Hi,
MTF measures contrast at different resolutions. Take my P45+ back. It has 6.8 micron pixels, so it can resolve 1000/(2*6.8 )=73.5 lp/mm.
If you look at pixels in an unsharpened image, they will be pretty soft. That is because lens and sensor together have low contrast at this level. If you zoom out to "fit on screen" the image will look OK for the lens transfers a lot more contrast at lower frequencies.
MTF can be measured either by analysing an image, or on an optical bench. Here is one of my MTF-plots. Any MTF-plot of an unsharpened lens looks like this.
Please note that the pixels here are at 73.5 lp/mm everything to the right of that is information that the sensor cannot handle. That is the source of moiré. The resolution limit of the sensor is often referred to as "Nyquist" meaning the "
Nyquist Limit". All MTF above "Nyquist" is causing
aliasing artifacts.
This kind of MTF plot describes only a single point in the image in one direction. So it is useful in some contexts, but actually says very little about the lens.
So the normal way is to measure MTF at many points and then show curves for 10/20/40 lp on vertical axis and distance from the centre on the horisontal axis. Like this one, from a Planar 120/4 at closest focus. Normally they make one for full aperture and one stopped down:
Why two curves? Because they measure MTF in two directions, sagittal and tangential.
In general, the curves should be as high as possible. If the curves drop evenly from centre to corner like in the figure below, the problem is probably field curvature. The Hasselblad Macro Planar is designed for short distances and has a lot of field curvature at infinity:
The Zeiss 120/4 Apo Macro for the Contax 645 is a much more advanced design, while the Hasselblad version is a traditional "Double Gauss" like virtually any "Normal lens", the Contax Macro contains an extra lens group and has floating elements, that is one group moves with focusing, thus keeping field flat at different focusing distances. I have no MTF curves to show for that lens, but they are good.
What is my "rule by thumb" for a sharp lens? MTF at 20lp/mm should be above 80% across the diagonal. Curves for 40 l/mm should be above 60%. If the curves separate I take the average.
Very good lenses, like the Zeiss Otus can reach above 80% of MTF at 40 lp/mm.
A very good (but long) tutorial on reading MTF is here:
http://www.zeiss.com/content/dam/Photography/new/pdf/en/cln_archiv/cln30_en_web_special_mtf_01.pdf
http://www.zeiss.com/content/dam/Photography/new/pdf/en/cln_archiv/cln31_en_web_special_mtf_02.pdfA very good video on MTF by Canon's Larry Thorpe is here:
https://youtu.be/iBKDjLeNlsQThis one by John Galt is also look worthy:
https://www.youtube.com/watch?v=v96yhEr-DWMBut these two videos are video oriented.
One thing to keep in mind is that MTF at lower frequencies dominates perceived sharpness. So, say the 20 lp/mm curve is far more important than 40 lp/mm. But when we pixel peep we actually look at 73 lp/mm (on the P45+).
MTF curves describe how the optical system behaves, but they normally only illustrate sharpness in the plane of focus and most often at infinity. The plane of sharpness can be very thin, so out of focus rendition is also important, and the normal set of MTF curves will not tell about that.A bad property of almost all lenses is what now days often is called longitudinal chromatic aberration. This causes magenta/green fringing at large apertures. Very few lenses are without it, here is a list I know of: Coastal Optics 60/4 macro, Otus 85/1.4, Otus 55/1.4, Zeiss 135/2.0 APO Sonnar, Voigländer APO Lanthar 125/2.5. I could add the Zeiss Superachromats, I would think. Zeiss is taking great pride in the Otuses being without colour fringing at full aperture.
This video offers some excellent insights: https://youtu.be/9cnEnRADDLoHope this is some help…
Best regards
Erik
Erik, thank for your efforts. I confess, I do not know how to interpret these graphs. Perhaps a simple tutorial?
Obliged to you,
David
