RE: [Rollei] Schneider - Zeiss debate

[Richard Knoppow <dickburk >]

At 04:46 PM 03/25/2002 +0100, you wrote:
>> with all respect, these comparisons ain't worth nuthin, since
>> they are only
>> practical and without any figures and numbers or even scientific
>> test-patterns ;-)
>
>Jan, I fully agree. That's the reason why I wrote that the site owner was
>suggesting(!) that his(!) Schneider is contrastier than his(!) Zeiss ;-)
>
>I actually have tried a more scientific approach of my "bookshelf" test by
>including two USAF1951 test target and shoot from open aperture to f16. I
>was doing this with the Rolleikin (why wasting 120 when I can only evaluate
>35mm ;-). But during my attempts, I broke a spring in the Rolleikin counter
>when advancing the film. This left me really frustrated and in my
>frustration, I did not completely rewinded the film before opening the
>camera and my tests with the 2.8F Planar was wasted :(
>After that, I started "repairing" the counter but could not get this counter
>to count again :(
>I probably still have the results of the 3.5F Planar on film but I need to
>finish this roll first with the 3.5E3 and Vb(which is now taken apart for
>cleaning).
>
>> But to become a little more serious, I think you can't tell much
>> by judging scanned
>> prints of unknown proveniance. I'd need some Velvia (or Provia
>> 100F) slides and my
>> magnifier (or even better Technical pan negatives).
>> Jan
>
>What kind of magnifier do you suggest using?
>
>Siu Fai
>
  Good Scotch (Oh, wait, that's pacifier not magnifier). 
  Tests like this tell you something about the system performance,
including film and any focus errors introduced by the finder and film
flatness, but not a whole lot about lenses, unless they are grossly
different. 
  Real lens testing should be done by examining the aerial image,
preferably using an electronic sensor of some sort although visual
examination using a very high quality low power microscope was the
traditional method for decades. 
  It takes very little defocusing to affect a 100 l/mm resolution chart. 
  The shape of MTF curves can be useful in analysing the type of aberration
a lens has and will show defocusing effects. But to be really meaningful
many MTF curves must be taken at various points of the image, from center
to edge, and at a variety of stops. 
  To really answer a question like Xenar vs: Tessar or Xenotar vs: Planar
would require plotting the spherical aberration, astigmatism, chromatic
aberration, etc., using an optical bench. MTF will also vary with color,
depending on the degree of chromatic correction of the lens. 
  Lenses also have an effect similar to acutance in film. If you look at an
MTF curve as a frequency response curve, as for a loudspeaker, its shape
will affect the appearance of the image. Some lenses start to fall off in
the midrange but fall off slowly, so that they still have output at high
resolution numbers. Some stay fairly flat in the mid range but fall off
rapidly after some point. The second type is apt to appear sharper to the
eye than the first kind, although its _limiting_ resolution may be much
lower. 
  The shape and cut off of the resolution curve is often tailored for
lenses used on electronic sensors to avoid aliasing.
  The total resolution of a combination of lens and film can not be
calculated in any simple manner. The curves for the two must be combined by
convolving. The old rule of 1/t = 1/L + 1/F or using the square root of the
sum of the squares is _very_ approximate as applied to the limiting
resoluton but not really useful in giving any idea of the apparent sharpness. 
  On film resolution can be measured but the possible errors of the method
must be understood and, preferably, eliminated as much as possible. The use
of glass plates or a vacuum back is mandatory of film curvature is to be
eliminated. The glass reseau available on some models of Rolleiflex
certainly eliminates film curvature and wandering of the film plane, but
introduces a flat plate, which brings some errors along with it. This is
not a trivial problem.
  On a related subject, many years ago Kodak built a special camera for
measuring film resolution (its described in the Journal of the Optical
Society of America somewhere). In about 1948 they discovered the special
lens used in this camera had substantial and unpredicted aberrations. A new
lens was designed with limiting resolution of 500 lp/mm, and all Kodak
films were re-measured. The result was a near doubling of the resolution
numbers for the films. 
  The lens is described by Rudolf Kingslake in on if his lens design books
and I think in another JOSA paper. 
  Also: A single number for resolution is quite misleading. For what limit
is it stated? For electronic transmission and filters the bandwidth is
often stated as the half amplitude value but for sharp cut off filters
limites for other amplitude values must be stated to characterise the
system. Optical systems seem to use something like 10% response rather than
50% response. It probably depends on the application. For visual
determination, using the usual bar chart, the limit is one of visual
contrast. A caveat, bar charts can display false resolution. One fellow who
posted to the large-format group, insists that any decent lens is capable
of 400 l/mm resolution in the center of its field. This is simply hogwash.
What he was seeing was some false resolution due, probably, to the
aberrations in the microscope system he was using to examine the aerial
image. Even calculated values for ordinary lenses are never even close to
this value. 
- ----
Richard Knoppow
Los Angeles, CA, USA
dickburk  

------------------------------