Diffraction at small apertures, a limiting factor in sharpness?

Paperman

@Ade, Thank You, thank you, thank you ! Finally someone understanding what I mean ...

@Ironheart
"When you attempt to measure when diffraction sets in on a given lens, you are measuring the MTF of the whole lens and trying to separate out what might be due to diffraction. Someone prove me wrong. "

The theoretical limit is the utmost point achievable .We know that that limit is not reachable in reality,and we are not trying to determine the actual diffraction limit ( which would be lower/start earlier ) of a system. That isn't the point.

There is someone here saying lenses can go past theoretical limits stated by laws of optics.

Ironheart

Not me. I know bumblebees can fly ;-)

Ade

@PitchBlack

Many cameras & some photo editors have "diffraction correction". Sony, Pentax, Olympus, even Canon. "Diffraction correction" is no different than any other software (JPEG) correction, like distortion correction, CA correction, etc.

If you think about it, CA correction doesn't really correct the CA of a lens; It's a clever post-processing algorithm which masks the effects of CA (sometimes). Likewise, diffraction correction doesn't really correct diffraction. I think it works like Adobe's motion-blur correction, by trying to figure out some amount of "reverse blur" (deconvolution) followed by a liberal dose of sharpening.

@Ironheart

The key is to understand that diffraction sets a performance limit; it is not a measure. (Back to the extended analogy, think of the difference between speed and speed limit. The first is a measure, the second is a limit).

Here's a visual which might help:



The y-axis represents MTF. The x-axis represents f-stop.

Suppose we're designing a new lens. The red line is the diffraction limit. No lens -- no matter how perfect -- can exceed the MTF performance set by this red line. This line is governed by the laws of physics as we know it today.

The green line is the design limit of this lens. This line is bound by the uncorrected aberrations of the lens (e.g., spherical aberrations, CA, etc.) This is line represents the best the lens can do given design & manufacturing limitations.

The blue line represents the actual (measured) performance of the lens.

As you can see, for this lens, before around f/5.6 the lens MTF (blue line) is limited by aberrations of the lens (green line). At f/8 onwards, the the lens MTF is limited by diffraction (red line).

The red line is the same for all lenses, real or theoretical. Better lenses can reach the red line sooner. Poor lenses might never reach the red line anywhere. But no lens can perform any better than the red line, at any aperture.

It doesn't make sense to ask, "when does diffraction set in for this lens?" That question is backwards. A better question is, "at which apertures can this lens reach the diffraction limit?"

Also, we can't "separate out" the effects of diffraction on MTF. At any aperture, the lens performance either reaches the diffraction limit or not. At f/4, this lens has too many aberrations to reach the limit. At f/11, the lens is diffraction-limited.

Ironheart

@Ade, you don't have to explain it to me, I get it. I'm trying to help others understand that the MTF of every lens is a function of the red and green lines above. As I stated earlier:
As you stop a lens down, the effects of the aberrations decrease, but the effects of diffraction increase. Where the happy medium is, varies according lens design, and likely even from copy to copy.

So thanks for clarifying my point with a picture :-) I was thinking of drawing the same picture to explain it to you...

@paperman is the one asking how we determine this theoretical limit for a particular lens. I'm saying that the theoretical limit isn't lens dependent. The only thing you can do is test the lens and find the actual point where diffraction is unacceptable for you.

Paperman

@Ironheart, you can't be further away from what I am discussing :-S Is my expression of myself so helpless ? How can I / why on earth would I be asking here how to test a lens for the diffraction limit when I am the one screeeeaaming here from the start that DIFFRACTION IS THE SAME FOR ALL LENSES ?? :-S

Ironheart

I just said, "the theoretical limit isn't lens dependent" so you and I are in agreement, theoretically )
For some reason I thought you were on a different (practical) point. No worries.

TaoTeJared

The test of a first-rate intelligence is the ability to hold two opposed ideas in the mind at the same time, and still retain the ability to function.
F. Scott Fitzgerald,

Paperman

I guess we won't be getting the link on the Zeiss "diffraction shiftable" ( at an addiditional fee of $100 ) lens .....

SquamishPhoto

I would agree. I've been dabbling with focus stacking over the past few months and while it does yield some pretty cool results in the right circumstances, it is incredibly unforgiving (and impossible for objects that may move even a millimeter). For every decent image I create I have to trash files and go back to the drawing board twice because of something I did wrong.

You just need to adjust your methods. I've been doing free hand focus stacking of live bugs, flowers and landscapes for years and Im hardly alone:

http://thesmallermajority.com/2012/08/21/focus-stacking-of-live-subjects/

The reason I bought a macro lens was not to take photos of bugs, but to do macro beauty shots of makeup for makeup artist portfolios. You can't really focus stack a model's face or even whizzing bumblebees, so it's a solution with limited applicability.

Spoken like someone who's never actually tried.

proudgeek

You just need to adjust your methods. I've been doing free hand focus stacking of live bugs, flowers and landscapes for years and Im hardly alone:

http://thesmallermajority.com/2012/08/21/focus-stacking-of-live-subjects/

Very interesting. I'll be giving that a try. You learn something new every day.

Msmoto

@Ade

Finally a visual I can understand... :-*

rmp

Thanks @Ade for the visual-graph. It really helped. This was/is a very beneficial thread for me. Does anyone know of a reference or footnote for @Ade's graph?

Vipmediastar_JZ

So basically from the graph - at 2.8 the CA should no longer be and by F8 - f16 diffraction kicks in and by f22 it is the worst?

I think I understand it and thats why personally after hearing about diffraction on this site a year ago I have limited myselft to F8 unless I really need the F16-f22 for a longer exposure when I forgot the ND filter as an example.

Paperman

@Vipmediastar,
I have always thought the ND filter would cause more degradation than an f16, f22 ( but never had chance to compare. Probably a sure thing with the old Cokins I use. Wonder if anyone has done a test.

Vipmediastar_JZ

@paperman I had cokins in the past and it wasn't the best IMHO. there was some red tints with and possible degradition. I don't remember what was introduced but after I used the B+W I was more pleased with the results but I had also gained more knowledge in photography Post Cokin.
Here is one at f8

Maybe the foreground/lightpost is a bit degraded but I probably focused on the buildings most likely.

Ironheart

Remember diffraction comes on gradually. It's not like you should never ever use f/16. Just as you sometimes need to use a slightly higher ISO where noise starts to creep in to get the shot you want, you sometimes need to get into the diffraction zone to get the DoF you want. It's always a trade off.

That's why it's best to test your lenses and bodies to see how much diffraction you can stand, just like ISO it will be different for everyone. And also I should point out that @Ade's graph just an example, your sensor/lens combination will be different.

WestEndBoy

In an earlier post I stated that my 28 mm is diffraction limited at f/5.6. I had Ade's graph in my head when I said that.

I figure that my 20mm is diffraction limited around f/8.

You could say that these are the sweet spots.

And all other things being equal, the wider the sweet spot, the better the lens.

An important detail, depending on frame location (centre, edge etc.) this sweet spot will be different on any particular lens. Something to consider when edge sharpness is important.

SquamishPhoto

You can't really focus stack a model's face or even whizzing bumblebees, so it's a solution with limited applicability.

Spoken like someone who's never actually tried.

I tried to do multiple exposure shots on models to try and do HDR and it looked terrible. Mind you, that's pretty much automatic with the camera rapidly taking a number of photos at different exposure. Now you want to add messing with the focus ring to the mix? Good one.

You should have read it more closely. You focus through movement, not by racking focus manually.

Ade

@rmp

Zeiss has a great introduction to MTF and related factors in two parts: PART I and PART II.

You can find further discussion and a graph similar to what I presented above starting on Page 25 of Part II, plotting MTF of the Planar 1.4/85 ZA vs. f-stop and calculated diffraction limit (lines in red).

Reese

I have a theory, that for every two photographers there are at least 3 opinions on the same topic.

DaveyJ

I have a number of macro and close up video shots particularly with the D7100 often with the 70-300VR lens where bumblebee or Tiger Swallowtail Butterflies or Hummingbirds were pretty nicely photographed which just left to the default camera settings were impossible. These photos were NOT shot with my Micro Nikkor lens. I tend to try for about 800 ISO and at LEAST f11 or even f16 for stills. In video I have found the rules with the D7100 is 0.0 +/- settings and the camera picks the video settings the way I do it. The results are pretty sharp and on big screen HDTV very impressive. When I was shooting with 8x10 view camera I sometimes shot macro subjects at f64 and by that time employed as much swing and tilt as the subject needed. The results with the D7100 of close up moving subjects is actually QUITE impressive. I do try to use max f11 stop downs for more ordinary subjects. When I shoot photos I usually am shooting to maximize depth of field. Diffraction degradation does set in and often if possible I have shot some subjects that were less racy by shooting several shots and then weeding out the ones where diffraction or just failure to achieve sharp focus happened.

heartyfisher

Just got around to reading this thread .. interesting... is anything resolved?
I am still not certain of several point..
1) I would have thought that the amount of diffraction for a given aperture is different between lenses of different focal lengths. but it sounds like that is in contention.
2) even with lenses of identical focal and aperture but of different design surely diffraction will be different.
3) The smaller the sensor the larger the visual effect of diffraction.

Ironheart

1) since an f-stop is a fraction off the focal length (f/2 for a 50mm lens is 25mm, for a 100mm lens the f/2 opening would be 50mm) mathematically the actual focal length drops out and all that remains is the f-stop.

2) true

3) only because smaller sensors tend to have a greater pixel density/smaller pitch (a 16MP 1' sensor clearly has a smaller pich than a 16MP FX sensor). It's the smaller size of the pixels in relation to the airy disk that makes the diffraction more apparent.

I like these tutorials:
http://www.cambridgeincolour.com/tutorials/digital-camera-sensor-size.htm
http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm
And a quote from the first one:
Keep in mind that the onset of diffraction is gradual, so apertures slightly larger or smaller than the above diffraction limit will not all of a sudden look better or worse, respectively. Furthermore, the above is only a theoretical limit; actual results will also depend on lens characteristics.

Ade

#2 is an ambiguous question. For a given aperture, the diffraction limit is the same regardless of the lens design, so the proper answer is false. However, different lenses are able to reach the diffraction limit at different apertures (or sometimes not at all).

#3 depends on context. I'm going to nitpick @Ironheart's correct answer. Like DoF, the effects from diffraction depends on magnification. So other than when viewing at 100%, may become a factor not when the airy disk is large relative to the pixel pitch, but when the airy disk is large relative to the circle of confusion.

Paperman

@heartyfisher,
1) I would have thought that the amount of diffraction for a given aperture is different between lenses of different focal lengths. but it sounds like that is in contention.
2) even with lenses of identical focal and aperture but of different design surely diffraction will be different.
3) The smaller the sensor the larger the visual effect of diffraction.f

1) Definite no as Ade has explained. All that matters is f-stop
2) No again if we are talking purely about the effect of diffraction and NOT about combined effects of all other lens faults AND diffraction. Max resolution/IQ will be reached at different f-stops for each lens but it doesn't mean diffraction starts at different f-stops. It may not be visible - or let's say it may be hidden - due to other lens faults but that does not mean diffraction has not happened.(*)
3) Explained above.

No 2 can actually bring out another discussion - if all other lens aberrations & diffraction are cumulative or not ( diminished within each other ). For example, if you have a crappy lens capable of only 1500 lw/ph, would a f16 or f22 make it worse or won't have any effect as the IQ has already been pulled down to minimum by other faults .)