Cost-Effective Stereoscopic 6DOF VR Simulation Build

[vvortex3]

I am very interested in any input regarding my VR simulation. Especially pertaining to performance, or laptops capable of supporting dual external monitors. My build is listed in detail here: http://www.minus-reality.com/?p=18

[PalmerTech]

I have some experience with this kind of project at my job.

The issue with camera based tracking over any kind of range is that even the slightest changes in direction (Even the small bit caused by heat expansion of, say, a steel tubing frame that the camera is mounted on) will ruin the tracking, requiring a recalibration. We are using 20 camera modules from Raptor Photonics that provide very low noise signals, and even then, one of our largest issues it trying to figure out how to get the system to calibrate itself accurately enough without having to reset for every single run. You will have a very tough time getting accurate hand and head tracking with just a few cameras, and definitely not in an area nearly as large as you propose.

For output, you probably don't need a powerhouse. I would suggest using an ION based netbook (I can suggest a few, if you want to go that route) for the low power consumption and small size, despite packing a decent GPU punch. On top of that, a Matrox Dual Head adapter box would give you the dual outputs you need.

[cybereality]

That looks like an interesting project. Unfortunately, I can't help you much in the ways of technical support, but I wish you the best of luck.

[vvortex3]

PalmerTech, what tracking software are you using? It seems that OptiTrack is handling alot of the calibraiton issues with their software before the feed would be sent to my graphics engine on the client side.

One of the toughest things to get information about, it seems, is the range and number of cameras required if i used powered infrared LEDs. I have seen university experiments where the researchers found the infrared camera on a wiimote to track up to 400 feet.

I definately could use reccomendations on laptops confirmed to be capable of supporting dual external monitors WITHOUT a matrox splitter. The matrox splitter is not capable of powering two distinct displays. From the viewpoint of the graphics engine, it is a single display that is very wide which will not facilitate rendering a distinct view for each eye.

Have any of you had experience dealing with vrealities? I'm may order the dual input z800 soon ($2295, ouch). I have owned both the emagin oled dev kit, and the z800 single input before and I found the single input version to have noticeable flicker in stereoscopic mode and full of driver problems, so I'm pretty sure this will be resolved with dual input and 60hz per eye.

[vvortex3]

Another thing i'm thinking about is to use a different device for roll,pitch,yaw and use the cameras for position, if necessary. Maybe the cameras would be acceptable at range if used in conjunction with a good orientation sensor.

[PalmerTech]

I cannot really talk about our tracking software, it is proprietary. Sorry, not my choice.

You can fix a lot of calibration problems in real time, but it gets extremely hard at the kind of distances you want. Fine motor control poses a lot of problems. Using IR LEDs is a great idea, but it only gets you so far (Much better than nothing, though!).

The Wiimote tracking is quite a bit different, in that you are basically doing the reverse of what you are doing, tracking static LEDs using a moving camera. Considering that you only have one camera in that scenario, and that you do not need more than simple tracking on a 2D plane, AND that you have no limits as to how bright your widely spaced IR light source is... Yeah, 400 feet is doable for sure, but again, not useful for your purposes.

Your idea about using a separate device to aid tracking is a good one. The real problem is, doing fine resolution 3D motion tracking over such a large space requires exponentially more cameras than doing it in a small space. I wish I knew the exact numbers on if you used some IR beacons, I will see if I can get some info on that for you. It is not what we are set up for, but I will do my best (No promises, though!). The higher resolution the cameras are, the fewer you would need.

[PalmerTech]

I missed the part about where you said you will be generating an image for two distinct monitors. Does it have to be that way? IIRC, Torque uses DirectX 9 for rendering. Why not do that, and use IZ3D drivers on top (Are you familiar with them?). That would give you full separation and convergence control, support for pretty much every 3D display of any kind on the market, and a multitude of tweaking options. I love IZ3D to death, their monitor (Now discontinued) had issues, but the driver itself is phenomenal. And, it would support Matrox. Oh, and supposedly, their latest drivers for AMD cards will work with shuttering solutions and HMDs like the Headplay (Seen it?)

I have not had any experience yet, but it is odd you ask! Just two days ago, I came across one, I have not been able to test it yet, but I PM'd another member here to ask if he would be interested in buying it off me if I got it. I am pretty sure they use the eMagin OLED modules, but must be a different driver board, as it supports recent graphics cards. You interested? If so, I will send you a link via PM, don't want to advertise it publicly, lest some lurker snatch it up before us!

My current workhorse at home for personal use is the Virtual Research V8, 640x480 dual input, 60 degree field of view. Works great, probably not what you need, though.

[vvortex3]

I have acquired the HMD and tracking system and have posted the build along with videos on my blog: http://www.minus-reality.com" onclick="window.open(this.href);return false;