I previously discussed how camera shake and focus accuracy can affect maximum image sharpness with the Nikon D800 (or any DSLR, for that matter). Another contributing factor to image sharpness is lens performance, especially with respect to aperture. While the primary use of aperture is to control depth of field, most lenses just aren’t quite at their sharpest when used at their extremes.
Wide-open, most lenses will have sub-optimal performance. Contrast and sharpness will be reduced, and you might also see light fall-off (corner shading). As you stop the aperture down, both sharpness and depth of field increase, and light fall-off diminishes. How much you need to stop down to get razor-sharp images is related to the quality of your lens. Some of the telephoto prime lenses, like the Nikon 300mm f/4 lens, are very sharp wide-open, especially compared to a consumer-grade counterpart, like the 70-300mm f/4.5-5.6 VR Nikkor (a good lens, but not nearly as sharp wide-open). Higher quality lenses usually require less stopping down to get optimal sharpness.
However, beyond certain apertures, images will actually start to soften due to the effect of diffraction. Diffraction softening is a physical function related to the size of the pixels on your camera sensor. The smaller the pixels, the more noticeable diffraction effects can become.
In this review, I primarily wanted to examine diffraction with the D800e. I used my Nikkor 24-70mm f/2.8 AFS G lens mounted on a tripod. I focused with Live View for maximum accuracy and used MLU to eliminate any camera shake. Images are in-camera JPEGs with Picture Control set to Standard and sharpening at +4.
The full-frame image is here:
Here are 100% crops from the center of the image showing the sharpness as you go from f/2.8 to f/22 (click images to enlarge):
If you examine these images, you’ll see that the sharpest images are from f/5.6-f/11. Beyond f/11, you can see the effects of diffraction softening.
With any lens, you’ll get the sharpest images stopped down slightly from the maximum aperture. How much you need to stop down is mostly a parameter of your particular lens. With the 24-70mm f/2.8 AFS G Nikkor, you can get very sharp images at f/4, and perfectly good images at f/2.8. With the D800e, maximum sharpness is somewhere between f/8 and f/11. By f/16, you can see diffraction softness degrading fine details.
Is this important?
Some people might look at this review and conclude:
- I should never shoot wide-open
- I should never go beyond f/11 with the D800/e
If you come to this conclusion, I think you’re being a bit near-sighted. Certainly, for subjects that demand maximum sharpness, you’ll want to stay away from apertures smaller than f/11 with the D800/e. However, aperture is also about depth of field. If you want to isolate your subject, then go ahead and shoot wide-open. You can often correct minor softness in your post-processing routine, especially if you shoot RAW. The same goes for small apertures. If you need the extra depth of field or you need a slow shutter speed that can only be achieved by using f/16 or f/22, then use it! Don’t sacrifice creativity for nit-noid differences in image sharpness. Remember, the examples I show are 100% crops… the equivalent of viewing a 102″ (259 cm) wide print!
Here’s a crop of the f/2.8 shot after I processed it with some custom sharpening (High Pass and USM in Capture NX2).
8 thoughts on “D800 Sharpness: Diffraction”
Excellent demonstration of the effect of diffraction AND shooting “wide open”. As an amateur it’s hard to get past the idea that you aren’t using “all” of your lens to restrict oneself to f/4 to f/11! Your final advice is excellent though and people often get wrong ideas from “pixel peeping” on the computer at enlargement ratios that show “problems” that will never show up in the prints. I am waiting patiently for my D800E while Nikon churns out 1 a day. I assume your camera doesn’t have any of the left side focusing problem (which wouldn’t really alter the results of your test anyhow).
I agree 100%. In these examples the f5.6 looks the best to me. But that might just be some increased contrast when changing to f8? But if needed, the f22 would be very usable (like the f2.8), with proper sharping and contrast adjustments.
Looking forward to seeing some D800 landscapes from your next outing.
Thanks Jason, your tests are very helpful to me. Do you think you would see sharpness start to fall off on all lenses at F11 or not?
Another great analysis Jason. However, I am not familiar with the photographic technical term “nit-noid”. Perhaps the subject of your next article. LOL
I am not a pixel peeper. But I think according to MTF tests this lens is technically “best” at f/4, possibly f/5.6. f/8 and the rest of course are fine as well.
Even with it’s distortion, I use it for architecture all the time at f/8 or f/11.
Fair point, Geoff, but this test was to show two things:
1) how lenses change sharpness at different apertures
2) how diffraction becomes noticeable on the D800
Each lens has its own special characteristics and these results shouldn’t be taken as fixed recommendations (other than to see sharpness with the 24-70).
Thank you, Jason! Your tests are very good and are helping me a lot. I decided to change my Canon 7D with a Nikon D800 in my next trip to NY and you have enlightened me a lot. Bye.
Sao Paulo, Brazil
Absolutely Jason, I was actually trying to buffer your point.
Even though the technical MTF tests may show exact scientific numbers, one should never base their decision of aperture on those, instead relying on needed DOF and the individual lens characteristics.
As I mostly shoot architecture these days with my beloved D800, I have to shoot up at f/8 – f/11 for DOF, but do remind myself to shoot slightly wider when possible for max sharpness. As you mentioned, mostly splitting hairs though. I’ve found that within a general aperture range 5.6 – 14, if you DOF requirements are met, the raw conversion and sharpening techniques seem to have a bigger hand in real world final results. Again, I’m not a real pixel peeper, so perhaps I’m wrong :).