In Making Sharp Images, it is demonstrated that stopping down can improve image quality. But also as shown, stopping down too far damages image quality, particularly for high resolution cameras like the Nikon D800E. The issue is a law of optical physics: diffraction. The diffraction 'hit' is no different from film days.
Let us take one of the very best lenses on the market today: the Zeiss 135mm f/2 APO-Sonnar (reviewed in Guide to Zeiss). The Zeiss 135mm f/2 APO-Sonnar offers performance at the level of the best Leica R APO glass, yet offers the native mount capabilities for Nikon and Canon DSLRs.
MTF at ƒ/22
What happens when you stop down the Zeiss 135/2 APO-Sonnar to ƒ/22? Can it evade the damaging effects of diffraction?
Manufacturers rarely publish MTF for the full aperture series, let alone ƒ/22. But shown below courtesy of Carl Zeiss are ƒ/2, ƒ/5.6 and ƒ/22 MTF charts. These are actual measured MTF charts from a production serial-numbered lens (mine). Not computed charts. Not fantasy charts with skewed wavelength measurements. No, the real deal from a real lens.
The ƒ/2 and ƒ/5.6 charts are superb, the only necessary comment being “wow”.
The ƒ/22 chart shows strongly degraded contrast on coarse, medium and especially fine details (10, 20, 40 line pairs/mm). This is diffraction at work, and on the D800E it becomes visible at ƒ/8 with a top-grade lens, or even ƒ/5.6 (subtly) with the right choice of subject matter. By ƒ/11, diffraction dulls things visibly, ƒ/16 looks soft, and at ƒ/22 a gray haze permeates the entire image and destroys micro contrast. Of course, ordinary lenses don’t seem to fare so badly, because a mediocre ƒ/5.6 makes ƒ/11 look just peachy.
Because the MTF starts at such a high level to begin with, the 135/2 APO holds up better than most lenses, essentially bumping up against the limits of diffraction.
Observe that MTF for fine details (40 lp/mm, bottom pair of lines) drops from about 84% (world class) to about 48%, a dulling that few would find acceptable. In fact, 50% MTF is the cutoff for what one deems acceptable to be perceived as sharp to the eye. So we can see that ƒ/22 cannot deliver any sharp detail by that standard.
In this example at very close range, ƒ/2 suffers from having so little depth of field that I could not decide on focusing on the surface of the eye or the iris just below it (ƒ/2 can be very sharp, it’s essentially mostly out of focus here). The ƒ/5.6 result shows peak sharpness and contrast; compare to ƒ/22. All apertures well-sharpened as shown here. A massive dose of sharpening along with a contrast whack can make the ƒ/22 image usable, but it lacks all brilliance.
Actual pixels from Nikon D800E.
Stephen P writes:
This is one of the simplest explanations of diffraction I have ever seen. Seeing truly is believing!