Virtual Reality/CAVE Software?

WorthlessManatee asks: "I work for a company that is in the beginning stages of constructing our own CAVE and I would like some advice on a few topics. Which hardware vendors should I go through? I know of Fakespace, but are there other vendors I should be looking at? What software solution works best for a CAVE? So far I know of: VR Juggler, DIVERSE, and CAVELib? What are the pros and cons of each of them? Answers to these questions and any other hints or experiences you would care to share would be most appreciated. Thank you very much."

Comments

This is what a CAVE is:

[djcapelis]

This btw... is a small explaination of a CAVE.

but wait there is more...

[drfrog]

fishing in my weblinks....

unreal cave
projection links and type breakdowns
uaeu cave
elumens products are droool
fakespace
vrml 3d projection
not quite cave.... this is 3d projection
this one is further leading.....

io2tech

Re:Translation:

[NanoGator]

"Hi, I'm admittedly worthless, so I thought I'd bump up my karma by submitting an Ask Slashdot question about something totally obscure, since the editors seem to be on an 'accept all' kick today."

Hi, I'm a condescending jerk that wants to bump up his karma by trying to point out the silliness of an Ask Slashdot. I'm incapable of providing thoughtful input, and the idea of ignoring the subject I'm not interested in has escaped me, so I'm hoping that my critical attitude will demonstrate just how elite I am.

Cave Software

[Weyland Yutani]

If you think Performer is not exactly the best API for what you want, look at what a friend of mine has been doing here in Amsterdam ...

http://www.nat.vu.nl/~desmond/VIRPI/

http://www.nat.vu.nl/~desmond/Aura/

http://www.cs.vu.nl/~renambot/vr/html/intro.htm

CAVELibs vs VRJuggler

[Yerase]

CAVELib's is a very simple, glut-like API that lets you take existing opengl apps and make them "CAVE-Compatible" very quickly.. Unfortunately, it's difficult to do multithreading or view-dependent rendering the way it's setup. I prefer VRJuggler. It's Object oriented, and pThreads based. IMHO, it's much more "mature" than CAVELibs, and allows alot more flexibility. Unfortunatly, it's a bit harder to learn. Also, VRJuggler is developed by Carolina Vera-Cruz, the same person that designed the original CAVELibs. In her words, VRJuggler was designed to fix alot of the shortcomings and design mistakes she made in CAVELibs.

My favorites...

[icemax]

I work for the Penn State University visualization group and we use Chromium (which supports cave-like deformed viewpoints and stereoscopic rendering over them, as well as low-cost linux clustering and high-end myrinet and infiniband clustering) and OpenSG (A general purpose toolkit for writing scenegraph-based applications). Good luck! Visit our website for information on how we've used it: PSU Vis. Group

The UAEU-CAVE experience

[aokeil]

What do you mean by "constructing our own CAVE"? Really constructing your own CAVE as we did at the department of Architecture, UAEU? or have a company construct it for you?

In our case I selected to construct it myself using of-the-shelf components. We have a cluster of 3xP4s + windows 2000. more info found at: http://www.engg.uaeu.ac.ae/a.okeil/uaeu-cave/

A nswers to your questions depend on several determining factors such as:
1- What type of applications will you display?
2- Do you need navigation+interaction or only navigation? What type of interaction.
3- Do yoiu plan to integrate other input devices? tracking systems?
4- How do you plan to create your applications (on the long run might cost more than the initial cave investment).
5- Is the cave ment to be a showcase for your company or is it for internal use?.

In our case simplicity of creating the application came before anything else. We selected to start with programs students of architecture already use such as AUTOCAD, 3DS MAX and FormZ. The models are exported to the VRML format. The model is then displayed using a web browser + the cortona plug-in. A small application was developed to synchronize events on all screens. That is all you need to navigate through a virtual world in our cave. I think it can never get simpler. If you need some interaction then you have to start working on the VRML level using VRMLPAD for example.

This approach has saved us and our students the hasle of going into programming in languages none of us know. In 5 minutes you can have your 3D model out of Autocad running in the cave. I hope this reply answers some of your questions. Let me know if you have other questions.

Of course there is...

[cr0sh]

Your graphics hardware isn't the problem - we all know that a very decent rendering engine PC can be had for almost nothing. Let's say dual-head, high-end process, ram out the door, etc - probably do it for $2000-3000 (and that is a very liberal guesstimate - likely the final PC will be a lot cheaper).

The software is cheap/free, too - plenty of options there, from modded UT and Quake offerrings (among others), to Python/PyGame/OpenGL combos.

Where you will be hit is the HMD/output end. If you are willing to go with a lower resolution (640x480 - 800x600), there are plenty of offerrings on Ebay and elsewhere - you have to know whether your app will need immersion (60 degree H x 45 degree V FOV minimum), or if you need resolution (always a tradeoff). If you are looking for immersion (personally, I always prefer to be immersed - its VR, not looking down a tube!), look into a Visette 2 or Visette Pro HMD. These HMDs could be termed "arcade" or "prosumer", or low-end research HMDs - the Visette 2 HMD was used on the Virtuality 2000 series pods, and can be found used. One caveat though, it needs PAL (composite) video inputs. The Pro version takes SVGA.

Try to "try-before-you-buy", if you can (very difficult, I know). For the Visette 2, go to Vegas or a theme park and look for the Virtuality "VR Arcade" or whatever they call it, and try it out (stay away from the older Virtuality 1000 pods, the HMDs are heavy, and low res - but great FOV).

At any rate, you should be able to pick up an HMD for under $1000 if it is used (I have bought both a Visette Pro and a CyberEye CE-200M HMD for under $300 each off ebay). A new HMD like that will run from $1500-2500.00. A final option would be to build the HMD yourself - easily doable, people did it all the time back in the early 90's using handheld casio TVs.

The nice thing about the Visette 2 HMDs (off an actual Virtuality pod), is that they have Polhemus receiver coils built-in - meaning that you can easily interface them with Polhemus 6DOF tracking systems. Same with the Virtuality 2000 pod's joystick.

Tracking is where you will also run into money issues - expect to pay around $2000 for a good tracking system. If you just need head movement, though - and can live without "roll" of the head (only yaw and pitch), look into modding a GyroPoint GyroMouse for the application - simple, and cheap 2DOF head tracking. There are also a lot of simple tracking possibilities that can be homebrewed, as well - but if you want a commercial system, expect to pay some money.

Now, as far as gloves are concerned, you would probably do best to get a commercial 6DOF tracking system (put the sensor on the back of your hand or wrist), then buy some lycra gloves, sew some brass button "snaps" onto the fingertips and palm, and make a "pinch" circuit recognition system (rather than tracking finger bend - unless your app requires it). Gloves will run you an arm-and-a-leg - they are so niche (and still locked up in patent issues from the VPL patent set - it was a very broad and large patent suite that VPL had). You can spend a lot here. Roll your own, if you can (and if you aren't selling a product - most any glove that you homebrew will stomp on a number of patents). Another possibility is to hack up/apart a powerglove for the flex sensors, then sew them into/on a lighter lycra glove...

If you aren't going to want to "look around" your world, then simple shutter glasses, like they sell for game systems, can be had for cheap - and you don't even need dual monitors for it. If you want something more professional, a CrystalEyes system would be best, but be prepared to spend some cash.

Even if you go completely professional (and stick with a regular PC and free/open source VR software solutions), you can easily do this for under $10,000. If you really take your time, you can easily do it for much less (remember, homebrew VR on PCs used 386's back in the early 90's - cost for a VR setup really hasn't changed much if you homebrew all or most of it - its only when you bri