M9 diffraction limits

[jaapv]

Advertisement (gone after registration)

That is due to the tradeoff between aberrations and diffraction. Some lenses, like the apo-telyt R 280/4.0 are so well corrected that they are diffraction-limited wide open.

[okram]

Dear Okram,

 

Diffraction only depends on the aperture and the wavelength of the light, not on the focal length, and not on the format.

 

It is less visible on larger formats due to the lower magnification necessary to produce a print of a given size.

 

Best regards

 

Trond

 

 

Dear Trond, first is true only if we are not photographers.

 

Larger format means longer lenses, therefore larger diameter of aperture, therefore practically less diffraction. Of course magn. factor is more important.

 

 

 

Michael, with all due respect,and this can be a long discussion- lets make it simple: the smaller the hole- more diffraction. The hole marked as f is relative to focal length.

[ianspector]

Dear All,

 

This is a most interesting set of responses. One of the things that I learned from the linked article was

"Technical Note:

Since the physical size of the lens aperture is larger for telephoto lenses (f/22 is a larger aperture at 200 mm than at 50 mm), why doesn't the size of the airy disk vary with focal length? This is because the distance to the focal plane also increases with focal length, and so the airy disk diverges more over this greater distance. As a result, the two effects of physical aperture size and focal length cancel out. Therefore the size of the airy disk only depends on the f-stop, which describes both focal length and aperture size. The term used to universally describe the lens opening is the "numerical aperture" (inverse of twice the f-stop). There is some variation between lenses though, but this is mostly due more to the different design and distance between the focal plane and "entrance pupil.""

 

But does this allow for different lens construction methods, as for example a 50 Summarit is much shorter than my Noctilux.

 

I really appreciate your responses and the quality of the discussion. I personally find it more engaging than the 'what 6 lenses should I take on a first date' threads (although I do rad those as well).

Ian

[Guest trond]

Dear Ian,

 

You state this very well!

 

I agree, the physics behind diffraction is clearly understood and indisputable.

 

No one can change it, not photographers, not engineers, nor Leica lens designers.

 

I have nothing more to add.

 

Best regards

 

Trond

[okram]

Ian thank You for this explanation, it makes thing more clear for me now.

On the other hand it makes things possibly more complicated- since Airy disc exists at all apertures, that would mean that with telephoto lens it would be initially bigger. That would also mean that, for close focus, diffraction limit would kick in sooner...

 

Also- focal distance is a distance from lens nodal point, an that point can be almost anywhere, for different types of lens construction.

 

There is also that diffraction is different for red and violet on the other side of the spectrum.

Would this mean that diff. is bigger if the WB too red?

 

The exact answer for all this is too complicated for me to be used in practice.

 

I might just add that, according to Sean's calculator, I can use f22 for any lens with my 5D mk2 with no diffraction. This is just not true.

[ianspector]

Dear Okram,

 

Perhaps I am using the calculator wrong but when I try it suggests that f/8 is the smallest aperture without diffraction. I put in 21.1MP and 35mm as the sensor size.

 

Certainly violet will be less susceptible to diffraction at a given aperture than red, and also the article reminded me that the Bayer pattern uses twice as many green/luminance pixels as red and blue so it will be worse in green.

 

I feel confident even if I forget most of that which I have learned here; I will remember that on my M9 I am pretty safe from diffraction upto and including f/8. Thus with Pull /80 and 1/4000 and f/8 it will probably be only the nuclear holocaust that I am trying to capture that will stump me. By then, I guess, diffraction will be the least of my worries.

 

Ian

[xtoph]

Advertisement (gone after registration)

Shooting at 1000ft from Buckinghamshire to Dorset with the roof open I very soon realised that I was going to move the Noct off its normal f/1 setting. [...]

I can only assume that given the smaller lenses diffraction is the culprit here. What are the accepted Leica lens limits before diffraction, and is it the same for all leica lenses?

 

I guess an ND filter is in my future.

Ian

 

i am very surprised that you wanted such dark exposures. that's about three stops below iso standard exposure for a sunny day; allow an extra stop for altitude (which i doubt is that much of a factor at 1000 feet) and you're still several stops underexposed. you can see this pretty clearly in the landscapes. direct reflections off silver roofs and water are never going to be within dynamic range; you might as well try to expose not to clip the disk of the sun. your best option for that sort of scene is probably a polarizing filter (i shudder to contemplate trying to operate it on a rangefinder while flying a small plane though...), or try to make all those reflections the theme of the photo.

 

there have already been good posts about diffraction. it's worth emphasizing the difference between aperture and f-stop in these discussions, since usually we casually refer to them interchangeably; the absolute aperture size, not the f ratio, is the important thing for diffraction.

 

but airy disks aside, my practical experiments have shown that the diffraction limits which are often calculated mathematically don't always play out in practice. between the rather clever demosaicing algorithms and sharpening, i can often pull detail out of shots made at f/16 that wasn't supposed to be there. and of course the final output size/print should be kept in mind; if a particular reproduction is diffraction-limited when using a 12mp sensor, it will be no more worse for the wear if it had come from a 21mp sensor. with higher resolution sensors we have potentially more to lose, but a lot of that potential is very rarely fully exploited anyway; i would expect the vibration from that plane to rob you of more of the theoretical maximum resolution than diffraction any day of the week.

 

so unless one is using a steady tripod and maximizing resolution in other ways, then in practical terms i generally find diffraction to be the least of my worries on this format. if for some reason i need the extra dof from f/16 or even 22, i rarely hesitate on account of diffraction to go there.

[mjh]

On the other hand it makes things possibly more complicated- since Airy disc exists at all apertures, that would mean that with telephoto lens it would be initially bigger. That would also mean that, for close focus, diffraction limit would kick in sooner...

No, not at all. The only variable factors in the (approximate) formula for the Airy disk size are the wavelength (usually set to 555 nm or in other words green) and f/D where D is the diameter of the aperture. But since f/D is the relative aperture, the only factor is the relative aperture. At a given aperture value, the size of the Airy disc is constant, regardless of the focal length of the lens.

 

Larger format means longer lenses, therefore larger diameter of aperture, therefore practically less diffraction.

Again, the relevant factor is f/D. A larger format means longer lenses (f increases) and a larger diameter of the aperture (D increases). So in the fraction f/D, both numerator and denominator grow by the same factor, canceling each other out.

 

Also- focal distance is a distance from lens nodal point, an that point can be almost anywhere, for different types of lens construction.

Fortunately the location of the nodal point is not a factor here, so we can safely ignore it.

 

There is also that diffraction is different for red and violet on the other side of the spectrum. Would this mean that diff. is bigger if the WB too red?

No, it wouldn’t. White balance makes no difference. Now if you did set the color temperature to some really perverse value that was way off, then either red or blue would contribute much more to the image than it should and in such extreme cases it could, theoretically, make a difference. But that would be the least of your concerns then.

 

Green is the range of wavelengths both our eyes and the sensors of our cameras resolve best, so once we have made sure that diffraction won’t cause any problems in the green channel, there is no need to worry about red or blue.

 

I might just add that, according to Sean's calculator, I can use f22 for any lens with my 5D mk2 with no diffraction. This is just not true.

Given a pixel pitch of 6.4 µm for the EOS 5D Mark II it would be prudent to stay below 2 x 6.4 / 1.35 which is approximately 9.5.

[mjh]

but airy disks aside, my practical experiments have shown that the diffraction limits which are often calculated mathematically don't always play out in practice. between the rather clever demosaicing algorithms and sharpening, i can often pull detail out of shots made at f/16 that wasn't supposed to be there.

That’s because the Airy disc isn’t a disc of uniform brightness. It is not like light that should be focused to a point was spread to a disc resembling a circle of confusion. Rather the central part of an Airy disc (that is surrounded by concentric circles of diminishing brightness) is like a fuzzy blob, with the diameter corresponding to the distance from the center where the intensity is zero. So if two Airy discs get so close as to overlap, that doesn’t mean there is no more contrast – it is just that contrast is reduced and you can still salvage some of that lost contrast by sharpening. And neither does it mean that there is no loss of contrast due to diffraction as long as the Airy discs are small enough so they won’t overlap. Even then there are still the concentric circles overlapping. We can usually ignore their effect as it is so small, but it is still there.

[xtoph]

[...] the central part of an Airy disc (that is surrounded by concentric circles of diminishing brightness) is like a fuzzy blob, with the diameter corresponding to the distance from the center where the intensity is zero. So if two Airy discs get so close as to overlap, that doesn’t mean there is no more contrast – it is just that contrast is reduced and you can still salvage some of that lost contrast by sharpening. [...] We can usually ignore their effect as it is so small, but it is still there.

 

yes, that's a helpful explanation, and basically how i understood it as well, though not how it is always presented (eg over on luminous landscape).

 

i agree that one can detect the effect of diffraction in the absence of any other limiting condition... but for a large proportion of practical photography, other conditions--stability, focus accuracy, loss of detail at high iso, subject movement, etc--will be a greater limitation on final resolution than diffraction, even though diffraction is always present.

[ho_co]

Ian's opening post raised the question of why his stopped-down Noctilux shots were not sharp, and phrased the question in terms of diffraction.

 

But remember, the 50/1 has relatively low contrast for small details at f/5.6, and one must take into consideration the residual focus shift at this aperture, according to the MTF diagram and caption on p 31 of LFI 2/2010.

 

In other words, for this type of photography, a different 50 mm might be preferable.

 

This is a fruitful discussion of diffraction, but it may be missing the specific point that leads to the perception of unsharpness in this case.

[xtoph]

[...] remember, the 50/1 has relatively low contrast for small details at f/5.6, and one must take into consideration the residual focus shift at this aperture, according to the MTF diagram and caption on p 31 of LFI 2/2010.

 

In other words, for this type of photography, a different 50 mm might be preferable.

 

This is a fruitful discussion of diffraction, but it may be missing the specific point that leads to the perception of unsharpness.

 

i agree both with the suggestion that another lens would have been preferable, and that diffraction is probably the least source of perceived unsharpness in this particular case.

 

but are you suggesting that at f/5.6, dof at infinity is insufficient to counteract focus shift? that would surprise me. also, i expect focus shift to be more severe at f/4, but he was satisfied with sharpness there; also, the offending shots were at f/8, he said, not f/5.6.

[mjh]

This is a fruitful discussion of diffraction, but it may be missing the specific point that leads to the perception of unsharpness in this case.

Indeed it might. Also we are dealing with far away landscapes here so scattered light may have compromised sharpness as well.

[ho_co]

but are you suggesting that at f/5.6, dof at infinity is insufficient to counteract focus shift? that would surprise me. also, i expect focus shift to be more severe at f/4, but he was satisfied with sharpness there; also, the offending shots were at f/8, he said, not f/5.6.

 

x, those are valid points that I can't answer with certainty. Consider also that a particular (high-frequency?) aircraft vibration may have appeared with the f/8 pictures that wasn't present earlier. It was later in the day; Ian may have been tiring or getting cold; the wind situation in the cockpit might have changed; etc. I think his remark in post #6 is to the point: Without extended tests, we are all simply guessing.

 

To your specific questions:

 

As for f/8 vs f/5.6, I don't think MTF will improve at f/8; that's why it's usually shown for open aperture and f/5.6. I think Andy Piper made some experiments with the M8 which he published here, indicating IIRC that about f/5.6 was the aperture beyond which quality deteriorated. (I know, M8 isn't M9.) Erwin Puts' tests also tend to imply that for most lenses, stopping down further than about f/5.6 is seldom helpful for optical quality. I doubt that there is much visible difference between f/5.6 and f/8 with most modern lenses.

 

I won't be drawn on DoF. I've expressed my opinions often enough on the topic here. I simply cited the LFI caption.

 

 

I think the important thing to notice here is that although the classical rules still apply, with digital other aspects of image formation can become more important. We can't simply extrapolate from analog to digital without considering factors such as sensor depth, Bayer matrix, pixel pitch etc.

[xtoph]

x, [...] Consider also that a particular (high-frequency?) aircraft vibration may have appeared with the f/8 pictures that wasn't present earlier. [...] we are all simply guessing.

 

As for f/8 vs f/5.6, I don't think MTF will improve at f/8; that's why it's usually shown for open aperture and f/5.6. Andy Piper made some experiments with the M8 which he published here, indicating IIRC that f/5.6 was the aperture beyond which quality deteriorated. (I know, M8 isn't M9.) Erwin Puts' tests also tend to imply that for most lenses, stopping down further than f/5.6 is seldom helpful except in increasing depth of field.

 

I won't be drawn on DoF. I've expressed my opinions often enough on the topic here. I simply cited the LFI caption.

 

[...] digital adds aspects that didn't apply with film. We can't simply extrapolate from analog to digital without considering sensor depth, Bayer matrix, pixel pitch etc.

 

actually, i already suggested that vibration in that little plane would be the first suspect in unsharp results. i agree that we're all just guessing.

 

i wasn't suggesting that stopping down to f/8 was likely to make a big difference in mtf. i was pointing out that dof would be even greater at f/8, which would be even more likely to cover any focus shift, which would still be a rather small concern at infinity. that's been my experience with lenses which exhibit focus shift; i was wondering if there were some specific property of the noct that made a difference. i don't know anything about your or the forum's history talking about focus shift, i was simply trying to elicit information.

 

while i agree that there's lots of digital-specific concerns to imaging, with regard to diffraction i think too often people get sidetracked by calculating the theoretical limits of resolution by a sensor of n megapixels, and can forget that in most photos they will never come close to achieving that theoretical maximum for mundane reasons having nothing to do with diffraction. in this respect, digital is no different from film, i'd say.

[ho_co]

... i don't know anything about your or the forum's history talking about focus shift, i was simply trying to elicit information.

Typo? Are you referring to my comment on not wishing to be drawn on depth of field? If so, sorry, but that's a can of worms on which I have firm opinions. The forum has had several long and heated discussions on the topic. I don't want to go there now.

 

... with regard to diffraction i think too often people get sidetracked by calculating the theoretical limits of resolution by a sensor of n megapixels, and can forget that in most photos they will never come close to achieving that theoretical maximum for mundane reasons having nothing to do with diffraction....

Christoph, I agree. There's often too much concentration on minute technical details that have little use in making pictures. And I have a tendency to get involved in those discussions as much as anyone else.

[ianspector]

I believe that my initial assumption was wrong when I attributed the lack of focus of my small aperture shots as compared to my larger aperture shots as diffraction. I am satisfied that it was not the case but will use f/8 as a ceiling.

 

The images were only taken a few minutes apart so tiredness was probably not an issue but perhaps even at 1/4000 the wartime Gypsy Major may have been causing new vibrations. A factor not obvious from the posted images was how much haze there really was which really nailed the histogram on the right hand side.

 

The next comment on why a Noctilux in a Chipmunk. As I placed the precious metal and glass around my neck and found it trying to hit every sharp object the RAF put in its planes I asked myself that question. Firstly it is the only 50mm I own. I have the 28 Cron but I would need to be at 500ft for that to give me the effect I wanted. My 90 Cron would have made any vibration problems worse.

 

I will run some tests in more controlled conditions and report back.

 

Thank you so much for your help.

 

Ian

[lars_bergquist]

So how it really is?

I have read on ReidReviews that faster lenses become softer sooner due to diffraction, than slower lenses.

Lars is also mentioning that.

 

I have heard elsewhere too, that slow\fast doesn't matter when we talk about diffraction.

But no-one neglects, that in most cases faster lenses become softer sooner, than slower lenses...

 

So probably this is a fact, but root cause is not diffraction, but lens construction?

 

As I pointed out, the location of the 'sweet spot' depends very much on the level of residual aberrations in the lens, at any f-stop. The higher that level is, the more you have to stop down in order to have moved enough aberration out of the way, for the rising level of diffraction to be felt (and seen).

 

Very fast lenses must have their aberrations more rigorously corrected than slow lenses; they would be unusable otherwise. So with modern, very advanced high speed lenses, it is not unreasonable to find that, as shown by MTF graphs, the general contrast of the lens can begin to drop slightly (a sign of having passed the diffraction limit) already at f:4, while a slightly slower lens may show this effect only from f:5.6, and an ancient lens maybe not at all within the available range of f-stops. Really old Leica lenses often have stops down to very small apertures, while current lenses normally put a stop to the game at 16. Now you know why.

 

Now the nature of the sensor may of course introduce complications because of its regular pixel grid. Life was simpler with the completely stochastic grain of film. Visible grain could mask the effects of diffraction; instead of sharp micro-detail you got sharp grain! But diffraction as such occurs IN THE LENS, though we see it behind the lens -- after the fact. The sensor impinges on HOW we see it.

 

The old diffracted man

[lars_bergquist]

[ ... ] while i agree that there's lots of digital-specific concerns to imaging, with regard to diffraction i think too often people get sidetracked by calculating the theoretical limits of resolution by a sensor of n megapixels, and can forget that in most photos they will never come close to achieving that theoretical maximum for mundane reasons having nothing to do with diffraction. in this respect, digital is no different from film, i'd say.

 

I am completely with you in this respect. This is also my opinion (or rather, my observation).

 

The old diffracted, refracted and reflected man

[Advertisement]

Hallo,

Look here: Leica M

[ho_co]

Ian, I meant no disrespect in referring to tiredness and cold. I only wanted to say there are many possibilities.

 

(As for shutter speed, I have a tremor that sometimes makes shots at 1/2000 unsharp, though that's on an image-stabilized D-Lux 4, not apropos here. On my M8, I need to shoot a minimum of about 1/250 to get reasonably consistent results, and I haven't seen problems at 1/1000 or higher.)

 

I think the amount of haze you describe is likely to have more influence on the matter than the other explanations we've advanced.

 

 

Lars, I think your definition of the place of diffraction within the range of other lens aberrations is spot on, and a textbook example of clarity. However, I think Michael's comments on relating the size of the Airy disk to the pixel separation is also important. That's a case where the simple fact of the existence of the sensor can lead to effects not seen on film. There's a fair amount of material on the Web these days that, rightly or wrongly, refers to these artifacts as 'diffraction.'