P3D Permissible Depth Range (was: "Too much depth")

[abram klooswyk <abram.klooswyk@xxxxxx>]

John W Roberts recommends to use a distinct term to identify
the maximum recommended depth.
George Themelis: "excessive" would be very difficult
to define in a general manner.

The issue of the permissible depth range in stereo pictures
is more than 100 years old.

Chris Jones (P3D 3570, 29 Oct 1999):
>(...) the problem seems to be that the brain processes stereo
>information in more than one way. It's used to using static
>cues (...) for objects more distant than about 2 m (...)
>But when you get up close, you don't naturally fuse the left
>and right images that way. (...) since it gradually becomes
>less seeing the same object displaced laterally against the
>background and more a case of seeing two very different
>views of the same object.
>For close objects your eyes will use small variations in
>direction and focus to enhance the stereo information,
>which doesn't really matter since you can only fixate on a
>small area of vision anyway - you have to move your eyes to
>appreciate most close-up objects in their entirety.

Bruce Springsteen (P3d 3568, 28 Oct 1999):
>Isn't the problem of "too much depth" really more about
>the near limits of stereopsis?
and (P3D 3571, 29 Oct 1999):
>(...) a handgun (...) the end of the barrel is about one
>foot from my face (...) a full moon just over his shoulder.

If I understand Chris Jones right, I'm afraid I disagree to
some degree.

Let's first distinguish between natural vision and viewing
stereo pictures.
There is no boundary at 2 m (in either mode) for any depth
cue. In natural vision, there is no near limit for stereopsis
other than the limit of accommodation and convergence.
Children and myopes can convergence 60 degrees and also
accommodate so close, with perfect stereopsis.

I am moderate myopic (about -5 diopter), I still can converge
at 10 cm, but accommodation fails so close. But I can fuse
cross-eyed stereo pictures at 20 cm (8 inches), which is
converging at 10 cm, and see them perfectly in stereo.
(This is not uncommon for myopes, parallel free viewing at
 20 cm is also easy for them - I recommend myopia :-))

Accommodation is a so weak cue for depth that it can be
discarded in practise. The main function of convergence
in depth perception (apart from assuring fusion) is to set
the scale for stereopsis, but tiny depth differences can
only be seen with stereopsis.
Children and myopes recognize the depth of tiny wrinkles
when holding a hand at 10 cm before the eyes, also when
the eyes are fixed at one point, without eye movements.
(That stereopsis is independent of eye movements is known
since the mid 19th century).
When surface points of an object are seen sharply by both
eyes, stereopsis works, at any distance, unless the distance
is too *large*.

When you see different sides of a gun barrel pointed at
you, rivalry is the result, but if you get the time you
can see depth in its nearby rim.

Now for viewing stereo pictures. In free viewing or with
stereoscopes there seems to be no near limit for
stereopsis either, but you shouldn't expect comfortable
viewing, certainly not with a distant moon.

I the 1930's the Dutchman van Albada (known for a view
finder, but also a stereo expert) has publishes stereoviews
of a landscape ranging from about 50 cm to infinity.
This was a careful shot with gradual depth differences at
all distances. He published it to show that the depth
constraint of 70 minutes of arch, advocated in Germany
(infinity to 10 feet for normal base camera's), was not
justified as a general rule.

The NASA has published pictures of the 1976 Mars Viking
expeditions in the book "The Martian Landscape"
(Washington D.C. 1978, Lib. Congress No. 78-606041),
which contains 19 stereo pictures. The book came with a
Taylor-Merchant "Stereopticon 707" book stereo viewer
(the cardboard version). The Vikings had each two
camera's about one meter apart, and I guess some two
meter above the ground, on Mars they could see the horizon
as well as nearby rocks.

Some of these stereo pictures are probably among the
pictures with the largest deviations ever published.
(nobody in any stereo club would dare to show such
images to other members:-).)
With infinity separations of about 59 to 61 mm, the
near point separation is often less than 30 mm,
down to 25 mm. Viewing at 13.6 cm from the paper, a
deviation of 35 mm is comparable with 8 or 9 mm deviation
in a Stereo Realist pair, which *is* extreme.
Is this viewable? Hardly. But in looking first at
the distant rocks on the Martian soil, and gradually
looking closer, I can fuse the near rocks too, I only
feel some tension in the eye muscles ...

Now for the handgun shots by Bruce, I guess they can
be fused too in stereoscope viewing, but not comfortable,
and not the gun and the moon at the same time.
In projection fusion would be more difficult. I don't
know why, but I suppose because you also see a part
of the room in the periphery of your field of view.

The gimmick in 3D movies is to throw objects at you,
like the stones after the landing of the aliens in
"It cam from outer space", and the ball on an elastic
string in front of the "House of Wax".
In these cases the deviations are even larger, but for
very short times. Still the stereoscopic effect can be
seen.

Pat Whitehouse in her famous dissolve stereo shows often
projected slides which would be difficult to view when
they would be shown longer, but she did the dissolve
so fast in *those* type of slides, that you only could
look at the part of space where your attention was fixed
in the slide *before*, and it was great.

So this long argument only confirms what George Themelis
said: "The bottom line is that you can quantify depth via
the stereoscopic deviations but "excessive" would be very
difficult to define in a general manner.
(I will say more on stretch, distortion, deformation
another time)

Abram Klooswyk