hultquis@unc.UUCP (Jeffrey P. Hultquist) (09/01/85)
A book, published in '66, mentions a "volumetric display" and describes such an animal in this way. Suppose you took a fishtank and filled it with some phospor with a non-linear excitation response. You could then pass two lasers through this medium at half the required triggering intensity, and thus excite any point in this volume by intersecting the beams. The book continues to state that no one had yet done this. Questions: In the past twenty years, has anyone tried to build such a device? If so; who, when, where, how, and did they publish their findings? Many thanks. PS: Please mail and I will post a summary to the net. --- decvax!mcnc!unc!hultquis
gnome@olivee.UUCP (GNOME) (09/20/85)
> A book, published in '66, mentions a "volumetric display" and > describes such an animal in this way. Suppose you took a fishtank > and filled it with some phospor with a non-linear excitation response. > You could then pass two lasers through this medium at half the > required triggering intensity, and thus excite any point in this > volume by intersecting the beams. The book continues to state that > no one had yet done this. > Well, it isn't really the same but I had some success with a spinning screen (wire grid) and an x/y deflected and modulated laser. Bad part is the velocity needed to get a complex figure drawn. Looked neat though. Gary
jr@bbncc5.UUCP (John Robinson) (09/24/85)
BBN desgined a display system that used a vibrating mirror to cause an oscilloscope face to trace out a volume in the virtual image behind the mirror (got that? hope so). By timing the images on the scope, you could place them at a particular depth. The mirror was excited by a regular ole' loudspeaker at 30 hz (maybe it was 60); it was counterweighted so that its vibration was parabolic rather than spherical. Eventually, Genisco bought the technology and reengineered the package to sell as their own. I don't believe they have had much luck selling it. I thought the images it produced were quite striking. I could chase down references if anyone's interested. /jr
fournier@Navajo.ARPA (09/26/85)
> BBN desgined a display system that used a vibrating mirror ...... ....... Eventually, Genisco bought the technology and > reengineered the package to sell as their own. I don't believe they have had > much luck selling it. I thought the images it produced were quite striking. *** REPLACE THIS LINE WITH YOUR 8-CONNECTED STRING OF PIXELS *** Genisco marketed it under the name "Spacegraph". They had one at Siggraph about three years ago. It was in the $100K range if memory serves. Henry Fuchs' group at North Carolina/Chapel Hill worked on a similar design. The beast appears in one of their demo videotapes.
carl@bdaemon.UUCP (carl) (09/28/85)
Demonstration models of volumetric displays using vibrating mirrors,
rotating screens or similar concepts have been around for a long, long
time. One of the first I saw was in 1962 at the ITT Labs in Nutley, N.J.
Static 3D displays were common at the turn of the century and the subject
of a fair amount of research by perceptual psychologists at the time.
The reason why we have seen many demonstration devices and no production
displays is very simple - the huge bandwidth required.
Consider the case where ordinary television resolution is acceptable:
1. A 2D picture, just like your vanilla TV, requires about 3 MHz
for about 300 line resolution.
2. For a 3D picture with about 300 planes in the Z axis the
bandwidth would have to be 3*10^6 * 300 or 900 MHz, a non-trivial
accomplishment for a display with limited useful applications
(Would you really like to have an air-controller control *your*
flight with a standard TV quality display?).
Obviously, resolution is a function of the third power of the bandwidth, so
I am not going to hold my breath waiting for practical 3D displays to become
reality.
Yours for an appreciation of what simple arithmetic can do for old,
time-worn ideas.
Carl Brandauer
daemon associates, Inc.
1760 Sunset Boulevard
Boulder, CO 80302
303-442-1731
{allegra|amd|attunix|cbosgd|ucbvax|ut-sally}!nbires!bdaemon!carlgwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (09/29/85)
> Obviously, resolution is a function of the third power of the bandwidth ...
That's not obvious at all. Once you get near the eye's resolving
ability, further improvement buys nothing. Not only that, but very
effective stereoscopy has been done with two 512-pixel square images;
pixel-to-pixel coherence makes the depth resolution argument less
significant.
Interesting point: It was reported in the 1950's (when stereo
photography was popular) that something like 30% of adults have
defective binocular depth cueing. Perhaps this helps explain why
stereoscopy keeps failing to catch on with the general public
(eyestrain brought on by defective techniques didn't help, either).jqj@cornell.UUCP (J Q Johnson) (09/30/85)
Although it might seem that a 3d display requires bandwidth o(n^3) for resolution n, in fact a great deal of data compression is possible. We don't actually see in 3d, but rather in 2-1/2 d or so (who cares about the inside of a solid?). Consider a representation that encodes only range data at each point on a 2d display -- it requires bandwidth o(n^2*log(n)). Seems to me that most 3d images should be encodable with a small constant multiple of the range data (e.g. as up to k visible z values at each x,y position; by continuity, the k is not a function of the resolution). What representation actually achieves o(n^2*log(n)) for a 3d display? Well, how about a hologram?
carl@bdaemon.UUCP (carl) (10/02/85)
> > Obviously, resolution is a function of the third power of the bandwidth ... > > That's not obvious at all. Once you get near the eye's resolving > ability, further improvement buys nothing. Not only that, but very > effective stereoscopy has been done with two 512-pixel square images; > pixel-to-pixel coherence makes the depth resolution argument less > significant. Are rare instance when Doug misses the point. A stereoscopic display consisting of two 512-pixel square images is fine if all you want is a nice *static* picture of Granny in front of Old Faithful. However, a dynamic sequence showing Granny walking to the right spot, Old Faithful gurgling and spitting before finally starting to spout at full speed etc., etc. will require about 30 * 2 * 512 * 512 = 1.57 * 10 ^ 7 pixels per second if each pixel is either on or off and if we want to avoid excessive flicker. If we assume that 8 bits are needed for a decent gray scale, 1.26 * 10 ^ 8 bits per second are necessary, a fairly hefty bandwidth for a pretty crummy picture. Carl Brandauer:{allegra|amd|attunix|cbosgd|ucbvax|ut-sally}!nbires!bdaemon!carl
gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (10/02/85)
> > > Obviously, resolution is a function of the third power of the bandwidth ... > > > > That's not obvious at all. Once you get near the eye's resolving > > ability, further improvement buys nothing. Not only that, but very > > effective stereoscopy has been done with two 512-pixel square images; > > pixel-to-pixel coherence makes the depth resolution argument less > > significant. > > Are rare instance when Doug misses the point. A stereoscopic display > consisting of two 512-pixel square images is fine if all you want is a nice > *static* picture of Granny in front of Old Faithful. However, a dynamic > sequence showing Granny walking to the right spot, Old Faithful gurgling > and spitting before finally starting to spout at full speed etc., etc. will > require about 30 * 2 * 512 * 512 = 1.57 * 10 ^ 7 pixels per second if each > pixel is either on or off and if we want to avoid excessive flicker. If we > assume that 8 bits are needed for a decent gray scale, 1.26 * 10 ^ 8 bits per > second are necessary, a fairly hefty bandwidth for a pretty crummy picture. Well, thanks for calling it "rare". I believe the original "third power" argument was that DEPTH needed the same resolution as HEIGHT and WIDTH. TIME had nothing to do with it.. Maybe I really did miss it.. The bandwidth for a movie is significant, all right, but 512x512 is about the same as Super-8, so stereoscopic movies are not much of a problem. 512x512 is also about what a good TV signal might attain so all these are in the same general ballpark. Of course one would have to get better effective bandwidth than today's commercial TV broadcasts. The Japanese are working on it, but I don't think they're planning on stereoscopy (just better pictures). It is interesting to note (if one hasn't thought about it before) that considerably less bandwidth is needed for typical moving pictures than simple calculations show. Back in the days of the PicturePhone, the BSTJ was full of articles on delta modulation and related techniques for exploiting this feature of typical source material. I don't know whether there are any good methods for doing these things at video scan rates..
jqj@cornell.UUCP (J Q Johnson) (10/03/85)
In article <309@bdaemon.UUCP> carl@bdaemon.UUCP (carl) writes: >> > Obviously, resolution is a function of the third power of the bandwidth ... >> >> That's not obvious at all. Once you get near the eye's resolving >> ability, further improvement buys nothing. >> . . . >A rare instance when Doug misses the point. A stereoscopic display >consisting of two 512-pixel square images is fine if all you want is a nice >*static* picture of Granny in front of Old Faithful. However, a dynamic >sequence showing Granny walking to the right spot, Old Faithful gurgling >and spitting before finally starting to spout at full speed etc., etc. will >require about 30 * 2 * 512 * 512 = 1.57 * 10 ^ 7 pixels per second if each >pixel is either on or off and if we want to avoid excessive flicker. Ah, but the point is that quite a bit of depth info is available for only a factor of 2 increase over 2d. Granted, digital animation is expensive! But even there temporal coherence can be used to reduce the needed bandwidth; the trees in the background aren't changing. The real problem with stereo is that it doesn't give much 3d data. It's static perspective, and works only if you have exactly 2 eyes (what about the poor Venerians?). I want to be able to move my head and look behind an object!