15 -
between a laboratory test and practical performance.
Nothing of this sort is obtainable directly from the
MTF. It might be considered that such information is
unnecessary, and that the MTF being a test of the lens
alone, should stand on its own. This might have some
plausibility if all MTF's had the same shape. The MTF's
of photographic lenses in general fall very rapidly from
zero frequency to some low value such as 0.2 or 0.3, and
continue at a very low level, say 0.1 or less, to the
cut-off frequency for the aperture. Depending on the
state of correction, different MTF's may cross at various
levels. When such crossing occurs, how shall the lenses
be evaluated? Some weighting technique, taking account
of the emulsion and the human observer is clearly required.
It has been suggested (4,5) that since each emulsion has a
"threshold", indicating the minimum modulation it can
detectably record at any spatial frequency, the area bet
ween the MTF and the threshold would be a figure of merit
for the lens in combination with that emulsion. This is
not a rigorously proved relationship, but might be a useful
practical guide. Thus in figure 1 it is not clear from
the MTF's alone if lens A is superior or inferior to B.
But the effective area of A is relatively greater than B for
use with Tri-X emulsion, while without reference to the
emulsion, B might be considered superior because its MTF
continues to higher frequencies at a much higher modulation
than A. Experience shows that lens A would give much better
image quality in general aerial photography and the illus
tration emphasizes that a lens cannot be evaluated solely
from its MTF without reference to an emulsion. A fairly
realistic appraisal of the relative importance of different
parts of the MTF can be obtained if it is multiplied by the
MTF of the emulsion with which the lens will be used. Diff
erences between lenses are then reduced in apparent import
ance and regions of very low MTF may appear to become in
significant.
Since the MTF is in frequency space a figure of merit
might appear to be obtainable from the efficiency with which
it passes the spatial frequency spectrum of "the typical
scene" at image scale. However, there would be great difficulty
in agreeing on a typical scene and measuring its spectrum, and
the requirement for including the emulsion and observer would
still have to be observed.
The most practical approach to this problem involves
the translation of the MTF into a graph of contrast versus
size for some specified form of target. There are some advant
ages in using a single bar target for this purpose. (6) The
size/contrast graph provides information in a readily compre
hended form. In conjunction with empirical data for minimum