home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.graphics.animation
- Path: sparky!uunet!statsci!almond
- From: almond@statsci.com (Russell G. Almond)
- Subject: Animation and 3-D (Re: Anyone using stereo 3-D?)
- In-Reply-To: schuette@quip.eecs.umich.edu's message of 15 Sep 92 03:02:59 GMT
- Message-ID: <ALMOND.92Sep15180129@bass.statsci.com>
- Sender: usenet@statsci.com (Usenet News Account)
- Organization: Statistical Sciences, Inc., Seattle, WA USA
- References: <1992Sep15.030259.5923@zip.eecs.umich.edu>
- Date: Wed, 16 Sep 1992 02:01:29 GMT
- Lines: 79
-
- In article <1992Sep15.030259.5923@zip.eecs.umich.edu> schuette@quip.eecs.umich.edu (R. Wade Schuette) writes:
-
- Wade Schuette Writes:
-
- Just curious about who in the world is doing what with low-end
- stereoscopic 3-D on amiga's or PC's or Mac's or Suns or whatever's
- low-end for you. (like, using Sega glasses, not StereoGraphics
- CrystalEyes or a pair of microscopic Televisions attached to ... ) .
-
- You don't need stereo pairs if the animation is good enough, the
- motion is sufficient to give you the illusion of depth. This has been
- known in the Statistics world for some time, and a lot of the work on
- rotating point clouds is to allow us to look at 3-D data on 2-D
- screens. (In answer to the original post, there is currently a lot of
- Statistics software out there which does this. Xgobi, I believe is in
- the public domain. MacSpin is a classic example, and StatSci's S-PLUS
- has those features.)
-
- There is an interesting story about rotation and stereo pairs. An
- early version of PRIM-H supported both rotation and stereo, on a
- relatively high end system. We did a classic experiment where we took
- a 3-D normal point cloud (equal variances, no covariance) and removed
- a "ball" consisting of about 20% of the points from the center of the
- distribution. Any 2-D projection would look approximately normal, and
- neither rotation nor stereo alone was sufficient to show the problem,
- however, both rotation and stereo pairs would.
-
- We later ported the program to an Apollo workstation without the
- stereo stuff. We decided to build a movie based on our work called
- "Visual Aspects of 3-D Data Graphics: A Movie" (Almond, Huber,
- Ramos, Roseman and Thoma: Peter Huber produced it; I directed and
- photographed). For this purpose we produced a special animation
- version of PRIM-H which would allow us to: (1) input commands we
- normally imputted through mouse and key interaction using a simple
- script language (I needed to write a macro preprocessor for the script
- language), (2) Would "click" a camera shutter through a serial line
- which was hooked up to a "blue box" animation/time-lapse controller
- which drove the camera through a sound sync motor. An interesting
- feature of this settup was that we could use essentially time lapse
- photography. I used 8 second exposures to get good color saturation
- (with a high T-stop) and the computer would click the shutter after
- the image was drawn. Thus, even though it took half a night to shoot
- 100' of film, we could show it back at 24 frames per second.
-
- While the narration explained the "experiment" with the normal
- distribution, the movie displayed it; telling people how they really
- couldn't see the effect. At the higher speed of 24 frames per second
- and with correspondingly smaller increments for the rotation (we were
- getting about 6/frames per second off of the 1983 hardware) the
- audience could all see it quite clearly!
-
- BTW, the trickiest portion of the whole enterprise was lining up the
- animation with the verbal clues in the narration. In order to do
- this, I first ran some test talk throughs of the narration, to get
- timing marks for the animation. I then twiddled the animation script
- so that it would match the queues. Finally, I had to have the
- narrator match the verbal queues with the video test print when we
- recorded the sound track. A real trick. After the first movie we
- learned to avoid phrases like "We color the affected points yellow,
- NOW" for phrases like "We color the affected points yellow;" that
- gave us more margin for error.
-
- Does anybody have any experience trying similar tricks in some kind of
- hypermedia environment? I've been thinking that a perfect on-line
- tutorial would have the computer running through some script
- demonstrating the functionality while a recorded narration described
- what the program was doing. Can this be done? Has anybody tried it?
-
-
- Russell Almond
- Statistical Sciences, Inc. U. Washington
- 1700 Westlake Ave., N Suite 500 Statistics, GN-22
- Seattle, WA 98109 Seattle, WA 98195
- (206) 283-8802
- almond@statsci.com almond@stat.washington.edu
-
-
-
-
-