A study in lens diffraction f2.8 - f 22

Discussion in 'Casual Photo Conversations' started by invisibleflash, Feb 21, 2021.

  1. DoF and diffraction. Depends on how you understand DoF.

    Diffraction increases the minimal attainable CoC size, so DoF as blur below the visual acuity of whoever is viewing the image under a given set of circumstances decreases.

    DoF as the visual difference between sharp and unsharp increases. More DoF, but overall less sharp.
    Better lenses have less DoF at the same aperture.

    And yes, if the medium is not up to the job of resolving what a lens projects, we do not get to see differences in the range beyond the medium's capabilities. So more DoF, but less resolution, regardless of lens and diffraction.
    Last edited: Feb 23, 2021 at 11:45 AM
  2. The usual method of manual focusing is to establish two points on either side of the focal plane (subjectively) eually out of focus, using the focusing ring, then divide the distance. This works reasonably well with a ground glass, and particularly well with a digital camera with the assistance of focus magnification.

    Objectively, these two ring positions are well within the marked DOF, regardless of the method.

    Defining DOF as the observable circle of confusion in an 8x10 print, viewed at a distance of 10 inches, is a convenient fiction, with little application to real life. The greatest fiction is the hyperfocal distance. Most of us have found landscapes turn out better if you focus precisely on the object of central interest, rather than use the hyperfocal setting which often renders every major point OOF. The same is true if you use an aperture of f/11 or smaller, except when "published" inside the 1000 pixel limit on PNet.

    Alternately you can use the Scheimpflug Effect, or focus stacking.
  3. Either that or use the "double the distance" method explained here: Double the Distance Method Explained (photographylife.com) (do a search on that site to discover a few more articles on focusing in landscape photography. I never got along with hyperfocal and had only limited success with Merklinger's approach; "double the distance" has the advantage of being easy to apply in the field.
  4. Using the DoF markers and distance scale on lenses works and is very easy. That is, if your lenses still have such thingies.
  5. One of the key advantages of MILC is that you have 100% live view, where the viewfinder (EVF) is bright enough to actually preview the DOF and its effect on the composition. I don't avoid OOF elements, I use them to isolate and complement the desired subject. Sony engineers must have similar thoughts, because even their AF lenses focus using the preset aperture. My preferred lenses are manual (Zeiss Loxia), which are strictly mechanical with DOF scales. Zeiss AF lenses (Batis) also have DOF scales, which are presented in an OLED display on the lens. However my only use of that display so far is to set the focus accurately at infinity for astro photography. (You can focus on stars only in suburban situations with a relatively bright sky. In really dark skies the EVF has too much noise to see anything.)

    For those who have not been introduced, even peripherally, to the law, the concept of "a reasonable person" in determining justificaion is an example of a legal fiction. We are all expected to act reasonably, but the definition is left to the judge or jury.
    Last edited: Feb 23, 2021 at 6:03 PM
  6. Who said I define DOF that way? Currently my approach is to define acceptable DOF as a CoC of .019mm.
    Using the above CoC as my reference, it works fine for me. I determine my focal length, nearest point of interest, figure out if I can get from there to infinity at f/8 or f/11 and if not decide if I want the foreground or the background to go out of focus. While the usual approach is to sacrifice the background because its details are larger and more distinguishable, when I'm shooting with something like the Golden Gate Bridge (another example might be the Eiffel Tower) then I will choose that over the foreground. Further complicating matters for me (YMMV) is field curvature, so my Tokina AT-X Pro 14-20mm f/2 which is stellar even wide open where it's in focus has pronounced field curvature so where hyperfocal distance formulas indicate f/8 is enough it actually requires me to go to f/11 because I have to overcompensate to keep the corners sharp (a pet peeve of mine).
    You either ignored or otherwise disregarded my earlier posts (above) about post-processing as it relates to diffraction at f/11.

    Anyway, here's a photo I made back in 2016:

    Ignore the EXIF data regarding the f/stop, it was taken at f/11 and f/2.8 was just the maximum aperture of the lens. These days I figured out that you can put the same lens into the non-CPU lens selection in multiple banks on my Nikon cameras, and have f/8 and f/11 options for my T/S lenses (if I can just remember to set it correctly before taking the shot).

    With sharpened 100% crops:


    Compressed some to keep the file sizes here at Photonet down in size. It looks good, even viewed up close, as a 20x30 inch print up on my wall.
    Last edited: Feb 23, 2021 at 7:21 PM
  7. That is the official description, on which lens engravings are based. A COC of 0.019 mm (0.02 mm) may be neither appropriate nor achievable on a mural sized print.

    Interesting photos, but soft and of low contrast even by PNET standards. I'm not convinced I want to follow your example.
    Last edited: Feb 23, 2021 at 8:50 PM
  8. My DOF calculator allows me to input my own CoC. Also, the "official description" is for 135 format, mine is actually based on DX format and I carry it over to FX/135 format. I strive to have the DOF be applicable to 20x30 inch prints.
    It isn't going to get much smaller than that and still be achievable with our current sensors. If it's bigger than that then why would anyone worry about diffraction?
    It's one photo and 100% crops from said photograph.
    Speaking of mural-sized output, those crops translate to something like an 8x12 foot print viewed up close. As far as the pixels themselves being soft, well the lens is so-so and I chose it because it had shift capability, and those pixels wouldn't be any sharper at f/5.6 than they are at f/11 (as shown above using a better lens in my previous posts in this thread).
    FWIW, here is the version from my Imageevent website resized to 800 pixels and 327 KB:


    Like I said, it looks good, even viewed up close, as a 20x30 inch print up on my wall.
    By all means don't. It's always such a "pleasure" to get drawn into nasty exchanges.
  9. The irony is that we are now discussing DoF - an apparition that only sometimes, and only briefly, could convince someone that it would be real - with someone who calls diffraction - a 'hard' physical reality - "hypothetical".
    You have to smile.
    Last edited: Feb 24, 2021 at 4:05 AM
    Ed_Ingold likes this.
  10. So you're shooting a landscape and want to get the DOF you need. So how do you decide whether to trade-off DOF against diffraction or vice versa? I've always calculated the DOF I need and then added a bit just for good measure. Of course, that adds the possibility for more diffraction. Sounds like I:m screwed either way.
  11. Or to put it another way, at what point would the average viewer notices diffraction over not having enough DOF?
  12. It's indeed a balancing act, Alan.

    You have to consider lens faults and their effect on performance too. Stopping down will generally improve a lens' performance and only when stopped down to middle apertures and beyond will the effect of diffraction begin to reduce performance again. Very few lenses are diffraction limited wide open, and those will then only be diffraction limited in the center of the image. So stopping down generally improves performance, because at first lens faults reduce performance more than diffraction would.
    Upto a point. There is that often mentioned "sweet spot", "optimum aperture", where the improvements due to stopping down (elimination of peripheral rays and the reduction of abberations) are in balance with the degradation caused by stopping down (diffraction).

    DoF however needs what it needs. If an image requires little or no contract between obviously sharp and obviously not sharp, i.e. needs 'infinite' DoF, you need to do what you need to do to get that, even if that reduces the maximum sharpness in the image. Conversely, if you need all the resolution you can get, you may have to trade in some DoF for that.

    Personally, i never aim for maximum DoF, but do what others have mentioned before: put focus on the main subject, or main point of interest (or main 'anchorpoint' for the viewers eye) if the subject is extended quite a bit (landscape? Usually there is some focal point in a landscape too that you can put focus on). But sometimes you just need DoF to extend between two given points, and all you can do is try to get there without doing too much damage.
    In landscape photography, using front tilt helps. It does not increase DoF, but repositions the plane of focus, from which DoF extends. Someone pointed to an article describing a "double the distance" method, and there too tilting the camera is recommended. However, doing that as described in that article will not help that much, because you can only tilt a fixed camera as far as framing will allow. You need a camera with movements to get the real and full benefit of tilt. (I do not think much of that article and the method it described, by the way. Using the DoF scales on a lens offers more and more precise control. But if it works for the author...)

    Extreme and high magnification photography is a field in which DoF is extremely shallow. You can stop down quite a bit in an attempt to get as much DoF as possible, but the gain in DoF will be very small, and image degradation due to stopping down will be quite dramatic. That was very problematic, until digital capture and image processing became available. Focus stacking provides an easy, though laborious and time consuming, solution. Before, we had to use techniques like slit illumination photography, where the subject was lit by a verry narrow slit, positioned in the plane of focus, moving the subject up through that plane towards the camera. Worked, but not suited for many subjects.
    Last edited: Feb 24, 2021 at 8:09 AM
  13. There's a difference between the theoretical limit on resolution set by diffraction, and a slight and very visible degradation in contrast.

    The absolute (theoretical) limit on resolution set by diffraction at f/5.6 is around 260 lppmm @ a monochromatic light wavelength of 520nm. This is obviously well below what most digital cameras, and any conventional film can resolve. However, the effect of diffraction over multiple wavelengths - i.e. real life use - is obviously greater, and gradual rather than a sharp cutoff.

    Regardless of what theory dictates, the effect of diffraction at mid apertures (f/6.3 - f/8) can still be seen with a good lens and a 24 megapixel DX or ~40 megapixel full-frame sensor. It can also be countered to a large degree by applying a small amount of digital sharpening.

    It has to be remembered that detail in a real-life subject, as opposed to res-test line-pairs, doesn't conveniently limit itself to aligning nicely with the pixel spacing of the sensor. The point spread (Airy disc diameter) of diffraction falls between pixel spacings and reduces pixel-to-pixel contrast; long before it completely smudges the difference between those pixels - or rather photosites.

    The effective aperture also has to be taken into account. E.g. at a lifesize (1:1) reproduction ratio the effective aperture number is double that marked on the lens scale. And it's the effective aperture number that dictates the degree of diffraction.

    Anyway. Long and short of it is that the effect of diffraction can be observed even at apertures like f/5.6. And even more so at macro distances.
    Last edited: Feb 24, 2021 at 9:18 AM
  14. Under what conditions is it visible? When pixel peeping? I'm much more concerned with what is noticeable viewing a print from a normal distance than with what can be detected with a magnifing lens. People rarely decide great paintings aren't good because of you stand 4 cm from the painting, you can see some of the marks of individual bristles. A common standard is whether difraction is visible in an 8 x 10 print at 1 foot. That's of course arbitrary, and the effective limit will depend on print size and viewing distance.

    So my approach is simple: try to stay below the limit calculated using the standard above, and go above it when necessary, being aware that prints may show some degradation. In practice, they rarely have shown any degredation that anyone but me notices, and I rarely do myself.
  15. I doubt that the "average" viewer notices much beyond shapes and color. Diffraction is noticed by the critical viewer, bordering on (or surpassing) obsession. A lot depends on knowing your equipment and what it can do at its best. There's not much you can to spoil the sharpness of a 6 MP image, short of a sloppy dog lick on the lens. At 50 MP, the obsessive owner will notice that pine needles on the horizon are no longer defined. As q.g._de_bakker stated, it's a balancing act. At f/16, you get 6 MP results regardless of the sensor.

    Unlike DOF, diffraction (and residual aberrations) paint with a broad brush. All parts of the image are affected, and you lose fine detail where you would expect it. The subject matters a great deal. You expect a mountain panorama to be blisteringly sharp, but not a candid scene on a train platform. You can exclude elements which would distract from the composition. A row of OOF plants across the bottom of the scene is distracting, but can add to the composition if they frame a key object. Selective focus is another technique. Morning frost on an agave plant in the foreground had better be sharp, even if the escarpment behind is a little soft. If the lower third is in the dark, make it darker or crop, rather than half revealing distracting detail. Since diffraction doesn't depend on focal length, increasing the film (image) size reduces its relative effect.
  16. The average viewer sees a lot that they wouldn’t be able to specify or articulate. Resorting to the average viewer as a guide seems dubious to me. Lots of subtleties inform the viewer that the viewer doesn’t notice or couldn’t reasonably be expected to know or talk about. I don’t spend time or energy on things because I think most viewers will understand what, precisely, I’m doing or care. Nevertheless, viewers are affected by these choices, often without noticing. Cinematographers don’t necessarily utilize deep focus techniques because they think their audience will notice that. They’d probably prefer their technique to go unnoticed by audiences. The important thing is how these choices make the viewer feel, not whether they are noticed.
    Ed_Ingold likes this.
  17. I remember reading that back in 2009:


    That's when I realized that there are ways to deal with diffraction and that it isn't a brick wall you can't go past, at least with the cameras and lenses I used back then and since. Like I said, it's a slope and not a cliff; sensor resolution on the other hand, is a cliff, as you can see above (I need to stop mixing my metaphors).
    Scroll up to where I referred to the scenario you proposed as hypothetical, and later where I point out that we already can see the practical limits of diffraction with very small sensors (which if you scaled them up to 135 format are in the range of 600 or more MP). I will say it again (in not so many words), such concerns about diffraction are hypothetical in the same sense that you might be worried that you won't be able to see to infinity with an improperly focused lens.
    It may be different at the extremes, but the rule of thumb I work with is two f/stops doubles your DOF. I also rarely stop a lens down more than f/11, but I routinely use f/11 when the photo I envision calls for it:


    100% crop:


    I also sharpen for output, and how I sharpened the 100% crop would be too much sharpening for the 1080 pixel final output I used for this photo intended to be seen on a computer monitor.
    Yes, it can be observed when looking closely at the original file, and I do that all the time. OTOH, as the photographer I am ultimately in control of what the viewer observes.
    Last edited: Feb 24, 2021 at 11:31 AM
  18. I'd rethink that one. You're in control of what you put in front of the viewer, which gives you much control over what the viewer observes, but not ultimate control over what the viewer observes. Viewers have different abilities to see and to understand what they see. One viewer may be much more observant and attentive than another. Letting go of that sense of control can be as important as knowing how to use it.
  19. Yes, do scroll up and notice i did not propose any scenario. I did no more than explain how diffraction reduces attainable resolution.
    You called that "hypothetical", and that is rather ignorant nonsense. Period.

    And yes, do scroll up and see that you - like now - are the one offering "scenarios" in which diffraction would be not much more than a fictional thing. You did not think diffraction limits. Zeiss, for instance, think otherwise, and call it (i quote their data sheet of one of the few diffraction limited lenses) "the final, the unavoidable of all limitations: the diffraction". You did not think that limit was as hard as a brick wall ("it's not as if there is a brick wall"), more like a "steep slope". You again claim it is not a brick wall and that there are ways to "deal with it". Migh, migh... those ignorant people at Zeiss call it "unavoidable". Imagine that! You said you were "totally uninterested in such a hypothetical scenario" (there it is again) such as the existence of a hard law of nature that cannot be stopped, cannot be deterred by any scenario, from doing what it does.
    In short: a bunch of nonsense that cannot be put right by whatever you say. Except, perhaps, by an admission that it was indeed nonsense.

    So stop protesting. Stop misrepresenting people who told you how things are. Acknowledge what we all know to be a fact.
  20. Your "perfect lens" having its diffraction recorded at a wide apertures is currently a hypothetical scenario. As for my lenses being diffraction "limited" past f/11 that's not what I see and I bet you can't show it either (and it's worth noting here you haven't even tried), they are (as I said) diffraction impaired. Once the sensor has enough resolution to record all of that impairment then we could characterize it as limited, but we have a ways to go before we get there.

    As I said earlier, it's "fun" to get drawn into nasty discussions. I've had enough of that. I'm done responding to your offensive and increasingly rude posts.
    Last edited: Feb 24, 2021 at 12:27 PM

Share This Page