li 
bo 
i 
M 
1 
13 
ac 
ac 
pe 
sh 
pa 
ing 
ra 
op 
tal 
is 
wi 
ple 
cai 
aci 
fui 
dis 
siz 
sin 
cal 
Inc 
of 
be 
str 
ap! 
lig] 
we 
vit 
on 
tyE 
the 
tha 
cer 
litt 
tiol 
bas 
det 
eve 
jec 
not 
refi 
ong 
ica 
of 
anc 
equ 
enc 
wh 
ord 
me: 
pra. 
it will lead 
desirable feature and also it will lead to simpli- 
10 
the light available for scanning and, indirectly, 
plotting speed. Projection plotters require a 
small aperture in order to achieve a useful depth 
of field. Non-projection plotters are free from 
this limitation. 
A less fundamental, but very practical, 
limit to plotting speed results from inertia of 
the carriage and other parts that move during 
plotting. Inertia increases the size and expense 
of servo-mechanisms, and also governs the 
magnitude of the acceleration forces to which 
the machine is subjected during rapid plotting, 
and which cause vibration and loss of adjust- 
ment. Model size is the most serious factor af- 
fecting inertia, since it controls the size and 
weight of the carriage assembly, and also the 
distance through which it moves. Inertia varies 
between the third and the fifth power, of the 
linear dimensions of the model, depending on the 
type of instrument, and is therefore greatly re- 
duced by plotting at reduced scale. 
The Helava Analytical Plotter * should 
satisfy the basic optical and inertial require- 
ments for a fast automatic machine. The optical 
system is simple, and facilitates the use of 
large aperture. Plotting is at photo scale, and 
the orientation adjustments are made on the 
computer, where they are not subject to the dis- 
turbing effects of vibration and stress. The speed 
potential of such a combination is difficult to 
assess, but a contour or profile tracing speed 
about one hundred times that of a human oper- 
ator can probably be achieved. 
Static Accuracy 
  
The static accuracy of an automatic stereo 
system can be no better than the precision of the 
plotting instrument to which itis attached. Auto- 
matic stereo may introduce additional errors 
however, and their magnitude will depend upon 
the precision of the parallax sensing and clear- 
ing operations. 
The precision of parallax sensing'is re- 
lated to the effective resolution of the plotting 
instrument, including the size of the scanning 
spot. A spot diameter should be chosen that 
will not degrade the resolution of the instrument 
appreciably, bearing in mind that the amount of 
scanning light available is also related to spot 
size. 
Experience indicates that Stereomat can 
clear parallax to about a fifth of the system reso- 
lution. For Stereomat on a projection plotter, 
parallax is regularly cleared to better than 0.01 
millimetres at photo scale. 
Dynamic Accuracy 
Errors arising out of servo limitations do 
not appear to be serious in Stereomat on pro- 
AUTOMATIC STEREO 
jection plotters. Delays in the smoothing net- 
works necessary to suppress quantum-noise 
irregularities are the greatest source of dynamic 
error. 
Network characteristics may be chosen in 
relation to the type of performance required. A 
short time-constant favours small dynamic 
error, at the expense of increased “jiggle” and 
static error. Dynamic and static accuracy can 
both be improved only by utilizing a stereo in- 
strument with greater optical efficiency. 
Contour accuracy is difficult to assess. 
Stereomat on a projection plotter traces contours 
at a speed of about four times that of a stereo 
operator, with more than 90 percent of the con- 
tour lying within 0.5 millimetres of the correct 
height at model scale. 
Profile accuracy has not been precisely 
determined at the present time, but it appears 
to be equivalent to contour accuracy in terms of 
height error. Average profiling speed is 1 or 2 
inches per second at model scale. 
Speed and Accuracy of Orientation 
About 5 seconds are required to clear the 
Y parallax at each parallaxing station, and ap- 
proximately 2 minutes are sufficient to clear all 
five stations for the first cycle. Relative orient- 
ation of an average model is complete in about 
5 or 6 minutes, and requires four or five cycles. 
Initial tests indicate that Stereomat clears Y 
parallax more precisely than an operator by a 
factor of about two3 
Fallibility 
The following conditions affect the ability 
of the correlation circuits to sense parallax and 
develop useful data signals: 
1. Model slope. 
2. Heigbt ambiguity. 
3. Density of the transparencies. 
Model Slope 
In areas of severe slope, a large part of 
the horizontal scanning pattern may be so far 
from the model surface that the resulting X 
parallax is greater than the correlation circuits 
can handle. Experience indicates that a model 
slope in excess of 20 or 30 degrees reduces the 
performance of Stereomat considerably. A pro 
posed solution to this problem requires a separate 
C.R.T. for each photograph. By using scanning 
patterns that are not quite identical the resulting 
pattern plane can be made to assume an angle 
in space and be tangent to the model surface at 
all times. The ability of the pattern to resolve 
=> — yere ver rvererrrvorrvorttp77€7] 1VET 11 CUU Y 665 UIUL TU 
to the unification which is a most 
adequate record of the discussion is available 
for publication. 
small 
by thi 
to cor 
model. 
under 
Heigbt 
, 
erratic 
one S! 
trouble 
the ar 
flying 
commo 
when t 
is visi 
presen 
which 
variou. 
area ol 
I 
this c: 
troduc« 
betwee 
of the 
signal: 
7 
reduce 
the mo 
the pro 
Densit 
1 
jection 
and of 
serve | 
should 
when a 
in diam 
as gre: 
density 
in high 
F 
pared 
unshar| 
caution 
remove 
and en 
nition. 
fine st 
only in 
Plates 
printer, 
plate is 
of abo: 
the eye 
printer, 
quire a 
centime