Sharpening vs diffraction

Diffraction limitations on image sharpness are a constant issue with DSLRs. In most cases the optimal lens sharpness seems to be about f8 yet a lot of photographers seem to be opting to use smaller apertures such as f11-16 and accept some of the loss of sharpness in order to increase depth of field. I have seen several authors suggest that this loss of sharpness can be compensated for using Photoshop but they never exapnd on that and give any limitations of the method or general guidelines.

So my question is how much of this loss in sharpness can we realistically expect to recover using sharpening routines in Photoshop, etc.? How far can we push the envelope say using f22 and then get back to the same level of sharpness as would be expected at say f8 with less post processing? Certainly there must be a point beyond which even sophisticated sharpening routines can recover.

Also, the level of contrast in the original scene must come into play since sharpening as I understand it is heavily dependent on the contrast in the digital image data. Is there a general rule of thumb that says for example you can get away with more diffraction losses if the image is very contrasty? How many f-stops can we go above our maximum lens sharpenss and still expect to recover?

Larry

Larry, one technique I've seen folks use on this site it to shoot two shots at f/8 or f/11, one near-focused and one far-focused, then mask 'n blend the two in Photoshop. That would make one sharp image, and I need to practice that technique. However, the whole thing does suggest a "forget trying to shoot at f/16 or 22" scenario, if they are going to those lengths to get a fully sharp image. The acceptability amount of sharpeness will vary between viewers of an image I guess, so the best way to nip it in the bud seems to be two f/8 exposures, blended. I guess that doesn't actually answer your question though!

It's simple, use a TS-E lens and tilt or swing and shot at f/8 or f/11

F/22 kills detail and the only time I ever us it is if I want a sunstar.

First, diffraction is an issue with any lens, not just those on DSLRs. Shooting at f/22 on film or on a DSLR with the same lens will yield the same diffraction softening. Of course, how much it shows up in the image will depend on a few different things (noise, sensor size, DOF, etc.) As an example - a smaller sensor will show diffraction effects sooner, but it has more inherent DOF (all else equal) so you don't need to stop down as much for front-to-back sharpness. As usual, everything is a trade-off.

This loss of sharpness due to diffraction can't be compensated for any more than softness due to the lens itself, or the filters in front of the sensor. They can all be recovered by various sharpening flows/routines, but none perfectly. So there's no way to "undo" the softness caused by diffraction. I've never seen any sharpening that specifically addressed diffraction either.

I think you might just have to try a couple of scenes on a few different apertures and judge for yourself what you think is acceptable. It WILL likely vary by scene, as you noted, because a low contrast scene is not going to need to look as sharp as a high contrast, high detail scene.

Note that that's not the same as saying that a low contrast scene can't be as sharp as a high contrast scene. It's just that preference and perception is for high contrast to be sharper than low contrast. "Sharpness" itself is defined by the transitions from one tone to the next. Something like a soft foggy scene will not have big fast transitions, so it won't need to be as sharp.

For me, I've shot at f/16 and made prints up to about 12x16 and didn't see any issue with it at all. But I'm not super critical about sharpness and detail since I print things mainly for myself...

Looks like I'm reading this thread about 24 hours after David's post, so I hope folks will still see it. This is a question I wondered about several months ago myself when reading either another thread or a magazine article on the same subject The short lesson was don't stop down too much to achieve DOF because you will lose sharpness due to diffraction. Yet, and this is where I got confused, there was either a photo w/caption or an article with photos by either (I can't remember) Guy Tal or Marc Adamus displaying a mountain scene shot at f22 which appeared in OP. When I saw the aperature number, I wondered why they would stop down so small in view of the diffraction issue. So, short of doing what Jerry suggests with a tilt/shift lens, is increased DOF always a trade-off with sharpness--trying to make the far ground sharper with increased DOF only to lose sharpness to diffraction???

Confused-
Ken

...is increased DOF always a trade-off with sharpness--trying to make the far ground sharper with increased DOF only to lose sharpness to diffraction???

In a word: yes.
Diffraction increases as the aperture gets smaller.
DOF increases as the aperture gets smaller.

Without a view camera or a tilt/shift lens on an SLR, the best advice is to stop down just enough to get the DOF you want but no further. And stop worrying about diffraction because you can't do anything about it. Insufficient DOF is usually more noticeable than diffraction.

Cheers,

One way to consider the trade-off between loss of IQ from diffraction and greater DOF from a small aperture is to remember that a good part of what we think is in focus in a photo with relatively much DOF isn't. But it's close enough to being in focus to let our eyes and brains perceive sharpness. In contrast (this is particularly evident in extreme close-up and macro shots), when we shoot at larger apertures there are pretty obviously OOF areas that we have no difficulty at all in seeing as unsharp. One post-processing technique I've come up with is to focus blend not just several f/8 shots with differing POFs but to include an f/16 exposure (and sometimes f/22, f/29...) in the stack. The focus-blending software will select the main subject from the f/8 shots because they're sharpest (no diffraction effects) and it will also select parts of the diffraction-compromised small-aperture image(s) because they're close to being sharp. The viewer of the result will see the main subject in fine detail and will perceive quite a bit of the diffraction-compromised area as part of the apparent DOF.

But "it all depends": on the subject itself, on the size of the photo as viewed, on the output medium, on what one thinks one is looking at (is it a photograph, so I zoom in to 100% and notice even hints of softness? or is it a mountain or flower, so I stand back and appreciate the subject itself?

Unless you buy T/S lens or shoot LF, this is an age-old trade off. Image stacking is one solution - not too easy to execute as changing the focus point can move the camera/tripod. But with softwares that can align automatically, this process is gaining popularity. But I still find this process to be cumbersome when shooting under fleeting light. When you have a lot of time in your hand, it's fine.
I hope that the future DSLR will have some kind of feature like "Focus bracketing" - just like "Exposure bracketing" - should not be too difficult to implement for manufacturers. But I am not sure whether they have this item on their list to do

Arnab - I about fell over to learn that Pentax was offering in-camera HDR. So, I bet someone is already working on in-camera "image stacking."

Pete D.