Alex, I would take a look at something like the ROAM algorithm to dynamically tessellate the landscape based on viewer position and performance. I don't know of anyone who has done this within Java 3D yet, so a good deal of prototyping may be in order. A friend of mine did use it using GL4Java however and claimed good performance.
More details here: www.vterrain.org Sincerely, Daniel Selman -----Original Message----- From: Discussion list for Java 3D API [mailto:[EMAIL PROTECTED]]On Behalf Of Alex Terrazas Sent: Tuesday, January 08, 2002 4:50 AM To: [EMAIL PROTECTED] Subject: Developing the Imposter Hey- I never heard back from anyone regarding collaborating on some of these imposter methods. Anyway, I am prepared to go on and I hope others will join me or at least correct me when I am wrong. There is some possibility of making small academic money for helping with this. A little background. I am studying spatial memory and scanning peoples brains while they navigate in virtual environments. The subject will sit in an MRI machine and navigate the MSU environment. We will also be looking at transfer of memory across reality and VR. Finally, the subject will be able to drive a video equiped robot around campus while lying in the scanner. The subject will wear special goggles but we also want to have subjects navigate over the internet on flat screens, in our Immersadesk using shutter glasses, etc... The Java 3D view model is really working nicely in this regard-thanks to a lot of help from Mark Hood at Sun--see his replies to my questions on the archives if you are interested. Project Goals: Making an accurate virtual MSU allows me to perform accurate comparisons within/between subjects. I may want to distort the virtual envioronment later but I want to start from the most realistic that is practical. I am having students model different buildings using 3DStudio MAX. We started with Beaumont Tower. In the HMD, this thing is really nice. It's one of the only truly immersive experiences of my life. We are teaching a class on VR and I am going to have this project as a part of that. So, where to go next. I am making some progress on loading the terrain as a whole. We are beginning in the park area around Beaumont and working our way outward. But this is where it gets tough. How do I load the next terrain element--and how do I break up my terrain into elements. Some sort of data structure is necessary, right? Also, is it best to go through the LOD mechanim to load up the terrain and neighborhood? How should I organize my terrain model? Sure, someone will say just dynamically load and cache the geometry--but I am not sure how to implement that. Maybe it is best to deform a surface dyanimically according to the position of the viewer. For example, the terrain model for MSU is in a .dwg format which I believe is a vector format. The model gives 1 ft elevations for the entire 7000 acres campus (MSU is huge). Could I write a custom Java NIO to read the format, skip the appropriate elements and update the terrain model? Anybody have suggestions? Alex =========================================================================== To unsubscribe, send email to [EMAIL PROTECTED] and include in the body of the message "signoff JAVA3D-INTEREST". For general help, send email to [EMAIL PROTECTED] and include in the body of the message "help". =========================================================================== To unsubscribe, send email to [EMAIL PROTECTED] and include in the body of the message "signoff JAVA3D-INTEREST". For general help, send email to [EMAIL PROTECTED] and include in the body of the message "help".
