Depth of Field Oddity/Experiment

[jcore]

I conducted a Depth of field (DOF) experiment to prove background image clarity (I know it's Photo 101 but really wanted to play with some of the settings).

Test:
Tripod, a few objects set at varying distances, Canon EOS Rebel XT with EFS 17-85 lense, daylight (in shade), metering set to evaluative, AWB auto, focus on ceramic girl's eye, timed shutter release.

- Photo #1: ISO 100, lowest f-stop (5.6), 1/60th second (eg: least DOF)
- Photo #2: ISO 100, highest f-stop (f32), .6 seconds (eg: most DOF)

Question:
Understandably the background is much clearer (as expected) but curious why the girl's eyes and the cherry liquor bottle are slightly less clear in Photo #2? Is this typical? I repeated the experiment with all metering settings and with Daylight and Shade WB settings and the same phenomenon occurred. (Hopefully, the attached images are clear enough to see these differences).

[sk66]

At f32 you may be stopped down beyond the "circle of confusion"

[OsmosisStudios]

Looks to me like the focus was actually on the edge of the bench near the doll. As SK66 mentioned, f/32 induces diffraction, which will soften things.

[jcore]

Thanks both of you. I'll play with other stops.

[BCampbell]

SK66 has it right, specifically read the section about diffraction limits.

[Elmo]

Past about f11 diffraction is a killer on digital stuff. You might want to expand your test to using different focal lengths which also change DOF

[BCampbell]

Originally Posted by Elmo

Past about f11 diffraction is a killer on digital stuff.

Enhhh not exactly. It depends heavily on the camera and the aperture where it actually starts having a real-world effect is much higher.

Two good places to start on this are the DLA chart and explanation below it here at The Digital Picture and This excellent explanation of diffraction cutoff frequency.

[OsmosisStudios]

That may be true, but as a general rule on an APS-C sensor, f/11 and smaller is going to be diffraction inducing.

[BCampbell]

Originally Posted by OsmosisStudios

That may be true, but as a general rule on an APS-C sensor, f/11 and smaller is going to be diffraction inducing.

Okay, but when will it affect your final output?

[OsmosisStudios]

Originally Posted by BCampbell

Okay, but when will it affect your final output?

That depends. Is your output to the web or to print? If the former, then probably never. Most web images are seen at less than 1000px on the longest side, and with even a 7mp image, that's a 300% DOWNsize. That being said, I have a few shots at f/22 that I can tell are soft enough even at that size.

In print, though, things are different. Even an 8x10/12 is big enough to start seeing some serious issues with diffraction. If it's bad enough, you can even see it in 4x6s and 5x7s.