when it comes to iterating on this architecture for gen2.

 

[pjbliverpool]

Have any of you guys tried Google Earth VR? It's pretty awesome. It's officially Vive only but can be easily spoofed to work on Rift.

 

[hughJ]

Yeah, I've tried it on the Vive. It's pretty cool, although I think the mileage you get from it depends an awful lot on whether the places you want to visit have geometry data for the structures. There are a lot of places/streets/etc I would like to revisit from my childhood for example, but in this they're all pancaked blurry textures. I figured something like the Giza pyramids would be a safe bet to experience real world scale - but again, no geometry for those either. It sort of reminds me of when Google Earth/Maps first launched - pretty rough around the edges, but it's easy to imagine how it could be a platform for something amazing in a couple years.

edit: I would also say it's the first application I've used that makes a pretty compelling case for gigabit fiber. It'll never be feasible to cache all that data locally, so you're basically stuck with streaming an earth sized amount of photo data on the fly.

 

[manux]

On the lowest end of the spectrum, the new HMDs will offer 1200x1080 resolution per eye. That’s striking, because that’s the same per-eye resolution of the two titans of the industry, the Oculus Rift and the HTC Vive. On the high end, Microsoft expects its hardware partners to offer up to 1440x1440, which would outpace the Rift and Vive.

However, note that a major difference is that these mainstream HMDs will offer just a 60Hz refresh rate at that resolution (versus 90Hz for Rift and Vive), which could prove problematic for the user experience. Again, however, at the high end, these new HMDs should meet or exceed that refresh rate.

http://www.tomshardware.com/news/mainstream-vr-hmds-intel-microsoft,33217.html