Mailinglist Archives:
Infrared
Panorama
Photo-3D
Tech-3D
Sell-3D
MF3D
|
|
Notice |
This mailinglist archive is frozen since May 2001, i.e. it will stay online but will not be updated.
|
|
Text of Zeiss paper (looooong)
- From: T3D john bercovitz <bercov@xxxxxxxxxxx>
- Subject: Text of Zeiss paper (looooong)
- Date: Sat, 18 Oct 1997 08:35:53 -0700
Resolving Power and Contrast
Erich Heynacher and Fritz Kober
Reprinted from
ZEISS INFORMATION No. 51
Interest in photography - be it
professional or as a hobby - is growing
constantly. It is therefore not at all
surprising that more questions are
asked about the performance of lenses
before a person decides on the purchase
of a camera or a new lens. Such
questions cannot be answered in just a
few words because, in evaluating a
photographic lens, a great number of
factors has to be taken into
consideration. The following article
may furnish valuable hints to anyone
interested in the study of image quality
in photographic lenses.
Lens manufacturers are frequently asked
what the resolving power of their
lenses is, in the assumption that the
resolving power is a criterion for
quality. However, the German optical
industry does not disclose any figures
concerning resolving power for the
following three reasons: 1. The
disclosure of figures on the performance
of lenses is rather problematical. A
proper evaluation of the quality of a
lens is possible only if several
numerical values are known, which -
for example - give information on the
sharpness of the lens over the entire
image area from the center to the very
corners of the field, the image quality
at different lens openings, starting with
full aperture, the image field
illumination (vignetting to- ward the
corners), the distortion, etc. These
performance data must be care- fully
balanced and different emphasis must
be placed on the various values
depending on the intended use of the
lens, e. 9. for general photographic
purposes, for portraiture, for
enlargements, or copy work.
Consequently, only a person fully
experienced in the testing of
photographic lenses will be able to
evaluate such data without the risk of
misinterpretations. Furthermore, any
data on the performance of a certain
lens is useful only if these values can
be fully applied in the regular
production run of a given type of lens.
2. When determining the
resolving power, it is possible to
influence the results quite considerably
by using photographic emulsions and
development techniques which deviate
from the common practice. As a result,
as long as no standard specifications
exist, manufacturers may be tempted to
apply measuring techniques which may
upgrade the results but do not
correspond to actual conditions.
3. Finally- and this is the main
reason why the German optical
industry refrains from giving lens
resolution figures - the re-solving
power is not as important a criterion
for image quality as is generally
believed. This statement will be
proved in the following paragraphs.
A number of photos were taken with
perfectly uniform image quality over
the entire field, so that it is not
necessary to balance center sharpness
against edge sharpness. Neither do
these pictures show any visible
vignetting nor distortion. We may,
therefore, use these photos without any
reservation for comparative image
quality tests. Let us first consider
photos 1 and 2. They are both of poor
quality. If you had to choose, however,
which of the two would you prefer? At
first glance, you would probably select
photo 2. At least, that is what
everyone did who saw the pictures up
to now. Photo 2 appears to have much
higher contrast than photo 1, the latter
giving the impression of being fuzzy.
However, if you take a close look, you
will notice that photo 1 has a
considerably higher resolution and
better definition than photo 2 which,
upon close examination, is rather
unsharp. We do not know for which of
the two pictures you will finally settle.
Your decision will largely depend on
your personal preference. In any case,
however, you will certainly not find
photo 1 so much better as the
resolution figure would have it.
Actually, the lens resolution figure for
photo 1 is twice as high as that for
photo 2. Let us now turn to photos 3
and 4. Their image quality, or at least
the image quality of photo 4, is much
better than that of the pictures
previously studied. And yet, it is
photo 3 which has the higher
resolution. This is easily recognizable
in the ornaments of scepter, crown and
robe. But unless we take a very good
look, we do not realize it. The higher
resolution, therefore, is of no
consequence for the impression created
by the picture. Far more striking
examples could be presented in photos
of really good image quality.
Unfortunately, this cannot be done here
because the screens employed in the
printing process would destroy the fine
detail which we need to prove that a
poorer photo actually can have the
higher resolution. But with the aid of a
trick we can create similar conditions
to those en- countered in very good
pictures. For this purpose, we need
only choose a larger viewing distance,
in other words, we observe photos 3
and 4 from a distance of, say, 3 or 6
feet instead of from the normal reading
distance of approximately 10 in. There
will then be absolutely no doubt about
which of the pictures is the better one,
and we realize of how little avail high
resolution can be. The fine detail
reproduced in the photo with the
higher resolution can no longer be
clearly seen. In other words, it does
not matter whether it is resolved or
not. The above examples have
demonstrated that it is possible to have
poor photos show- in high resolution,
and good pictures with moderate
resolution. Two photos of identical
resolution may be entirely different in
image quality. Just compare photos 1
and 4 from a distance of about 3 feet.
Both pictures have the same lens
resolution figure - but what a difference
in image quality. It is evident that the
resolving power - or at least the
resolving power alone - is not the
decisive criterion in evaluating the
quality of photographic lenses, and this
is what we have tried to show. At this
point the question may be raised
whether the poor image quality in
photos 1 and 3 may not primarily be
the result of softer printing as compared
to the printing of photos 2 and 4. The
answer is no because all four examples
were taken on identical photographic
material and treated alike during
processing. There is, of course, a
possibility of improving photos 1 and
3 to a certain extent during processing
(by using high-contrast photographic
material with the added control of
dodging, redevelopment etc.), but the
improvement will be rather
insignificant even with a subject of
very few intermediate tones, as is the
case in our examples. With true half-
tone pictures, the above technique
would be practically use- less, because
any attempt at improving the image
through the use of high-contrast
photographic material would
automatically lead to a decrease of
tonal gradation. But, you will ask,
what is it that distinguishes photo 4
with its modest resolution from photo
1 which has the same resolution as 4,
and photo 3 with high resolution?
What then, if not lens resolution, is a
valid criterion for image quality? Our
comparative photographs have shown
that the image quality is not so much
deter- mined by the definition of fine
detail as by the manner in which the
more easily perceptible, larger
structural elements in the picture are
reproduced. The more faithful the
contrast ratio, the better the image
quality. It is obvious that the degree of
accuracy of contrast rendition in an
image depends on how coarse or fine
the respective structural elements of the
image are. The contrast in very coarse
structural elements will always be
reproduced to a fairly accurate degree.
On the other hand, there is no such
thing as "true contrast rendition" as
soon as we go beyond the limit of
resolution. Details, however, the size
of which lies between these two
extremes, will not be reproduced
absolutely true, but still with more or
less good contrast. In order to get a
better idea of the image-forming
properties of optical system, it seems
appropriate to look at contrast
rendition as a function of the size of a
given detail. A very simple method is
employed for determining contrast.
Screens are used with equidistant white
and black lines of the same width. The
number of lines in 1 mm space serve as
a measure of the size of a detail. Figure
5 shows sketches of two such screens.
Figure 6 indicates the contrast
rendition as a function of the screen
size employed for the test. Line 5
represents the ideal contrast rendition
in a perfectly "true" optical imageU
Curves 1 to 4 represent the contrast in
our four sample photos. The numbers
of the curves correspond to the
numbers of the illustrations. It will be
noted that in photo 1 the contrast of
even coarse details (a few lines per
millimeter) is strongly reduced. This
is the reason why the picture looks
"soft" or "fuzzy-. By comparison, the
contrast of photo 2 falls off more
slowly from the ideal value. This
photo therefore seems to be richer in
contrast. It is true, of course, that
photo 1 contains much finer details
than photo 2U though at the sacrifice of
contrast
In order to render fine details visible,
little contrast is required. Under
favorable viewing conditions, about
5% will suffice. This value is marked
in the curves by a circle. Each circle
indicates the limit of resolution, and
the resolution figure is given by the
respective number of lines/millimeter.
Thus, in photo 1 the resolution is
about twice that of photo 2, while it is
identical to photo 4, as was mentioned
before. - What was said about photos 1
and 2 is applicable also to It is non-
existent in practice because a certain
unsharpness is caused by diffraction of
the light at the diaphragm opening
even in a perfectly error-free lens
Moreover, an additional unsharpness is
produced by the scattering of light in
the photographic emulsion It is known
that any unsharpness leads to a loss m
contrast transfer. pictures 3 and 4 with
the only difference that the curves of
contrast transfer show higher values
than the curves 1 and 2, thus
representing a higher image quality.
All this can be read at a glance from
the contrast transfer curves. But that is
not all.
If we include in our study the limit of
resolving power of the human eye, as
represented by the vertical line 6, this
line will divide the range into two
regions, one in which fine detail is
perceived by the eye (left of the line),
and the other in which fine detail is
invisible to the eye and thus not of
interest (right of the dividing line).
Line 6 applies when looking at the
photos from normal viewing distance.
As the viewing distance increases, the
number of perceptible details decreases,
and the dividing line is displaced to
the left. For observation from a
distance of approximately 3 ft., for
example, line 7 applies. Therefore,
within this range of detail perception,
contrast transfer curve 4 is closer to the
straight line 5, representing the ideal
lens, than contrast transfer curve 3.
Consequently, photo 4 must have a
better image quality than photo 3.
Considering all these facts, there is no
doubt why the resolving power as such
is not a suitable criterion for image
quality. It is only a point on the curve
of contrast transfer, and we have seen
that in good images this point even
lies far outside the important image-
forming range. What really matters is
the contrast rendition within that part
of the range in which detail can
actually be perceived by the human eye
at normal viewing distance.
Fig b: Examples of One screens as
used for the measurement of contrast
transfer. The screen size is expressed
by the number of lines/millimeter
(frequency).
Fig. 6: Contrast transfer curves for the
ideal image (5) and sample photos 1 to
4 (curves 1 to 4) The limit of resolving
power of the eye for a viewing distance
of 10 in is shown by the straight line 6
for a viewing distance of approximately
3 ft. by line 7. Multiplied by to the
figures given for detail size could be
the conjugate numbers of
lines/millimeter in the negative.
------------------------------
|