Diffraction at small apertures, a limiting factor in sharpness?

MsmotoMsmoto Posts: 5,396Moderator
edited May 2014 in Nikon Lenses
Maybe we need to hash out the data on diffraction in a new thread. For sure, the sweet spot of the lens IMO is about one or two f/stops from wide open, and as it gets down four or five f/stops I become concerned about loss of sharpness as a result of diffraction.

Maybe the resolution of some of the modern lenses with better coatings makes the differences in stopping way down more apparent when we look back forty years to those who were in the f/64 crowd. Interestingly enough I still have a Schneider 90mm f/8 Super Angulon from the 1960's which my guess is would be fairly sharp at f/22, f/32.

So, let us see what everyone has to say on diffraction.
Msmoto, mod
«13

Comments

  • AdeAde Posts: 1,071Member
    Similar to DoF, the size of the sensor plays a role in evaluating diffraction. An image from a small sensor must be magnified more vs. an image from a large sensor, so the visibility of diffraction effects is also magnified.

    While on a 35mm we might tolerate diffraction at (say) f/11, on a 4x5" diffraction might only become visible at f/32. And images from an 8x10" camera might show little diffraction at f/64.

    On the flip side, then when using larger formats, you have to stop down to those tiny apertures to achieve the same DoF as a 35mm.
  • IronheartIronheart Posts: 3,017Moderator
    But how much are we talking here, 10%? I've seen the test charts, and while yes there is a reduction in sharpness, there is also an increase in DoF, which may be needed for a particular shot. So like everything else it is a trade-off.
  • sevencrossingsevencrossing Posts: 2,800Member
    edited May 2014
    When doing landscapes, I tend to shoot at ~ f 8 ( say ~24mm on a D800) but I don't need a huge dof
    How do people balance diffraction vs dof when shooting macro and close ups ?
    Post edited by sevencrossing on
  • proudgeekproudgeek Posts: 1,422Member
    I don't know this for certain, but I'm going to guess that macro lenses are engineered to show less diffraction when stopped down more (at least when compared to other lenses). I've never tested it, but I'd be willing to bet that my 105mm looks different when stopped down to say f/20 than my 70-200 would at the same aperture. Again, no proof, just conjecture.
    The other option, when practical, is focus stacking.
  • jonnyapplejonnyapple Posts: 130Moderator
    Ade is absolutely right, and so are you, Ironheart. Smaller sensors show diffraction earlier but sometimes it's worth the trade-off. This is one advantage of FX over DX that is clear-cut. While the amount of diffraction that happens depends only on the focal length of the lens and its aperture, and FX sensor will image more area and so diffraction is less important when framing the same image.

    The reasons for loss of sharpness on either side of a lens's sweet spot are for different reasons. At large apertures, it's because of the difficulty in designing and grinding a lens that takes all the light from each point in the scene and redirects it to a single point on the sensor. Doing it for the whole spectrum of visible light makes it even more difficult, since the materials used will bend light of one color more or less than other colors. Stopping down makes the lens less important because your camera becomes more like a pinhole camera, so redirecting the light becomes easier and easier. However, this come with the trade-off of increasing diffraction. You could imagine plotting two things on one graph as you change aperture: the size of the circle of confusion due to aberrations (lens imperfections) and the size of the circle of confusion due to diffraction. The aberration line would start big at large apertures, but drop down to zero at small apertures. The diffraction line would start small, but steadily increase as apertures get smaller. The point where they cross is the sweet spot that Msmoto is referring to. At apertures smaller than this point, you would call the lens diffraction limited.

    The longer I live the more I think Ansel Adams had it right to start with a discussion of pinhole cameras in his book The Camera. If you understand a pinhole camera, you are very close to understanding what happens when you replace the pinhole with a lens.
    CC is welcome. DC is also welcome when I deserve it.
  • jonnyapplejonnyapple Posts: 130Moderator
    If you'd like to see a nice calculator that compares diffraction for different sensor sizes and lens focal lengths, look at the bottom of this page:
    http://www.tawbaware.com/maxlyons/calc.htm

    It's a little hard to wrap your head around, and this is simplified for a single color, but the idea is there. An interesting comparison is D800 vs D7100 at 50mm focal length
    D800: 35.9mm × 24mm
    D7100: 23.5mm × 15.6mm

    It says at f/8, the D800 should already be able to resolve the fuzziness from diffraction (because it's over the 34 megapixels listed there). To make the circles of confusion of the same relative size on the DX sensor with the same field of view (by a 33mm focal length lens), you need to open up that lens to f/5.2.
    CC is welcome. DC is also welcome when I deserve it.
  • IronheartIronheart Posts: 3,017Moderator
    From http://en.wikipedia.org/wiki/Diffraction-limited_system:

    This resolution-limited f/# depends only on the pixel pitch of the sensor. It does not depend on working distance to the subject or focal length

    As examples, we can compute the pixel pitch of the sensors for a full frame "professional" SLR, a "compact" SLR or mirrorless camera, a "respectable" point & shoot, and a sub-compact "toy" camera. As of March 2014, these are reasonably well represented by the 36 Mpixel Nikon D800 which has 4.88 um pixels, the 16 Mpixel Olympus OM-D/E-PL5 which has 3.75 um pixels, the 12 Mpixel Canon S120 which has 1.9 μm pixels, and the 16 Mpixel Powershot A4000 which has 1.25 μm pixels.

    If we equate the pixel size with the diffraction limited spot size, we can calculate the smallest aperture which is (theoretically) usable without blurring the image. For the D800 it is f/11, for the OM-D it's about f/8.5, for the S120 it is f/4.3 and for the A4000 it is f/2.8. (On the A4000, this is faster than the built in lens is capable of achieving, so it is always operating with 'empty' magnification)

    From me:
    Interesting that they reference the D800, this also means that the higher pitched D7100 should see diffraction sooner than the D800.
  • snakebunksnakebunk Posts: 845Member
    edited May 2014
    Interesting that reviewers sometimes write about how a lense handles diffraction.
    Post edited by snakebunk on
  • Golf007sdGolf007sd Posts: 2,840Moderator
    D4 & D7000 | Nikon Holy Trinity Set + 105 2.8 Mico + 200 F2 VR II | 300 2.8G VR II, 10.5 Fish-eye, 24 & 50 1.4G, 35 & 85 1.8G, 18-200 3.5-5.6 VR I SB-400 & 700 | TC 1.4E III, 1.7 & 2.0E III, 1.7 | Sigma 35 & 50 1.4 DG HSM | RRS Ballhead & Tripods Gear | Gitzo Monopod | Lowepro Gear | HDR via Promote Control System |
  • AdeAde Posts: 1,071Member
    Good video @Golf

    At the same time, they only explored macro distances. As mentioned somewhere above, at close-up and macro you have so many megapixels (resolution) concentrated in a small area that you can afford to lose some sharpness to diffraction.

    E.g., that macro shot of the person's eye had ~ 36 mp just on the eye. So yeah you can shoot f/32 here and still have lots of details of the eye.

    As a counter example, when shooting a full-length portrait, the subject's eye would only be a tiny fraction of the frame... maybe 200 pixels x 150 pixels. That's just 0.003 mp. Shoot that at f/16 and that already small amount of details may get smudged by the diffraction.
  • SquamishPhotoSquamishPhoto Posts: 608Member
    Easily the most reasonable and productive conversation on diffraction that I've ever read on this board. :)
    Mike
    D3 • D750 • 14-24mm f2.8 • 35mm f1.4A • PC-E 45mm f2.8 • 50mm f1.8G • AF-D 85mm f1.4 • ZF.2 100mm f2 • 200mm f2 VR2
  • TriShooterTriShooter Posts: 219Member
    Excellent thread. The number of pixels on the subject like Ade says is key to seeing diffraction. When the D7100 came out there was a test they did in Japan where they shooting skylines. The diffraction at f/16 was pretty dramatic on the building that were smaller in the background as I recall, and less apparent on the close buildings that were getting a pixel count.

    I use Helicon software when I want more depth of field with sharpness. It is inexpensive, and the software combines the images beautifully but this kind of shooting is on tripod in dead still conditions, or on things that do not perceptively move to be effective. There are a fair number of programs like this available to get around diffraction in some situations that produce incredible sharpness.
  • PapermanPaperman Posts: 468Member
    edited June 2014
    Carried on from
    http://forum.nikonrumors.com/discussion/2984/newbie-question-on-lens#Item_30

    @TaoteJared

    This is a hopeless discussion as long as you think diffraction is a lens fault/aberration. Saying different lenses will have diffraction limits is no different than saying every lens has a different Depth of Field.

    Calculations for both are similar, mostly taking into consideration the size of the lens opening and and one's ability to "see" ( Circle of confusion etc ).

    Here is one article ( that I know won't help you ) :

    http://www.luminous-landscape.com/tutorials/resolution.shtml

    The article is about resolution but one of best I know that explains/details diffraction.

    Though being drowned in such detailed information is inescapable for most, one can easily pick up a few paragraphs to grasp the idea. Table2 summarizes diffraction limits. One can also understand that if there is any way of reducing diffraction, it would come from sensor/pixel design and not from optics/lenses.

    I am hoping to get a similar article from you instead of the usual mumbo jumbo ( lots of words, personal opinions, with nothing scientific to back ) . Please also provide the 2 links to Leica's diffraction shiftable lens ( $100 extra cost ) and the Sigma Art again which has an "altered" diffraction limit . ( not the links to lenses obviously, but to the articles backing your statements on two )
    Post edited by Paperman on
  • donaldejosedonaldejose Posts: 3,199Member
    As a practical matter, my experience has been the same as MsMoto's "rule of thumb:" For best sharpness with any given lens stop down two stops but not more than four stops. However, that is just a rule of thumb. I like to view the graphs on DPReview provided by DxOMark. For example, the new Sigma 50mm f1.4 Art is sharpest across the board at f4-f5.6. http://www.dpreview.com/lensreviews/sigma-50mm-f1-4-dg-hsm/4 I will yield my "rule of thumb" to measured data. I don't yet have that lens but when I obtain a copy I will shoot it at f4 to f5.6 when I can. Each lens probably has its own "sweet spot" when measured and for a zoom lens that "spot" may well vary by focal length. To me diffraction is a trade off with depth of field; with the greatest conflict being during macro photography when you must shoot at f16 and f22 regardless of diffraction.
  • SquamishPhotoSquamishPhoto Posts: 608Member
    To me diffraction is a trade off with depth of field; with the greatest conflict being during macro photography when you must shoot at f16 and f22 regardless of diffraction.
    The best macro photography is usually done by focus stacking, so this is hardly a must.
    Mike
    D3 • D750 • 14-24mm f2.8 • 35mm f1.4A • PC-E 45mm f2.8 • 50mm f1.8G • AF-D 85mm f1.4 • ZF.2 100mm f2 • 200mm f2 VR2
  • sevencrossingsevencrossing Posts: 2,800Member

    The best macro photography is usually done by focus stacking, so this is hardly a must.
    I must admit I have never used focusing sacking, I assume it wont work with live things like live bees

  • IronheartIronheart Posts: 3,017Moderator
    edited June 2014
    @Paperman and @TaoTeJared are both correct, which is why we are having this discussion.

    As with any scientific discussion, we need to be careful about our terms and definitions. The Wikipedia snipit I referenced above is absolutely correct; the diffraction limit of a system depends only on the pixel pitch (digital) or grain size (film) being considered. The focal length of the lens, distance to subject, are not relevant. I'll call this the maximum limit or theoretical limit. This is likely what @Paper is referring to.

    In the real world lenses are manufactured to certain tolerances and other factors come into play. Poor contrast and CA are examples of factors that decrease resolution at any aperture. The higher the quality of the lens, the closer you approach the theoretical limit, but in no case can you exceed it. I would call this the practical limit, and this is likely what @TTJ is referring to.

    As you stop a lens down, the effects of the aberrations decrease, but the effects of diffraction increase. Where the happy medium is, varies according lens design, and likely even from copy to copy. Also your tolerance for diffraction may be different than mine. And as @sevencrossing points out, I may decide to accept a bit more diffraction to capture the decisive moment.

    Finally I'll point out that the luminous landscape article that @Paperman shared with us, puts all this into perspective rather nicely. And I'll also point out that this article suggests that the classic notions of DoF should be reconsidered in the digital era and should factor in pixel pitch (a notion I put forth here a few years back).

    From http://www.luminous-landscape.com/tutorials/resolution.shtml

    Table 1 reports the maximum resolution for a diffraction-limited (aberration free) lens at different apertures and for different levels of contrast (Norman Koren explains you how those numbers are obtained). See how diffraction increases the diameter of the Airy disks in the Table 1. In real photographic lenses, resolution values approach those presented in Table 1 only in first-class designs at medium apertures, when aberrations are reduced by stopping down. The sweet spot of any lens is at medium aperture values because aberrations are reduced and diffraction effects aren’t strong yet.
    Post edited by Ironheart on
  • PapermanPaperman Posts: 468Member
    edited June 2014
    @Ironheart

    You have with one sentence explained why I am right. Could you please explain why/how TaoteJared is also right?

    Just a reminder that we are talking about DIFFRACTION only; we are not discussing combined effects of aberrations & diffraction, sweet spots or what the best aperture for highest IQ is.

    The question here is , does diffraction change from lens to lens ? . I say NO because theoretical diffraction calculations are based on the PERFECT lens ( or for NO lens one can say ) which does not exist and it CAN NOT BE BETTERED .

    Can DOF change from lens to lens ? If the answer is no, then how can diffraction - which is also based on Circle of Confusion/Capability of eyesight - vary from lens to lens.?

    Waiting for the link on the Zeiss which comes with "altered" diffraction limit for an additional fee of $100 . ( Or is it coming with an " altered " aperture for best IQ ? )
    Post edited by Paperman on
  • proudgeekproudgeek Posts: 1,422Member
    edited June 2014
    I would agree. I've been dabbling with focus stacking over the past few months and while it does yield some pretty cool results in the right circumstances, it is incredibly unforgiving (and impossible for objects that may move even a millimeter). For every decent image I create I have to trash files and go back to the drawing board twice because of something I did wrong.
    Post edited by proudgeek on
  • haroldpharoldp Posts: 984Member
    We often take a one dimensional view of factors that are part of a complex outcome.

    Does diffraction limit sharpness, yes. - open the lens.
    Does being out of the plane of focus limit sharpness - yes - close the lens
    Does camera or subject movement limit sharpness - yes - open the lens
    Is your lens sharper at some apertures than others - yes, open or close the lens.
    Does high ISO limit sharpness - yes - lower ISO
    Does camera or subject movement limit sharpness - yes - raise ISO

    What kind of image do you want. ??
    How much DOF do you want ??
    Who or what is moving ??
    etc...............

    Discussion of any of these factors out of context of the others is almost nonsensical.

    Regards .... H


    D810, D3x, 14-24/2.8, 50/1.4D, 24-70/2.8, 24-120/4 VR, 70-200/2.8 VR1, 80-400 G, 200-400/4 VR1, 400/2.8 ED VR G, 105/2 DC, 17-55/2.8.
    Nikon N90s, F100, F, lots of Leica M digital and film stuff.

  • IronheartIronheart Posts: 3,017Moderator
    edited June 2014
    Another take on it. If we look at a pitcher throwing a fast-ball, and only measure the speed (diffraction) we are missing the big picture. I also want to know how accurate (aberrations) the pich is. If you can't hit the strike zone then who cares if you have the fastest pitch?

    @paperman, think of it this way. Contrast also limits resolution, as the contrast gets lower, your ability to distinguish black from white decreases until all you have is gray. That is why MTF charts are spec'd at either 50% or 80% contrast as the limiting factor. So knowing the diffraction limit will tell you where it will theoretically kick in, but measuring an actual lens will take into to account other factors, such as contrast that will cause diffraction to be a factor sooner than the theory indicates.

    I can't help but draw one more analogy. The oft repeated legend that "according to the laws of aerodynamics, bumble bees should not be able to fly". It seems that nobody has informed the bumblebee of this "fact".
    http://www.snopes.com/science/bumblebees.asp
    The real lesson to be gleaned from this myth isn't that scientists are so blinded by technicalities that they overlook what is painfully obvious to everyone else (namely, that bumblebees really do fly), but that one needs to understand there can be quite a difference between a real-life concept and a mathematical model of it.
    Post edited by Ironheart on
  • dissentdissent Posts: 1,268Member
    edited June 2014
    I must admit I have never used focusing sacking, I assume it wont work with live things like live bees
    If you poke around in the image galleries at www.photomacrography.net/forums , you can find image stacks of many live subjects. It helps a lot to get out really early on coolish mornings so that the critters are essentially motionless, but they are still live subjects. Takes some real dedication. See John Hallmén (morfa), Walter Piorkowski and many others.

    Post edited by dissent on
    - Ian . . . [D7000, D7100; Nikon glass: 35 f1.8, 85 f1.8, 70-300 VR, 105 f2.8 VR, 12-24 f4; 16-85 VR, 300 f4D, 14E-II TC, SB-400, SB-700 . . . and still plenty of ignorance]
  • PapermanPaperman Posts: 468Member
    edited June 2014
    @Ironheart

    The subject ( the part I carried on from other topic ) is being blurred again - why I am getting advice on general knowledge on image quality/ resolution/sharpness/MTF graphs/how many factors effect resolution/balance between diffraction & other lens aberrations etc..when the discussion is purely and ONLY about DIFFRACTION and if it VARIES from lens to lens or not ?


    Post edited by Paperman on
  • AdeAde Posts: 1,071Member
    edited June 2014
    Another take on it. If we look at a pitcher throwing a fast-ball, and only measure the speed (diffraction) we are missing the big picture. I also want to know how accurate (aberrations) the pich is. If you can't hit the strike zone then who cares if you have the fastest pitch?
    Sorry Ironheart, but that is a totally incorrect analogy. :( Paperman has it right.

    In your analogy, diffraction would not be the speed of the fastball. Rather, diffraction sets the speed limit of how fast any ball anywhere can travel for a given set of conditions.

    E.g., a fastball can't go faster than the speed of light if we were playing in a vacuum. :) It doesn't matter who the pitcher is, who manufactured of the ball, what kind of spin is imparted on the ball, etc. By the laws of classical mechanics, you can't get any better than this speed limit!!

    So, Leica can't produce a "velocity shiftable" baseball that can somehow break the speed of light.

    And Sigma can't sell an "altered" baseball that can somehow break the speed of light.

    Nor can they make photographic lenses which can resolve past the limits of diffraction. (At least without first winning numerous Nobel prizes for breakthrough advancements in theoretical physics).

    Frankly people who say otherwise don't know what they're talking about.
    Post edited by Ade on
  • IronheartIronheart Posts: 3,017Moderator
    Another take on it. If we look at a pitcher throwing a fast-ball, and only measure the speed (diffraction) we are missing the big picture. I also want to know how accurate (aberrations) the pich is. If you can't hit the strike zone then who cares if you have the fastest pitch?
    Sorry Ironheart, but that is a totally incorrect analogy. :( Paperman has it right.

    In your analogy, diffraction would not be the speed of the fastball. Rather, diffraction sets the speed limit of how fast any ball anywhere can travel for a given set of conditions.

    E.g., a fastball can't go faster than the speed of light if we were playing in a vacuum. :) It doesn't matter who the pitcher is, who manufactured of the ball, what kind of spin is imparted on the ball, etc. By the laws of classical mechanics, you can't get any better than this speed limit!!

    So, Leica can't produce a "velocity shiftable" baseball that can somehow break the speed of light.

    And Sigma can't sell an "altered" baseball that can somehow break the speed of light.

    Nor can they make photographic lenses which can resolve past the limits of diffraction. (At least without first winning numerous Nobel prizes for breakthrough advancements in theoretical physics).

    Frankly people who say otherwise don't know what they're talking about.
    @Ade, You stretched my analogy way farther than I meant it. The only point I was making is that one variable (diffraction) is pretty useless by itself. You need to consider the entire MTF if you want to live in the real world.

    @paperman, the only way to measure only diffraction is theoretically in an equation, which gives theoretical answers. When you attempt to measure when diffraction sets in on a given lens, you are measuring the MTF of the whole lens and trying to separate out what might be due to diffraction. Someone prove me wrong.
Sign In or Register to comment.