Subject: Viewing StereoDate: 2007-01-20 14:23:19From: Oleg Vorobyoff
Sam wrote:
discussion would shed some light on the physiology and psychology of
viewing stereo.
moment and think about what just one eye sees. Suppose you were in the
desert and constructed a large wall with a wall with a six foot square
window cut into it. Now you stand six feet behind the window and look with
one eye at the landscape beyond. Place a camera at the exact position of
your eye and photograph the window. Enlarge the photograph to six feet
square and insert the print in the window frame. If you could illuminate
the print brightly enough, your one eye would see almost exactly the same
landscape were you to remove the print. Now make similar print from the
position of the other eye and insert it in the window frame. I submit that
if you could force each eye to see in the window frame only the print it is
supposed to see, your mind will perceive a scene that is spatially
identical to the landscape behind the prints.
secondary effect. The prints in the window frame will not be exactly
square. Without actually doing the math, I imagine that there might be a
quarter inch difference in the heights of the prints at the left and right
ends of the window. If you want, you could always correct that bit of
keystoning in Photoshop, similar to the way you would correct a phantogram
to appear to lie flat on a table. But at this distance it is hardly worth
the effort.
whom I have had to wrest away my viewer were just being polite to me.
Actually, I only cited the example to illustrate how easy it is for most
people to decouple their focusing reflex from their eyeball convergence.
There must be some people for whom this feat is difficult, but I suspect
it is a small minority.
were vertically or rotationally misaligned, or had the chips mounted too
far apart, or had focal length mismatch. I do believe, however, that the
effect of viewer lenses is underappreciated. One of the problems is that
as the eyes rotate to converge on a homolog they also shift a bit off the
centers of the lenses. The closer the subject matter the more they have to
shift and the more degraded the view through lenses becomes. But the eyes
are not stupid. They will not linger on subject matter that is painful to
view. The trick is to is to keep such subject matter from the away from
the center of the view. A mirror based viewer should yield a more
comfortable view than a lens based viewer, at least with respect to
convergence. But with mirrors it is difficult to maintain alignment of the
left and right images.
projection, I remember the IMAX space station movie. It opened with a
simulation of an astronaut stepping out of the station from his own point
of view. You could see his hands grasping the ladder right where your
hands would be. I extended my hands to meet the image of the astronauts
hands. The image of the astronaut's hands was at exactly arm's length. I
suspect that most of the audience looked past the hands - the view of the
earth beyond was so compelling - but that they could have converged on the
hands had they thought of it. For me it was easy. But I do recognize that
what is easy for me might be difficult for others. The trick, as with any
art, is to make sure that the reward is worth the difficulty. Ideally, a
work has several levels of difficulty, so that it appeals to audiences of
various levels of sophistication.
Oleg Vorobyoff
>Oleg, now you've really confused me. What exactly toI have to apologize. I should not have mentioned names. I was hoping this
>you agree to with Chuck and disagree with me?
discussion would shed some light on the physiology and psychology of
viewing stereo.
>So does that mean you feel the camera DOES see theExactly. Let me restate my thought experiment. Forget about stereo for a
>world as the eyes see it?
moment and think about what just one eye sees. Suppose you were in the
desert and constructed a large wall with a wall with a six foot square
window cut into it. Now you stand six feet behind the window and look with
one eye at the landscape beyond. Place a camera at the exact position of
your eye and photograph the window. Enlarge the photograph to six feet
square and insert the print in the window frame. If you could illuminate
the print brightly enough, your one eye would see almost exactly the same
landscape were you to remove the print. Now make similar print from the
position of the other eye and insert it in the window frame. I submit that
if you could force each eye to see in the window frame only the print it is
supposed to see, your mind will perceive a scene that is spatially
identical to the landscape behind the prints.
>Then how do you explain the keystone distortion issue,Yes, keystoning does enter into my thought experiment, but only as a
>which does not exist through your eyes, but is as plain
>as day in a photographic image?
secondary effect. The prints in the window frame will not be exactly
square. Without actually doing the math, I imagine that there might be a
quarter inch difference in the heights of the prints at the left and right
ends of the window. If you want, you could always correct that bit of
keystoning in Photoshop, similar to the way you would correct a phantogram
to appear to lie flat on a table. But at this distance it is hardly worth
the effort.
>> How many times have you shown a stereo view to a completeI can't believe all those people, both novices and jaded stereo buffs, from
>>novice only to have him or her exclaim how real it looks?
>How many "novices" have said an ugly baby was actually ugly?
whom I have had to wrest away my viewer were just being polite to me.
Actually, I only cited the example to illustrate how easy it is for most
people to decouple their focusing reflex from their eyeball convergence.
There must be some people for whom this feat is difficult, but I suspect
it is a small minority.
>Are you saying the ONLY reason the large majorityNo. I have not seen Chuck's image. Images I have found painful to view
>of us cannot view the stereo image Chuck is
>suggesting is because of viewer lenses?
were vertically or rotationally misaligned, or had the chips mounted too
far apart, or had focal length mismatch. I do believe, however, that the
effect of viewer lenses is underappreciated. One of the problems is that
as the eyes rotate to converge on a homolog they also shift a bit off the
centers of the lenses. The closer the subject matter the more they have to
shift and the more degraded the view through lenses becomes. But the eyes
are not stupid. They will not linger on subject matter that is painful to
view. The trick is to is to keep such subject matter from the away from
the center of the view. A mirror based viewer should yield a more
comfortable view than a lens based viewer, at least with respect to
convergence. But with mirrors it is difficult to maintain alignment of the
left and right images.
>Are you saying that if that exact stereoscopic imagePrecisely, as long as it is properly aligned and illuminated.
>was enlarged to a pair of 4 foot square prints and
> viewed through a mirror stereoscope, we would see
>it as perfect as in real life?
>Why then does it also not also work in a projectedIt does. I have seen giant anaglyphs that look beautiful. As for
>stereo pair or an anaglyph?
projection, I remember the IMAX space station movie. It opened with a
simulation of an astronaut stepping out of the station from his own point
of view. You could see his hands grasping the ladder right where your
hands would be. I extended my hands to meet the image of the astronauts
hands. The image of the astronaut's hands was at exactly arm's length. I
suspect that most of the audience looked past the hands - the view of the
earth beyond was so compelling - but that they could have converged on the
hands had they thought of it. For me it was easy. But I do recognize that
what is easy for me might be difficult for others. The trick, as with any
art, is to make sure that the reward is worth the difficulty. Ideally, a
work has several levels of difficulty, so that it appeals to audiences of
various levels of sophistication.
Oleg Vorobyoff