Diffraction?

[j.weegee]

I was always under the impression that diffraction occurred on the aps-c format around F16, but I have heard some say it starts alot sooner. Is diffraction strictly dependent on format size or does the lens have some say in when it occurs and if so what in a lens exactly causes more diffraction than other lens'. Is it the number of aperture blades or rounded vs. non rounded blades? Any information will be greatly appreciated.

 

[480sparky]

Diffraction occurs in all lenses at all apertures, across all formats. The differences are at what point they become objectionable.

 

[Solarflare]

That depends upon lens and camera. For example, this test claims my 35mm f/1.8 on the D5100 (which has the same sensor as the D7000 used in this test) performs best around f/4. After that, resolution already goes down again.

 

[o hey tyler]

That depends upon lens and camera. For example, this test claims my 35mm f/1.8 on the D5100 (which has the same sensor as the D7000 used in this test) performs best around f/4. After that, resolution already goes down again.

Yeah, resolution goes down if you ONLY judge center sharpness, which is only a portion of the overall image quality. You need to learn where sharpness really starts to taper off, because I'm sure it performs pretty well up until f/7.1 or f/8 without even looking at the MTF charts.

 

[amolitor]

Diffraction depends on the wavelength of light, but any lens will diffract light.

A lens that is limited only by diffraction projects your image onto the sensor with some softness related to diffraction. For instance, an infinitesimally small dot will resolve as (roughly) a small dot of diameter equal to about the f-stop in microns. At f/2.8 you're getting a maybe 2 to 4 micron dot, depending on what color of light we're talking about. It's not a dot, it actually smears out all across the frame, but it's "mostly" a dot, so we think of it as a dot.

Generally people assume that as long as this theoretical dot is smaller than the dots their sensor sees, then diffraction "doesn't count" which isn't quite right, but is pretty close. So, yeah, your normal crop sensor from about now runs up against diffraction limits around f/5.6 or f/8.0 or somewhere near there, depending on colors and the details of the sensor and so on. A full-frame sensor with the same number of pixels won't hit diffraction limits until later, since it's using big fat pixels. A D800 which uses a huge number of itty bitty pixels again, will run into limits about the same place a current crop sensor does.

Basically, don't worry about it too much. When you stop down a whole bunch, you're gonna get a little softness.

Search terms: Rayleigh Limit, Airy Disc

 

[pgriz]

It's actually a pretty simple test to do, to determine when diffraction affects your lens. Put a newspaper with fine print (classified would do) about 10 ft. away for a normal lens and maybe 20 ft. away for a telephoto, put your camera on a tripod, focus on the print using live-view, and take successive pictures while varying the f/stop from the widest to the smallest. Then compare the images at each f/stop with each other. You'll generally find the fine detail to be a little soft wide open, then becomes sharp anywhere from f/5.6 to f/11, and start losing detail above that. It is lens specific, so you have to check each lens. But for my lenses, I usually limit myself to f/11 and go higher only if I really need the DOF and the fine detail is less important.

edit: It's important to also check the edge detail, not just the center point. You might be surprised at how much the edge detail deteriorates.

 

[j.weegee]

Perhaps, it was worth mentioning I am currently shooting with a Canon T3i/600D with a Canon EF-S 18-55mm F/3.5-5.6 II kit lens. In my experience with this setup F11 is a sweet spot for great sharpness and limited to no diffraction. Its my go to F number if I want to do a landscape shot.​