About diffraction and its real impact

[nongfuspring]

Interesting posts. I've just accepted the fact but never really thought about it too hard since I'm usually much more concerned with mitigating camera shake than diffraction and when I do use f16 it's because I've maxed out my shutter and don't have a choice anyway. I've had a lot of photos ruined by unwelcome blur but none because of having too small an aperture.

Out of curiosity, does the significance of diffraction change relative to focal length at all or is diffraction only about the relationship between aperture and sensor/film size?

 

[ColSebastianMoran]

I learned a simple formula: Diffraction limited MTF50 in lp/mm is 800/f, where f is the f-stop. That is, at f/8, your MTF50 is limited to 100 cycles/mm. Works for all formats.

So what? On film, it's hard to record 100 cycles/mm. Or looked at the other way, 100 cycles/mm on 35mm film or sensor would be 35MPx. f/8 is going to be fine for 35mm format.

Each stop reduces linear resolution by 1.4x and MPx by 2x. So, on 35mm, f/11 would be limited to 18MPx, f/16 to 9MPx and f/22 to 4MPx.

Now, let's look at APS-C: f/8 = 17MPx, f/11 = 8MPx, f/16 = 4MPx. f/22 is 36.4 lp/mm which is 2Mpx. That's a loss that will be visible. Whatever the effect of diffraction, it happens at one stop wider on APS-C than on 35mm.

LF might be shot at f/64 with MTF50 at 12.5 cycles/mm. For 4x5, this is roughly 2450x3000 px or about 6MPx. Has produced a lot of nice photographs, but resolution would be higher at wider apertures.

Caveats: MTF50 isn't everything, digital sensors have different resolution characteristics than film. Lens aberrations, DOF, and camera shake also affect resolution.

 

[Roger Hicks]

I learned a simple formula: Diffraction limited MTF50 in lp/mm is 800/f, where f is the f-stop. That is, at f/8, your MTF50 is limited to 100 cycles/mm. Works for all formats. . .

Quite. But see post 24. Too simple! As, in all fairness, you point out later in your post. This is why I discourage people from relying too much on this "simple formula", or any variation thereon: I first learned a variation of it from a physics Ph.D. student, who likewise pointed out its limitations and the assumptions inherent in it.

Cheers,

R.

 

[tunalegs]

As I understand it diffraction "kicks in" on all lenses regardless of focal length or actual aperture size at the same f-stop. Around f/8-f/11.

How noticeable the effect of diffraction is depends on the size of the image. Diffraction caused by using f/22 on a 35mm camera will be more noticeable than the effect of diffraction caused by using f/22 on an 8x10" camera - given that prints of equal size are made from both.

Of course given that sort of test all other sorts of defects will appear larger from the 35mm neg - literally magnified by enlargement.

 

[michaelwj]

It is said that the sharper the lens, the faster diffraction kicks in. Zeiss were proud to mention that their lenses are affected by diffraction as early as f/5.6.

Diffraction is always present. What diffraction does is spread out a point source into a disk, which becomes visible when the disk imposes itself on neighbouring pixels. To say that a sharper lens has less diffraction is a load of bull, and only the half the story. What is true is that with a very high resolution sensor capable of resolving all the resolution provided by the lens, diffraction will start to rob the sharpness earlier.

Put another way; when the size of the disk caused by diffraction is larger than both the disk caused by the optic anyway AND larger than the imaging pixels, then diffraction becomes noticeable for the point source.

As I understand it diffraction "kicks in" on all lenses regardless of focal length or actual aperture size at the same f-stop. Around f/8-f/11.

In a way, yes, the f/stop is the focal length divided by the size of the hole, so it takes into account magnification. This is why the aperture scale is much more useful than a physical measurement.

As an aside - I did my PhD in optical physics, and currently work on lensless diffraction imaging systems - and I still find it so interesting how it applies to photography. In the end, I bow to aesthetics, I do what I do to get the look I want in the image without getting bogged down in maths and diffraction and the like.

Cheers,
Michael

 

[EdwardKaraa]

Diffraction is always present. What diffraction does is spread out a point source into a disk, which becomes visible when the disk imposes itself on neighbouring pixels. To say that a sharper lens has less diffraction is a load of bull, and only the half the story. What is true is that with a very high resolution sensor capable of resolving all the resolution provided by the lens, diffraction will start to rob the sharpness earlier.



Cheers,
Michael

But but but, that is absolutely not what I said. Why do you say it is a load of Bull?

What I said, and Zeiss said, and I'm pretty sure even with your PhD that Zeiss knows better than you, that their lenses reach their peak performance by f/4 so that you can start seeing diffraction effects already by f/5.6. I said good sharper lenses get diffraction at wider apertures, earlier than less good lenses that reach peak performance at f/8 or even f/11. I never said sharper lenses have less diffraction. Please read more attentively.

PS. There is a white paper by Dr. Nasse on Zeiss website that describes this phenomenon. It shouldn't be difficult to find.

 

[michaelwj]

But but but, that is absolutely not what I said. Why do you say it is a load of Bull?

What I said, and Zeiss said, and I'm pretty sure even with your PhD that Zeiss knows better than you, that their lenses reach their peak performance by f/4 so that you can start seeing diffraction effects already by f/5.6. I said good sharper lenses get diffraction at wider apertures, earlier than less good lenses that reach peak performance at f/8 or even f/11. I never said sharper lenses have less diffraction. Please read more attentively.

PS. There is a white paper by Dr. Nasse on Zeiss website that describes this phenomenon. It shouldn't be difficult to find.

Sorry, I didn't mean to offend.

I reread my post, what I meant to say was "To say that a sharper lens has more diffraction is a load of bull". Sharpness has no effect on the amount of diffraction present. It's physically a property of the size of the opening and the wavelength of light. Not even the mighty Zeiss can change that, although their marketing people will have you believe that they can. They're trying to sell you something after all.

What I am saying is that diffraction is present at f/1, and f/4, and f/5.6, BUT, you will only see it when your sampling interval in the imaging plane is fine enough (and a bigger apertures you need to sample finer). IF you have a fine enough sampling interval to see diffraction at f/2.8 THEN you will also need to have a lens that can resolve BETTER than that to have any detrimental effect. This is what the marketing team twist into "Our lenses are so good that you can start to see diffraction effects by f/5.6".

Cheers,
Michael

 

[EdwardKaraa]

Sorry, I didn't mean to offend.

I reread my post, what I meant to say was "To say that a sharper lens has more diffraction is a load of bull". Sharpness has no effect on the amount of diffraction present. It's physically a property of the size of the opening and the wavelength of light. Not even the mighty Zeiss can change that, although their marketing people will have you believe that they can. They're trying to sell you something after all.

What I am saying is that diffraction is present at f/1, and f/4, and f/5.6, BUT, you will only see it when your sampling interval in the imaging plane is fine enough (and a bigger apertures you need to sample finer). IF you have a fine enough sampling interval to see diffraction at f/2.8 THEN you will also need to have a lens that can resolve BETTER than that to have any detrimental effect. This is what the marketing team twist into "Our lenses are so good that you can start to see diffraction effects by f/5.6".

Cheers,
Michael

Again Michael, you're reading wrongly. I didn't say (neither did Zeiss) that a sharper lens has more diffraction. I said very clearly that the effects of diffraction are visible earlier with lenses that reach their peak sharpness at wide apertures.

Your explanation is correct though, and in line with Dr. Nasse's white paper. It does not contradict what Zeiss marketing team says that their lenses are so good you start to see diffraction effects by f/5.6. This is what happens in practical terms, what the end user sees, and they are confirmed by the measured MTF charts provided by Zeiss and Leica.

 

[pgk]

FWIW I have carried out some very empirical testing with a very specific set-up, namely a Canon 5D2 with a 100mm macro lens when used underwater to identify and capture the details required to confirm the identification of British Gobies (small marine fish) - I know its a bit weird but someone had to do it😉.

I found that maximum resolution was between f/11 and f/16, below f/11 meant that depth of field was marginally less than desirable (I hesitate to say insufficient) whilst f/16 showed a small loss of resolution (not much but just about there) and the finest details wanted were less easy to be sure of. That said we are talking nuances here.

In purely practical terms (photography is practical isn't it?) this gave me the best compromise in terms of depth of field and resolution so diffraction of full frame 35mm is a factor at f/16 if that is, you really require absolute resolution in your images - something which is all too often considered an essential pre-requisite when in reality its the subject matter and its treatment which is the far more important factor.....

 

[teddy]

"So, what matters with diffraction is the physical size of the aperture that the light is passing through. This is why Ansel Adams (who shot with large format plate cameras) could often use apertures as small as f64. He did it because it maximized depth of field. He could get away with it because on a large format camera f64 is physically a lot bigger than it would be on say a 35mm camera and hence there is less diffraction even with an aperture that is nominally f64."

Peter, your explanation is one of the best I have ever come across. Thank you!

 

[mdarnton]

Like pgk, I do two jobs regularly that are basically archival cataloging, with a micro-Nikkor and D300. On both I can definitively see image degradation starting at f7.1. One job likes an amount of DOF and I can go to f11 and gain as much as I lose. This work is usually viewed on a screen and often at 100%, looking at and comparing details and no one would complain if I bought a D810 to do it. The other is flat work and I stay at f8 for that. Both them lean on the kind of micro-examination where detail is needed and appreciated.

The difference due to diffraction is real and observable in those non normal world contexts.

However, in normal photography I go for the DOF I need. Who in real life, not photo forums, looks for and counts ants on the sidewalk in the corner of the image? There's probably a difference but what normal viewer notices? My 10-20 Sigma is awful in the corners wide open but no one has ever commented on that aspect of the pix I take with it.

 

[ColSebastianMoran]

As I understand it diffraction "kicks in" on all lenses regardless of focal length or actual aperture size at the same f-stop. Around f/8-f/11.

Well, maybe, and probably a useful thought for 35mm format given sensor/film limitations.

But, it's really different for different formats (small sensor, APS, 35mm, 6x6, Large Format).

 

[tunalegs]

"So, what matters with diffraction is the physical size of the aperture that the light is passing through. This is why Ansel Adams (who shot with large format plate cameras) could often use apertures as small as f64. He did it because it maximized depth of field. He could get away with it because on a large format camera f64 is physically a lot bigger than it would be on say a 35mm camera and hence there is less diffraction even with an aperture that is nominally f64."

Peter, your explanation is one of the best I have ever come across. Thank you!

But not technically correct. If we were talking about an aperture, alone, without any lenses about it - this would be true. But because the lenses change the angle at which light passes through the aperture - diffusion produces basically the same effect at the same f/stop regardless of focal length or the actual physical diameter of the aperture.

The simplified version is as I wrote above that diffusion is more noticeable on small formats than on large formats. It is of course more complicated than that and ties into things like how lenses of identical focal length and speed will produce a difference in depth of field dependent on the format they are being used on. But as I cannot adequately explain that phenomenon I won't try to. 😀

Well, maybe, and probably a useful thought for 35mm format given sensor/film limitations.

But, it's really different for different formats (small sensor, APS, 35mm, 6x6, Large Format).

Which I did note in the part of my post you didn't quote.

 

[michaelwj]

It is said that the sharper the lens, the faster diffraction kicks in.

Again Michael, you're reading wrongly. I didn't say (neither did Zeiss) that a sharper lens has more diffraction. I said very clearly that the effects of diffraction are visible earlier with lenses that reach their peak sharpness at wide apertures.

I obviously misinterpreted your first comment. At large apertures there is so little diffraction that we'd need an unbelievable amount of resolution to see it. Anyway, I think we are on the same page in the end. Also, do you have the link to this paper you talk about, sounds interesting and I couldn't see it on the Zeiss website?

However, this is all academic since we all seem to do whatever works for a given photo.

Cheers,
Michael

 

[EdwardKaraa]

I obviously misinterpreted your first comment. At large apertures there is so little diffraction that we'd need an unbelievable amount of resolution to see it. Anyway, I think we are on the same page in the end. Also, do you have the link to this paper you talk about, sounds interesting and I couldn't see it on the Zeiss website?

However, this is all academic since we all seem to do whatever works for a given photo.

Cheers,
Michael

Hi Michael,

Indeed, we are both saying the exact same thing, while focusing each on different aspects.

As for the white paper, Dr. Nasse wrote several for Zeiss, but for some reason there is no direct link on Zeiss webpage. There is one about bokeh, three about reading MTF, and several about different Zeiss designs: Planar, Sonnar, Biogon, Distagon... etc. I am not sure in which he talks about diffraction, but I think it should be in the MTF paper. The links can be easily found via google.

Cheers,
Edward