Lens diffraction - truth or urban legend?

kossi008 said:
Oh wow. Lemme borrow your diffraction-limited 35/1.4... 😀
Click to expand...

Oh, snap!

The new Summilux ASPH FLE delivers an MTF-50 of slightly better than 40 lp/mm on center, and of course it gets worse as you move toward the corners. We'll be generous to Father L, and call it 50 lp/mm, wide open.

For those who were wondering: 50 < 1,100.

Shockingly, this $5,000 lens performs considerably better when stopped down from f/1.4 to f/5.6. Apparently, those who purchase it really should consider upgrading.

🙄
 
The older 50/1.4 was tested to ~75% contrast at 50LP/mm, wide open at F1.4. This test is from 1976. The "dot" is the measurement for the particular lens, the Bar represents the range of all lenses tested of this type. The Summilux had the best score used at F1.4 for center and 1/3rd out.

picture.php


50LP/mm works out to 100 pixels per millimeter, assuming no AA filter or color mosaic filter. So the KAF-1600 in my 1992 DCS200ir with its 9micron element size would be a good match for the Summilux. I'm not lending out the DCS200ir. I've used it to test some lenses as it does not have the mosaic filter or AA filter.

I'm sure that I have some diffraction limited optics left over in the Lab. It was expensive, and used in an optical computer built in the 80s.
 
Last edited:
Sparrow said:
OK, both at 100 asa, ignore the motion blur Alice interpreted "keep still" as "burst out laughing" [...]
click the pic for the original image, what do you see?
Click to expand...

What did you focus on? Of course you can compare things like these only within the plane of focus, otherwise the f/22 version's greater depth of field will make things look less blurry that are slightly out of focus at f/5.6.

Did you upload the original images to Flickr at different resolutions? The f22 one is 2048x1536 pixels, the f5.6 one is 1024x768. This makes it really hard to compare things. The f/22 one, for example, looks quite blurry at full resolution when you look at the jeans, and it would be nice to compare it 1:1.
 
semilog said:
For those who were wondering: 50 < 1,100.
Click to expand...

Exactly the point I was trying to make with my witty remark...

Oh yeah, and I used to work with an f/0.37 Zeiss lens that was close to the diffraction-limit at open aperture. Resolutions of 11,000 lp/mm were no problem for that baby.

The drawbacks were:
(a) stopping down was not possible
(b) focussing was not possible, fixed magnification of 1:4
(c) its weight was about 1,200 kg
(d) it's price (which I don't even know)
(e) it was corrected for a wavelength of 193 nm

Yes, I'm talking about photolithography... 😀
 
Last edited:
Sparrow said:
I'm digitally challenged so I just scale focused the lens, the DOF scale even at 5.6 easily covered the scene. The camera was on a substantial tripod, but the subject was giggling in the f22 shot so there is likely motion blur on her
Click to expand...

I see - I wonder what Ansel Adams had to say about giggling subjects. 😉 For the future I would probably prefer to do this sort of test with a precisely-focused lens without relying on the DOF scale, especially on a crop camera.

Anyway I have my doubts whether you'd see much impact in a 3 megapixel shot at all. If you print that, you get something like a 5x7" print, probably not enough to judge the effect of diffraction in the first place.
 
rxmd said:
I see - I wonder what Ansel Adams had to say about giggling subjects. 😉 For the future I would probably prefer to do this sort of test with a precisely-focused lens without relying on the DOF scale, especially on a crop camera.

Anyway I have my doubts whether you'd see much impact in a 3 megapixel shot at all. If you print that, you get something like a 5x7" print, probably not enough to judge the effect of diffraction in the first place.
Click to expand...

It would have been focused at 3m/10ft and was unchanged between shots.

I'll look to see if the RAW files are a higher resolution or repeat the exercise with a larger jpeg, the gf1 should have twice that resolution
 
Sparrow said:
It would have been focused at 3m/10ft and was unchanged between shots.

I'll look to see if the RAW files are a higher resolution or repeat the exercise with a larger jpeg, the gf1 should have twice that resolution
Click to expand...

Don't worry too much.... It's not like we have to prove that diffraction exists - nor that it's relevant for photography, for that matter; talk to any macro photographer who ever tried to get decent DOF into his scenes.
 
I'm not concerned at all, I just think it's important to get this type of thing in proportion, personally I've never seen diffraction in a photo, in this thread and despite some peoples claims we haven't seen a single example.

For the sake of any newcomers to photography I think it's important to debunk some of these yeti issues that seem to get blown out of all proportion here ...
 
Sparrow said:
For the sake of any newcomers to photography I think it's important to debunk some of these yeti issues that seem to get blown out of all proportion here ...
Click to expand...

Agreed Stewart. Not only was the issue blown out of proportion, but the advice to use f5.6 or below due to diffraction limitation is simply, technically wrong for many lenses.

In all lenses, aberrations and diffraction interact such that peak performance is reached at a certain aperture. While, for practical use, the peak aperture is rarely relevant, it can be found via MTF charts or testing.

An example: my pre-asph Summicron 35/2. Here are my typical test shots to check center and corner performance (focus is on antenna). I do these kind of tests whenever I get a new lens.

884077736_RMcBw-S.jpg


884077910_PzuSJ-S.jpg


Here is an assembly of crops at different apertures:

884079392_aFv3f-X2.jpg


Clearly, in the center, the lens performs highest at f11. In the corners it's already pretty good at f4, peaks at f8 and is still very good at f11. Showing that there is field curvature, among others, dominating performance. Note that, as a user, I like that behavior, because off-center infinity performance is more important than center performance, for the typical landscape photo.

For the newbie: use your lens. If you like the photos it takes, that's all that matters. If you want to check performance, read MTF charts or do your own tests and look at the results in their entirety. Testing is quite simple. All you need is a tripod.

Roland.
 
Last edited:
Sparrow said:
I'm not concerned at all, I just think it's important to get this type of thing in proportion, personally I've never seen diffraction in a photo, in this thread and despite some peoples claims we haven't seen a single example.

For the sake of any newcomers to photography I think it's important to debunk some of these yeti issues that seem to get blown out of all proportion here ...
Click to expand...

Its not myth at all.

The thing is that any system is only as good as its weakest link. When using film and wet printing, the weakest link is usually the enalrger alignemnt and the enlarger lens aperture. The latter because it can only magnify what it has been given from the camera lens. So if the taking stage was slightly off and the enlarger lens is slightly off, it compounds the situation. And if the enlarger alignment is off things get even worse except the bit where you place your grain magnifier to focus.
So for film users printing taditionally it is an important consideration for ultimate print quality.

But for digital users where there is no optical enlargement, you have completely removed the possibility (probability) of resolution/sharpness loss through that stage. That means diffraction will not influence final output as much as it can (probably) with wet printing from film.

So if there is any debunking to be done, it should be that diffraction IS significant especially when using film to traditional wet print.
 
tlitody said:
Its not myth at all.

The thing is that any system is only as good as its weakest link. When using film and wet printing, the weakest link is usually the enalrger alignemnt and the enlarger lens aperture. The latter because it can only magnify what it has been given from the camera lens. So if the taking stage was slightly off and the enlarger lens is slightly off, it compounds the situation. And if the enlarger alignment is off things get even worse except the bit where you place your grain magnifier to focus.
So for film users printing taditionally it is an important consideration for ultimate print quality.

But for digital users where there is no optical enlargement, you have completely removed the possibility (probability) of resolution/sharpness loss through that stage. That means diffraction will not influence final output as much as it can (probably) with wet printing from film.

So if there is any debunking to be done, it should be that diffraction IS significant especially when using film to traditional wet print.
Click to expand...

... and of course you will be supporting those assertions with photos no doubt?
 
Sparrow said:
... and of course you will be supporting those assertions with photos no doubt?
Click to expand...

I don't know if photos would support that logic or not. He's right -- in optical photography there are several sources of error that can cause unsharp pictures.

It is indeed a fact that a poorly calibrated enlarger will lead to a print that is not sharp... even if the negative is perfectly free of diffraction-induced fuzziness.

But so will a poorly cailbrated scanner.

I struggle to find why that logic applies to the diffraction discussion, but I also struggle with the notion that photos will prove/disprove anything related to that arguement. 🙂
 
p.s. From Rolands photos I would conclude that the physics of diffraction is mistaken -- diffraction actually happens at larger apertures. When the thesis was typed, the typist appears to have exchanged the terms "small aperture number" with "small aperture" in the sentence, "diffraction effects in a photographic lens occur at a ... " 🙂
 
Fuji Superia 400, 12mm Hellier, so probably 1/500 at f16'ish. That would be an iris aperture of just 0.75mm well into the danger zone, a rough and ready 6mp draft lab scan

click for 6mp version
 
ferider said:
Agreed Stewart. Not only was the issue blown out of proportion, but the advice to use f5.6 or below due to diffraction limitation is simply, technically wrong for many lenses.
Click to expand...

Correct. It depends on the lens, and on the sensor. In any case, f/11 is not usually problematic. In FF 35mm, f/22 and f/32 are where you'd more typically start to see perceptible degradation of IQ. Again, here's an example of an absolutely superb lens, that is nearly at its optimum from the f/2 max. aperture.

In all lenses, aberrations and diffraction interact such that peak performance is reached at a certain aperture. While, for practical use, the peak aperture is rarely relevant, it can be found via MTF charts or testing.
Click to expand...

Depends on the lens, sensor, and application. Generally it's enough to know that peak performance for most lenses is at middle to middle-wide apertures, and to know that you won't notice unless you're using a dense sensor (or fine-grained film) and a tripod (or a very fast shutter speed or, equivalently, strobe).

Clearly, in the center, the lens performs highest at f11. In the corners it's already pretty good at f4, peaks at f8 and is still very good at f11. Showing that there is field curvature, among others, dominating performance. Note that, as a user, I like that behavior, because off-center infinity performance is more important than center performance, for the typical landscape photo.
Click to expand...

Bingo. Also note that you have not shown focus-bracketed images, and that — especially at the wider apertures — focus errors are extremely common and can have (relatively speaking) large effects on practical lens performance.

Because of field curvature and focus errors (due to the user or due to miscalibration of the focus system or due to film curvature*), peak performance of the camera-lens system is often at a smaller aperture than peak performance of the lens alone — the increased DOF at smaller apertures compensates for these problems.

*This is where camera quality comes in to play, as pointed out in a recent article by Ctein at TOP.

Most systems will deliver much more than acceptable quality at f/4 or f/5.6 to f/11.
 
Last edited:
Gumby said:
I don't know if photos would support that logic or not. He's right -- in optical photography there are several sources of error that can cause unsharp pictures.

It is indeed a fact that a poorly calibrated enlarger will lead to a print that is not sharp... even if the negative is perfectly free of diffraction-induced fuzziness.

But so will a poorly cailbrated scanner.

I struggle to find why that logic applies to the diffraction discussion, but I also struggle with the notion that photos will prove/disprove anything related to that arguement. 🙂
Click to expand...

Well yes that's the point, diffraction is such a minor effect it cannot be considered the "weakest link" and it would be a waste of time for anyone new to photography to give it any thought at all ... which is pretty much the answer to the OP's question would you agree? 🙂
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom