f direct
ion over
le, how-
dequate
1. So, for
an. Cor-
mputations
)hs are
> presuppose
ects (e.g.,
)endicularly
tant. Let
m which the
vely (Fig. 4).
background
itaining the
iseline from
; denoted by
ixis will be
bserver's
the projec-
ely. The
lenoted by
a scale factor
9o that the photo
j. We shall
yhs at the points
und plane.
| space. In
virtual space the position of a point is defined by the place where the ocular axes
cross in regarding it, whatever the intervening optics. We erect a coordinate sys-
tem in virtual space in the same way as in real space. The x-axis is chosen as the
perpendicular bisector of the base line between the eyes and the y-axis passes
from right to left through the ocular centers (Fig. 5). The distance of the focal
plane of the stereoscope from the axes is denoted by x, and half the baseline LR
by b. The ocular projections into the focal plane of the images of P fallat
distances t and t
R L from the ocular axes projected onto the focal plane where
(3) tg = mk rp, ti 2mks
and m is the effective magnification of the stereoscope lens. The point P =
(£, n) of real space is mapped onto the point (x, y) of virtual space for which
tg = (y + b) Xe[X
(4)
ups (y - b) Xe/X .
From the equations (2), (3), (4) we obtain the coordinates of the point (x, y) in
virtual space corresponding to (£, n) in real space, namely,
x = ips = bn
mKBE, 7” R
Introducing the net effective magnification M where
M = mké /*,
we have simply
xat
(5) M B
y=bn
B
From equation (5) it follows that the stereoscopic presentation exactly
reproduces the sensory effect of real space if and only if x = € and y=n. This
means that b=ß and M= 1, a result which is plainly of little practical use. Conse-
quently, we turn our attention to approximate methods.
Let us choose as a specific stimulus in real space, a line segment symmetric
tothe £-axis of length 2a lifted off the background a distance h (Fig. 6). The co-
ordinates of the left endpoint of the segment are then given by & - 56 -h, nea. In
virtual space such an object presents a certain convergence disparity against the
“l=