Update: An enclosed crop of a flower from the same test image shows the difference on a 3D object. The flower in front is in the same plane as the test target, within a couple of millimeters.
I was thinking about those Schneider 55/2.8 LS curves showing a sharp drop in MTF at 22 mm. How bad is that?
In my book, a lens having say 50% MTF at 40 lp/mm would be a good lens. On the other hand, a 6 micron pitch sensor would resolve 83 lp/mm. As the lens drops to near zero at 60 lp/mm, something will be lost.
The drop in MTF probably depends on field curvature, there is some residual field curvature which is overcompensated near the edge. So field is wavy.
What I wanted to see was:
1) How much is the defocus?
2) How much sharpness will be lost?
The way I tried to find out I took a series of exposures with a 50 mm lens at 2 meters. I developed the images in Lightroom with no sharpening and measured MTF. I found that a defocus of 7.5 cm at 2 meters drops MTF to values corresponding to the Schneider plots. So the maximum curvature of the focus plane is about 7 cm at 2 meters. I guess it's about the same as the distance from an eye to a nose. Note: These tests were not made with a Schneider lens but with a 50 mm Minolta lens on 24MP APS-C DSLR. MTF-measurements don't care about sensor size, so I would suggest the methodology used is relevant.
The MTF curves are shown below:
The curves are MTF data measured by Imatest. I also added the approximate MTF data from the Schneider plots at 15, 30 and 60 lp/mm on axis and at the minimum MTF position.
The images were shot with a 3.9 micron pitch camera but downscaled to correspond to 6 micron pitch, that would be reasonably close to P65+ and IQ180.
The unsharpened images are like this:
For actual pixel view check this link: http://echophoto.dnsalias.net/ekr/Images/Schneider55/no_sharpening.png
I also sharpened the image using 1, 160, 1 in Photoshop and got the following image pair:
For actual pixel view check this link: http://echophoto.dnsalias.net/ekr/Images/Schneider55/Sharpened_1_160_1.png
The focus error is about 60 microns in the focal plane.
My conclusion? The loss of sharpness is clearly visible, but probably not catastrophic. If critical detail is placed in that area, refocusing may help. Stopping down reduces the problem.
My experiment also illustrates that the problem could pass undetected. There is some loss of image detail but I guess it would be hard to see if we didn't have a slightly sharper crop to compare with. Sharpening increases apparent edge contrast but the values used here are in my view a bit to heavy.