Wide aperture SHARP Narrow aperture NOT

[jaapv]

Advertisement (gone after registration)

If the need for a deep DOF overrides the distaste for image quality loss.

On another note, the larger the format the less these problems appear. Using a view camera, one can easily shoot at f 64 to get sufficient depth of field. One of the reasons Ansel Adams used these huge machines.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[ndjambrose]

Yes - that is normal.

 

Don't have a Summarit to compare, but I've never seen diffraction problems at f4 on any other lens. I'm generally comfortable with all of them until f8. For many of my lenses the conditions of greatest sharpness appears to be around f3.5-5.6.

 

I appreciate that diffraction as an optical effect begins at anything other than wide open by definition, but I've never found it was discernible by eye until stopped pretty far down. And certainly not at f4.

[jaapv]

In this pixelpeeping age with 100% crops or even more, it is quite possible to see it.

[ndjambrose]

In this pixelpeeping age with 100% crops or even more, it is quite possible to see it.

 

Doesn't entirely match my experience. It was because of pixel peeping that I figured out some of my lenses performed best around moderate apertures. Personally I'd be inclined to look for other causes in this specific case - such as miscalibration of sensor or front element, such that sharpness depends on a specific coincidence of light (like focus shift) -- or even variation in the testing process.

[jaapv]

It is lens dependent. For instance the 280-4.0 apo is fully diffraction limited, which means the best aperture is wide open, and any stopping down degrades the image. The Summillux 50 asph is iirc diffraction limited from f 2.0 onwards, etc. I could not say what the diffraction limit of the lenses the OP mentions is, but it may well be 4.0-5.6, as he noticed.

[zlatkob]

anything above 4 upwards and sharpness/details being to deteriorate

 

That doesn't sound normal. The M8 and M9 both have a pixel size of 6.8 microns. I don't know whether anyone has published the diffraction limited aperture (DLA) for the M8 or M9, but for Canon cameras with a similar pixel size (6.4 to 7.2), the DLA is calculated at about f/10 to f/11. This is is the point at which image sharpness begins to be compromised, and can be relevant for landscape or architectural photography. See the column titled "DLA" on the first chart at this link

 

In the small text under that chart, there is a link titled "diffraction comparison example" which leads to a test chart of the Canon 200mm f/2 lens. This lens superseded the Canon 200/1.8, which was the sharpest lens ever tested by Photodo.com (slightly sharper than even the 50/1.4 Summilux-M, the 280/4 APO-Telyt-R, etc.). The *mouseover* ISO 12233 Chart 100% Crops at this link are instructive as they show:

• extreme center sharpness from f/2 to f/8, with corner sharpness improving to f/8
  
• slight degradation at f/11
  
• significant degradation at f/16.
  

[jaapv]

Advertisement (gone after registration)

Lens diffraction is a lens property - that has nothing to do with sensor resolution and sensor DLA, which has to do with -as you say- pixel size and the size of the Airy disk thrown by the lens. There is even no difference between film and digital in lens diffraction.

[zlatkob]

Lens diffraction is a lens property - that has nothing to do with sensor resolution and sensor DLA. There is even no difference between film and digital in this respect.

 

Then why does the Diffraction Limited Aperture vary depending on the pixel size? It ranges from f/6.8 to f/12.7 on that chart, depending on the camera's pixel size.

[jaapv]

Diffraction Limited Photography: Pixel Size, Aperture and Airy Disks

[zlatkob]

Thank you for the link, but it seems to be consistent with my point above. Your link states:

 

"The size of the airy disk itself is only useful in the context of depth of field and pixel size."

 

... and ...

 

"As two examples, the Canon EOS 20D begins to show diffraction at around f/11, whereas the Canon PowerShot G6 (compact camera) begins to show its effects at only about f/4.0-5.6."

 

Before this thread, I had not heard that, due to diffraction, it is "normal" for a 50mm Leica lens to show a decline in sharpness when stopped down to f/4 and smaller. I would only expect to see it at f/11 and smaller.

[01af]

Anything above 4 upwards and sharpness/details being to deteriorate.

"Deteriorate"? Are you looking at pictures or at pixels? No-one sees any loss of sharpness due to diffraction in a picture taken with a 35-mm-format camera at up to, say, f/8 or f/11, and maybe the faintest hint of it at f/16 when looking very carefully.

 

Those who notice any losses beyond f/4 are the hard-core pixel peepers only. You should be glad when you're noticing a loss when going from f/4 to f/5.6 because that means a) your lens has exceptional quality and your technique is adequate to fully exploit that quality.

 

Apertures like f/11, f/16, or even f/22 still make sense because in images that depend on being sharp from front to back, more depth-of-field may outweigh the level of maximum sharpness at the plane of focus. If you need a lot of DOF but only have very little then your image will look basically blurred for the better part, no matter how excellent the sharpness may be in a tiny fraction of the image area.

 

So the general rule is, stop down as far as you need to but no further. And don't let diffraction loss considerations keep you from stopping down far enough for the DOF needed for your image composition at hand. There is a limit called the critical aperture at which diffraction blur is so bad that the COC criterion of acceptable sharpness is met nowhere across the whole frame ... however for 35-mm format this critical aperture is somewhere in the f/45 or f/64 range IIRC, so no need to worry about that.

 

 

Then why does the Diffraction Limited Aperture vary depending on the pixel size?

It doesn't. It's a myth that comes from not understanding how pixels (or grain) and Airy disks interfere. The pixel size is mostly irrelevant.

 

 

Before this thread, I had not heard that, due to diffraction, it is "normal" for a 50 mm Leica lens to show a decline in sharpness when stopped down to f/4 and smaller. I would only expect to see it at f/11 and smaller.

Well—I can clearly see a slight loss of sharpness in test images from f/4 on with my Summarit-M lenses when pixel-peeping carefully (haven't tried this with other lenses yet). At f/11, the losses are very obvious ... when pixel-peeping, that is. Pictures still look just fine.

[jaapv]

Before this thread, I had not heard that, due to diffraction, it is "normal" for a 50mm Leica lens to show a decline in sharpness when stopped down to f/4 and smaller. I would only expect to see it at f/11 and smaller.

Maybe if you glanced through Erwin Puts'"leica lens compendium" He gives the f-stop at which diffraction comes in for many lenses. It is often between 4.0 and 5.6. And never f 11.

I agree with 01af, however, that one should be aware of the phenomen, but never let it dictate picture taking. For one thing, the beginning of diffraction is indeed more easily seen on the test bench than on the final print, as it is often hidden by for instance camera shake or a slight misfocus, and for another the softening it may create at very small apertures is certainly offset by effects like a deep DOF. Sharpness is indeed a bourgeois concept.

[zlatkob]

OK, 01af, you disagree with Cambridgeincolour.com about the relevance of the pixel size. You state it's a myth and give no explanation. You also disagreed with Photozone.de about the validity of their MTF charts. You wrote they're not worthy of the electrons used to transmit them and gave no explanation.

 

Jaap, I followed your link to Cambridgeincolour and it supported my point. Now you want me to look at Ewin Puts.

 

If you choose to follow my link above, you will see that DLA for a camera with 6.8 micron pixels is about f/10 or f/11. You will also see, in ISO 12233 chart crops how one of the sharpest lenses on the planet shows visible degradation at f/11, not f/4. That doesn't seem consistent with your view that very sharp lenses show a decline at f/4.

 

The original post in this thread referred to the Summicron-M 50mm f/2.0. Photodo gives these weighted MTF scores for that lens:

f2 0.78

f2.8 0.81

f4 0.85

f8 0.87

Based on their tests, this lens should absolutely rock at f/8.

 

Unfortunately, the downloadable version of Erwin Puts' lens compendium omits diagrams. But for the 50mm Summicron, he writes (at p. 103):

 

"At 1:5.6 the overall contrast is slightly reduced, and the definition of the fine textural details in the field has improved a bit. Here we note that the user of these lenses should study his subjects and demands very carefully: for best overall contrast the optimum aperture is 4, but for best definition of very subtle textural shades of grey or colour, 5.6 might be more appropriate. Scientific tests can indicate these differences as measured values, but the user may or may not be able to see or appreciate them."

 

If according to Erwin Puts, the authority you cite, the very best apertures for the 50mm Summicron are f/4 (contrast) and f/5.6 (detail), why is it "normal" for the OP to actually see a decline in sharpness starting at f/4? Again, steinzeug wrote:

 

"however when i up the aperture to 4, 5.6 or higher, i get a deeper depth of field (of course) but, the area in focus is nowhere near as sharp as at 2, 2,5."

[jaapv]

:confused:because the Summicron is not a Summarit maybe? The Summicron is an older design, the Summarit a newer one. It is more than likely that the Summarit is better corrected for aberrations. That means that the diffraction will not become visible past 5.6 ( for the Summicron) as Puts implies in your quote, which incidentally also denies the rocking at 8.0 - but quite possibly at 4.0 (for the Summarit), as the OP mentions. And both are very far from the f 11 to f 16 you have been insisting on all the time. And the deterioration, as has been said by various posters, is not dramatic, we are talking about pixel peeping is test conditions, not real life. You are still confusing diffraction on the sensor with lens diffraction, which is a purely optical phenomen which takes place at the edge of the aperture blades and has nothing to do with pixel size, as long as the sensor is able to resolve the Airy Disk and which is absent when we extend the argument to film, which has no pixel size at all.

[zlatkob]

Jaap, I've followed your links and responded to each. You haven't responded to any of mine. So maybe I should stop linking to authorities, tests and examples, and just state unsupported opinions like everyone else.

 

Now you have a new theory: the Summarit possibly shows diffraction at f/4 ... possibly. OK, possibly it does. But I don't think that's enough to reassure the OP that what he is seeing is normal. The OP states he has the same problem with the 50mm Summarit and Summicron: worse at f/4 and f/5.6 than at f/2 and f/2.5. Those results are inconsistent with Erwin Puts, your authority on the 50mm Summicron, and inconsistent with Photodo. And yet you state they are "normal".

 

I am inclined to agree with ndjambrose that there is likely some other problem at work.

[01af]

OK, 01af, you disagree with Cambridge in Colour - Photography Tutorials & Learning Community about the relevance of the pixel size. You state it's a myth ...

I definitely do, because I understand how pixels and Airy disks interact to create sharpness (or lack thereof), and Sean McHugh doesn't. The explanation can be found in any junior highschool physics textbook ... or in your photographs if you dare to believe your own eyes.

 

The basics are very simple. When two resolution-limited components form a chain where the output of the first component (lens) is the input to the second (sensor or film) then contrary to common belief the system's overall resolution limit is not simply the minimum of the two components' limits. Let R1 be the maximum resolution of the first component. Let R2 be the maximum resolution of the second component. Then the system's total resolution R is 1 / (1/R1 + 1/R2). Or: 1/R = 1/R1 + 1/R2. Doesn't this formula look familiar? It should, because you'll find formulas of this shape in many contexts.

 

Anyway, one of the consequences of this simple fact is this: When you have a lens A that creates blur disks smaller than the pixel pitch and you have another lens B that is even better than lens A then the image from lens B on the sensor will be sharper than the image from lens A on the same sensor, even though lens A—the weaker lens—already "outresolves" the sensor. In other words, the optimal aperture (i. e. the aperture where increasing diffraction effects and decreasing lens aberrations balance each other out; i. e. the sweet spot of the lens) will be the same on every sensor. The pixel pitch is irrelevant for this. If a lens is best at, say, f/4 then it will be best at f/4 on every sensor (and on every film, too, of course).

 

 

I followed your link to Cambridgeincolour and it supported my point.

Sure it does. But they are wrong. You know, acquiring knowledge is not deciding who to believe. Neither is it blindly following the majority. Instead, it involves learning and understanding, and often you'll end up in the minority.

 

 

If you choose to follow my link above, you will see that diffraction-limited aperture for a camera with 6.8 micron pixels is about f/10 or f/11.

This statement is just nonsense. Camera don't have apertures; lenses have. If we follow your link then we'll see that at approx. f/11, the Airy disk has a diameter of approx. 6.8 µm. That's all. It does not tell us at which aperture any given lens will start to exhibit diffraction loss on a 6.8 µm sensor ... or any sensor. Because that has nothing to do with the pixel pitch but with lens' quality.

 

If you care to follow your own link then you will easily see that the mouseover ISO 12233 Chart 100 % Crops support my point, not yours. Choose a good lens from their drop-down list and find out at which aperture sharpness is starting to suffer from diffraction. Then re-do that with the same lens on a camera with a different pixel pitch, and you'll find that the limiting aperture will be just the same on the other sensor. Just as theory and common sense and practical experience predict.

 

 

That doesn't seem consistent with your view that very sharp lenses show a decline at f/4.

That's because the data presented at the-digital-picture.com aren't good enough ... or maybe they don't test really good lenses. I don't see any Leica lenses there. And those silly test charts shot at rather short distances don't make good test targets anyway. Why don't you create your own test shots and see for yourself rather than looking at irreliable third-hand web data?

[01af]

The OP states he has the same problem with the 50 mm Summarit and Summicron: worse at f/4 and f/5.6 than at f/2 and f/2.5.

I guess this particular problem has nothing to do with diffraction but with a defect of the lenses, the camera, or some kind of user error. But I'm just speculating here ... just as anyone else does as long as the OP doesn't show a few of his pictures.

[zlatkob]

01af, thanks for your detailed reply. I'm running out to a meeting, but will review your comments when I have a little more time. I agree, we are all speculating about the OP's issue.

[Advertisement]

Hallo,

Look here: Leica M

[Guest mc_k]

I definitely do, because I understand how pixels and Airy disks interact to create sharpness (or lack thereof), and Sean McHugh doesn't. The explanation can be found in any junior highschool physics textbook ... the system's total resolution R is 1 / (1/R1 + 1/R2). Or: 1/R = 1/R1 + 1/R2. Doesn't this formula look familiar? It should, because you'll find formulas of this shape in many contexts....

 

not sure how you got from Airy disks to the formula above, which I understand as a rule of thumb. Care to give some details?...

[01af]

Not sure how you got from Airy disks to the formula above ...

The larger the Airy disks, the lower the resolution. So the resolution basically is reciprocal to the Airy disk diameter.

 

 

... which I understand as a rule of thumb.

Basically it is a rule of thumb indeed. That's all you need in order to understand the underlying principle. If you want to go into details then things will get complicated because it's not so easy to describe lens resolution and sensor resolution by the same terms. But it's the principle that counts.