Chromatic aberrations on digital but not on film

Discussion in 'Casual Photo Conversations' started by mva, Nov 11, 2011.

  1. mva


    Hello photographers,

    I am the happy owner of a Nikon FM2n, a Nikkor AI-S 50mm f:1.2, a Sony NEX-5 digital camera, and a Nikon Coolscan 9000. (Well, I also have a few extra bits ;-) but they are not concerning my question here).

    The lens performs always very well when I use it on the Nikon, on film, in terms of chromatic aberrations: I don't see any in my 2000dpi scans.

    On the other hand, I occasionally use the lens adapted on the Sony, and I am very dissatisfied with the C.A. I see.

    I would like to understand why this happens: I have two options.

    The first: perhaps it is just one of those lenses that perform well on film, but not so well on digital. hence I would like to know your experience: do you use this lens on digital bodies? (On digital bodies that do not automatically correct CA?)

    The second: my adapter is not really a proper adapter, but it is a Lensbaby Tilt Composer. In other words, although I try to align the lens as much as I can, probably I do not achieve perfect alignment. I wonder if this could be the origin of CA.

    What do you think?


  2. Very fast lenses typically display quite a lot of optical aberrations wide open. If you use a DX camera such as the NEX-5, the image recorded is less than one-half of the area of the 35mm film frame. Thus when you make your final print, you will be magnifying the image by 1.5x. The aberrations which cause a somewhat soft image wide open on FX / 35mm film will now be a bit more obvious because of this magnification. Secondly, the photosite density of the NEX-5 is quite high, revealing imperfections clearly. Finally, digital sensors have optical layers on top of them (antialiasing filter, bayer filter, and microlens array) that may add to the CA if the light comes in at an oblique angle.
    Try a D700 with your 50/1.2. I think you should be reasonably happy with that combination.
  3. mva


    Hello Ilkka,
    So you exclude that CA could be caused by a non-perfect lens alignment?
    I am not sure that the angle you mention could be the cause here, because the rear lens of the 50/1.2 is very large. But I may be wrong.
    The D700 costs several times as much as a NEX-5 :cool:
  4. Digital sensors collect light at slightly different (narrower) incident angles than film. Perhaps that might be the source of the differences you seel.
  5. I think that once you shove a Lens Baby between the lens and camera body all bets are off when trying to compare the optical quality to the results from the same lens without the Lens Baby. If minor CA is the only payoff I'd consider yourself lucky and shoot away.
    Saying that the lens is good for film but not for digital seems extreme when you have introduced a moving adapter behind the lens for the digital comparisons!
  6. my adapter is not really a proper adapter, but it is a Lensbaby Tilt Composer.​
    Yeah, I agree with John. You're worrying about the deck chair arrangement on the Titanic.
  7. mva


    ...and I agree too, but I would love to hear from somebody who used the same lens on the same camera with a proper adapter, too. Otherwise, how can I justify the purchase of a proper adapter to use the lens?
  8. When I used a Pentax k20d, most of the lenses produced color fringing, including the esteemed 70 2.4 (regardless of aperture). I believe that CA tends to be emphasized by digital systems, although others can explain the phenomena better than I.
    I see very little in 120 scans with Mamiya or Zeiss lenses.
  9. My understanding (and I am not a real expert) is this:
    The problem is twofold. First, actual CA - which is a function of the lens optics and has nothing to do with the capture medium - is measured in the width of color aberrations as the are actually projected onto the image plane (the film of digital sensor). This width is amplified by the fact that when you are looking at a 100% magnification of an image on the screen you are looking at a much higher magnification (the ration of the size of the image you are viewing to the size that it was as projected onto the image plane in the camera) than you normally would with prints from 35mm film.
    Consider that many of us use cameras with sensors 33% smaller (in linear dimensions) than a 35mm film frame. My desktop monitor is 100 PPI and my DSLR (a D7000) makes an image that's close to 5000 pixels wide from a sensor that's 24mm wide. A 35mm film frame is 36mm wide. Do the math, and a 100% magnification on my monitor is the same magnification as a 75 inch wide print from 35mm film! At the size, you'll see any CA the lens projects.
    There's also a second type of aberration that is sometime mistaken for CA but is actually caused by the layering of filters and micro-lenses over the digital sensor. The sensitive layer on a film frame doesn't care if the light hits it from an angle. The layering of parts on a sensor does, because they will refract light in interesting ways as it comes through and they are designed to tolerate light at a certain maximum incident angle before the effect becomes a problem. With some lenses the effect can cause a noticeable fringe.
  10. mva


    Hello Andy,
    Thanks for the long explanation!
    I agree that looking at 100% digital capture, in the case of a 14 Megapixel image, corresponds to too large prints. However, I would say that the same happens when I scan film at high resolution. Doesn't it?
    The second aberration that you mention, intrinsic of digital capture: would that still look like red/cyan? or rather purple? Because the one I am referring to looks like red/cyan.
  11. Oh, one other thing. Digital is a bit different in that lens CAs look a bit different. IIRC (I'm not an expert so anybody who is should please correct me) a red/blue fringe usually means that your lens is focusing different wavelengths a bit differently - green is sharper and red and blue are a bit out of focus. The digital sensor specifically sees red, blue and green and will pick up the amount of defocusing of a wavelength more sharply than film (which softens things a bit due to a more forgiving color response) will.
    If you've got a scanner that does something like 5000 PPI (not flatbed scanner fake 5000 PPI) you should see CA on film but it won't have the same distinct coloration so you usually won't pick up on it.
  12. I'm no expert either, but this link's explanation is pretty good:
    Apparently there are more than one kind of CA. The kind I encounter is "color fringing" at the corners, as shown in the demo image at the top right of the mentioned page. I don't notice them in my film scans, regardless of the lenses used. But they are present in images shot with digital bodies, typically with wide angle lenses.
    Fortunately, fixing this kind of CA is easy. I view a raw file's corners at 100% to locate them. Then I adjust the CA Control slides in ACR's Lens Corrections panel to eliminate or alleviate them. This is an integral step in my raw conversion work flow.

Share This Page