Re: Lenses and Contrast

[P3D John Bercovitz <bercov@xxxxxxxxxx>]

This is a very complicated subject, mostly because of terminology, 
and I'm by no means as up on it as I should be before posting.  
But of course that has never stopped me from writing in the 
past... 

We have "resolution" which is what you get when you look at a 
resolution chart with a lens and try to figure out what's the 
finest pitch of line spacing you can see.  Very subjective call.

We have "acutance" which is sort of an edge phenomenon where you 
see how quickly the edge of black object on a white background, 
for instance, actually changes from black to white when you're 
looking at the image from the lens.  Does the sharp edge of the 
object get smeared out?

We have "contrast" (which is what we've been talking about here) 
which is the ratio of black to white on the output side compared 
to the ratio of black to white on the input side of the lens.

There's an excellent four- or six-page brochure on this and allied 
subjects available from Zeiss for the asking.  The Zeiss paper 
explains MTF which is Modulation Transfer Function.  You present 
the lens with a series of bars a lot like a conventional 
resolution chart except that these bars are really _gently_ 
_varying_ in brightness from white to black (in a sinusoidal 
pattern) rather than just black bars on white.  You read the 
output of the lens and compare the ratio of brightness to darkness 
at the output to the ratio of brightness to darkness that you 
presented the lens with as an input.  You do this for a large 
number of spatial frequencies (bar spacings) just as you would do 
with a resolution bar chart.  Then you make a chart:


          110% |     *
per cent  100% |
contrast   90% | *      *
           80% |         
           70% |
           60% |          *
           50% |
           40% |
           30% |              *
           20% |
           10% |                      *
            0% |______________________________________
               0  10  20  30  40  50  60  70

                 Spatial frequency, lines per mm

Obviously, you have to put a new line on the MTF chart for every 
f/stop setting of the lens but I've left the additional lines out 
for clarity.

Now I've shown the output "contrast" if you will, as greater than 
100% at 20 lines per mm.  This can actually happen if the 
frequency is just right and the lens has that characteristic.  In 
fact a lens can show you four bars on a conventional res chart 
where there are only three because of this reinforcement that can 
occur at a particular frequency.  Weird, huh?

With this chart, you can sort of imagine what happens when you 
look at the output of a lens.  If what the lens is looking at is a 
high contrast image like a res chart, your eye will be able to 
count lines even though the output contrast has fallen clear down 
to 20% or below.  However, if you have a low contrast bar chart, 
like a light grey against a slightly darker grey, you can see it's 
going to be very difficult to have any perceptible resolution.

A grease-smeared clear filter will primarily mess up the high 
frequencies or at least mess them up worse.  You can think of it 
like anything that scatters light gives the most intensity in the 
direction the light was originally going but gives some intensity 
a little ways off that axis and even less intensity as you get 
farther from that axis.  So if you have a broad thing you're 
looking at (a low spatial frequency target) the grease won't have 
much effect on the middle of a bar, but at the edges of the bar, 
you could have a lot of effect (this is acutance again).

I can't think of any snappy conclusions to draw so I give you this 
as merely a few things to think about.  I welcome corrections from 
folks knowledgeable in the field.  Jim C?  Shab?  Others?

John B


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