Received: from watserv1.uwaterloo.ca (watserv1.waterloo.edu) by karazm.math.UH.EDU with SMTP id AA28998 (5.65c/IDA-1.4.4 for ); Wed, 23 Oct 1991 18:48:28 -0500 Received: by watserv1.uwaterloo.ca id ; Wed, 23 Oct 91 19:44:19 -0400 Date: Wed, 23 Oct 91 19:44:19 -0400 From: Dave Stampe-Psy+Eng Message-Id: <9110232344.AA07249@watserv1.uwaterloo.ca> To: glove-list@karazm.math.uh.edu A reply to "Gary McTaggart" >-When objects are moving in the "world", would it be appropriate to only >recalculate the screen coordinates from the bottom up (world coordinates to >screen coordinates) only every X number of frames, depending on the nature >of the movement. > Not a bad idea... of course, you have to keep track of the object's position _somewhere_ in full resolution (in case the user "moves" closer). In effect, you don't update the world database, or flag the object as moved right away. The problem is, (especially if you're using eyephones) that the viewpoint is just as likely to move as the objects. This requires almost all of the objects on the screen to move significantly. It would work fine if the viewpoint doesn't change, though. >-If the movement of the object in the view coordinate system is: > 1. a small amount of rotation in on the X or Y axis > 2. any amount of rotation along the Z axis > 3. any small movement ("small" determined by experimentation) > > , could we not treat the object as a 2D image and use the screen >coordinates along with the previous Z view coordinate to approximate where >the object should be? > This is practical in some cases: it depends whether you do partial or full screen updates. Again, I'm not sure if it's always appropriate for VR, because the viewpoint often changes. If the object rotates on _any_ axis, the rendered image will change. Because of aliasing effects, even subpixel motions will affect the image. Whether the preservation of this effect is important, though, I don't know. The 2D image idea isn't too bad for distant objects, though, as it takes BIG changes in the viewpoint or the objects themselves to change them. So, if you have an "outdoor" VR world, it's worth it. "Indoor" worlds with all objects close might not be able to use it, though. This whole idea is a form of partial updating, which can be useful if there is only a few moving objects or the viewpoint is stationary. I haven't started serious work on partial updating yet, because it falls under the area of the graphics front-end that generates drawing lists. I've been concentrating on the renderer because it sets the standards for the rest of the software. It seems that a good VR renderer should also be able to redraw small areas of the screen, which is one form of what you were suggesting. I'm not sure whether keeping individual objects images is a good idea, because small objects can be drawn fairly fast, doing a "cutout" copy is quite expensive on a VGA video system, and close objects that look big tend to move and change a lot. Still, some good ideas to consider, some of which WILL affect the renderer. >I have not actually implemented the idea to test its effectiveness, and >being that I'm new at this sort of thing, this could be a common practice. Yes-- on lowend systems, and in video games. We have to see if it can be fit into a VR system. I'm pretty new to 3D rendering myself, although I have a lot of low-level graphics experience. I'm not too thrilled about programming the world database handler myself, as it's not really my field. And thanks for the input. -------------------------------------------------------------------------- | My life is Hardware, | | | my destiny is Software, | Dave Stampe | | my CPU is Wetware... | | | Anybody got a SDB I can borrow? | dstamp@watserv1.uwaterloo.ca | __________________________________________________________________________