MTF

[P3D John W Roberts <roberts@xxxxxxxxxxxxxxxxx>]

>Date: Tue, 23 Apr 1996 19:41:11 -0500
>From: P3D John Bercovitz <bercov@xxxxxxxxxxx>
>Subject: Re: Realist vs. 4"x5"

>MTF stands for "modulation transfer function".  Very simple 
>concept.  You have a special resolution chart which has bars on it 
>just like an ordinary resolution chart except instead of black 
>bars on a white background (or is it vice versa?) the lightness of 
>the bars changes in a sinusoidal fashion from black to white and 
>back, ad infinitum, ad nauseum.  So the lightness of the bars is 
>modulated.  As with an ordinary resolution chart, there are groups 
>of bars with different spacings, say 1 line per 1 cm, 3 lines per 
>1 cm, 10 lines per 1 cm and so on.  You point your camera lens at 
>this special resolution chart and then you read the sinusoidal 
>output pattern with a meter.  So let's say our special resolution 
>chart is perfect and goes from perfectly black up to some peak 
>light level in a perfect sinusoid.  Now the question is, how far 
>does the output sinusoid of the lens swing?  Say it swings from
>5% to 95% of the full input light swing.  So your transfer is 
>95 - 5 = 90% of the input.  You have an MTF of 90% or 0.9.

>The same thing applies to film.  If the lines on the chart are 
>spaced closely enough, you can well imagine that you don't get 
>the swing you should on the film; the lines just blur together 
>and you lose information.

>That's a non-technical explanation and I'm sure there's someone 
>out there who is much more knowledgeable in the area (John R?) who 
>can make corrections as required.

I would be very interested in finding out what sort of "meter" is used to
make this measurement. It sounds like the measurement is essentially contrast
ratio as a function of spatial frequency. I heard a very interesting lecture
two weeks ago on measurement of contrast ratio. For measuring displays,
apparently the traditional methods can introduce an error of 5% to 50%
or more, and very high contrast ratios or contrast ratios in images with
high spatial frequencies are difficult or impossible to measure. It might
be possible to get around some of these limitations for film or prints,
by putting the detection optics in direct contact with the image, but it's
hard to see how problems such as flare in the detecting optics could be
dealt with in measuring the performance of a lens (maybe a bare CCD array?).

>...PS: The plate resolution of a lens is limited by physical optics 
>(that's wave interference and stuff like that there) to somewhere 
>around 1400 to 1800/(f/no) for the number of lines per mm.  But 
>there are other problems which I'm not completely clear on that 
>make it more difficult to approach this limit as the lens is 
>scaled up.  So it's not all gravy for the lens designer if he 
>moves up to 4x5 or 8x10 format.

Apparently internal reflections among multiple optical surfaces, and other
phenomena associated with abrupt transitions in index of refraction result
in lens flare, which reduces contrast. And as far as I know, there's no such
thing as a "perfect" lens design - all designs involve a large number of
compromises, so performance is never ideal.

--------

>Date: Wed, 24 Apr 1996 00:27:43 -0500
>From: P3D Bob Howard  <bobh@xxxxxxxxxxxxxxxxxxxx>
>Subject: Re: PHOTO-3D digest 1294

>Re: Resolution MTF John B. and Eric G. comments:

>I think John B. has it backwards..the large format has poor resolution 
>of lenses and depends on its fine definition and gradation on little 
>enlargement of the film..near contact print. 

Bob, please quote the statement you think may be backwards - this may be
what you have in mind:

>>Date: Sat, 20 Apr 1996 12:46:53 -0500
>>From: P3D John Bercovitz <bercov@xxxxxxxxxx>
>>Subject: Re: Realist vs. 4"x5"

>>When you look at the MTF curves for lenses and for film and consider their
>>compounding, you find that in miniature format, the film drives the system
>>MTF, in large format, the lens drives the system MTF, and in medium format
>>they both contribute pretty equally.  So a fine lens in a larger format is
>>working at a bit of an advantage, not being film limited.
 
>>John B

It sounds like pretty much the same thing - a question of interpretation of
the words.

John "Remember - you can't put too much water in a nuclear reactor!" R.


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