Re^2: 3D-LCD Software??

[P3D Joerg Meyer Inf.] <jmeyer@xxxxxxxxxxxxxx>]

On Tue, 7 Jan 1997 Larry Berlin <lberlin@xxxxxxxxx> wrote:

> This method definitely works, but leaves gaps of information if you use
> alternate lines of each image. Remember the paired information is ideally
> exactly horizontal. To make it truly viewable with half the information
> missing, you have to NOT invert the line image. Use the exact same lines
> from each pair, and before combining them, shift one of them either one
> pixel up or down. This provides the most accurate and corresponding
> information in the interlaced results.

That's definitely true for still images. Since one image is transmitted
to the left eye only, and the other to the right eye only, the small
vertical shift between corresponding lines won't be detectable.
The amount of information is the same for even/odd lines and for even/even
lines, the latter case also providing half the resolution of a full
frame only. For still images this method also might reduce flicker when
viewed w/o LCD shutter glasses, because similar objects are displayed
twice (in two subsequent lines, two subsequent half frames).
   During animation sequences I noticed that motion is smoother if you
use alternative (even/odd) lines, because if an object moves in vertical
direction within one half frame, the motion blur, which makes a motion
picture more realistic than computer graphics, is reproduced in a better
way. If you don't have motion blur, the image seems to jump between two
time steps. It's a bit tricky to explain, but that at least is my             
personal experience.

> One other step that's important is getting the parallax right. Using the
> process described here it is imperative to first adjust the images size and
> position relative to each other so that a direct combination gives you what
> you want. With this method, I find it easy to use a conversion to anaglyph
> before making the interlaced versions, just to get it right. Make the
> adjustments and crop to your final images before applying the lines.

Good point! I also tried to overlay the two images first in order to
adjust horizontal and vertical alignment. Once you have your images on
the computer, it's an easy to job to convert them to red/blue, red/green,
or even pseudo-color anaglyphs!  :-)

Joerg


    _V_     | Joerg Meyer ----- University of Kaiserslautern, Germany
   /   \    | Department of Computer Science, Computer Graphics Group
   |O O|    | mailto:jmeyer@xxxxxxxxxxxxxx --------------------------
ooO--U--Ooo | http://davinci.informatik.uni-kl.de/~jmeyer -----------


------------------------------