Ain't talking?

[T3D Michael Gordon <mgordon@xxxxxxxxx>]

>I interpreted this to mean that he tilted the flattie in the 
>scanner: one direction to get a left view and the other direction 
>to get a right view.  But I may be way off base here.  Only 
>Michael can tell and he ain't talkin'.  8-)

Well, that's cuz I've been trying to get Stereoscopy out not so late!
Setting algebra equations takes a very long time and plus I think there's an
error which wouldn't matter so much but that thick nest of equations makes a
lovely screen made white and superimposed over the cover image; suggesting
that while we study these things scientifically, in the end, it is the image
that still matters.  So long as the equations are small, 12 point Century
Schoolbook, maybe not everyone is going to notice, but big on the cover,
anyone with high school algebra must at least suspect there's something amiss.  

Naw, what my roommate and lifelong professional photographer suggested to
do, was scan a flattie, make two copies of it (identical in every way) then
rotate the image in space, as for example about 2 degrees on the Y axis.
What would happen is, as has been discussed, some keystoning.  Several
bitmap editors have the ability to seem to rotate an image on the Y axis,
but that is no different than a picture pasted to a plane, and then the
plane itself rotated.  

A small amount of keystone distortion is present naturally all the time
especially for near objects -- your left eye will be closer to the left side
of an object, hence, it will be larger.  Since is the expected and natural,
you make nothing of it; and you also don't see the entire object all at
once.  You build up a "map" of it so you SEEM to see it all at once.  

However, viewing a flattie that has its clone rotated on the Y axis, you
have not only the usual and expected keystone effect on the left and right
edges caused by simple proximity to the view, but you have an extra
application of keystone.  Since the amount is small, it is within your
brain's ability to accomodate it; and it will interpret the effect.  Larger
amounts of keystone are caused by closer proximity.  Hence, the part of the
brain that adds to the "mix" of distance sensing, says, the edges of the
resulting view are closer to your eyes than the middle, hence, curved.  The
effect can actually be seen on a computer if you use a checkerboard; but it
is extremely subtle since the keystone effect does not, for me, contribute
much to my sense of distancing.  For others that do not use keystoning AT
ALL to judge nearness of objects, the image would seem to be a perfectly
ordinary stereo view of a flat photo (ie, rather pointless!).  

As for projection, the keystone distortion would make registration
impossible but since perfect registration is not only impossible, but not
even wanted, it is not much of a problem except for the folks on the front
row where the vertical size misalignments make accomodation impossible.

I have taken pencil sketches, colorized them, and made stereo images from
them; very extremely tedious work it is!  I do it by cloning the object,
then carefully shifting elements of it while viewing it stereo.  Objects
that are oblique, I mask and skew the object so that it obtains a
proportional change in shift, more toward the front.  It works reasonably
well, not quite spectacularly.  

Horst Hoyer has done some excellent conversions; he was going to publish an
article featuring the drawings in Alice In Wonderland, but has chosen to
postpone this feature for now.  

Sincerely,
Michael Gordon


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