Chromatic Aberration

Discussion in 'Digital Photography' started by Roger, Nov 23, 2004.

  1. Roger

    Roger Guest


    I've been searching for, but haven't found a suitable chromatic
    aberration filter to use with my Canon s60 images (plus others). I'm
    currently using PaintShop Pro and the new release 9 includes such a
    filter. However, my wife wants to switch back to Photoshop (we've had
    several years experience with a version 5.0). We would likely select
    Photoshop Elements 3.0.

    Is there stand alone software available to do CA corrections. I know
    there are some PhotoShop plugins, but many have caveats that they are
    not compatible with Elements.


    Roger, Nov 23, 2004
    1. Advertisements

  2. Try the free plugin panotools. In addition to fixing CA it will also
    do barrel or pincushion distortion. A web search will find a copy, plus
    lots of info on how to use it.
    Robert Feinman, Nov 24, 2004
    1. Advertisements

  3. Roger

    TAFKAB Guest

    If you're looking to solve purple fringing, Shay Stephens sells a very nice
    action for $10.
    TAFKAB, Nov 24, 2004
  4. SNIP

    Bart van der Wolf, Nov 24, 2004
  5. Roger

    Bill Perkins Guest

    Check this site for an excellent procedure that can be used in Photoshop or
    Photoshop Elements. It works!

    Bill P
    Bill Perkins, Nov 26, 2004
  6. Roger

    KBob Guest

    Since a lot of CA is created by light angles of incidence on the
    sensor surface that exceed about 15 degrees, there's really no special
    filter that deals with it directly. There are blurring filters
    intended to correct mosaic artifacts, but that's about it. For CA
    you're best off using the available postprocessing tools that allow
    you to separately adjust R-G-B diameters--often these are found in
    custom postprocessing programs from Kodak & Nikon that are used mostly
    with RAW images.
    KBob, Nov 27, 2004
  7. While we're at it, if chromatic aberration can be corrected by a
    computer program, and I think it can, why then don't we have
    digital cameras with cheaper lenses without any correction of
    chromatic aberration, but with a program built in that does the
    correction after taking the photo? Can't be that difficult,
    since the lens is sharp for each color and the aberration of the
    lens is precisely known.

    Hans-Georg Michna, Nov 29, 2004
  8. Roger

    imbsysop Guest

    The result of lens CA is color spectrum dependent .. How is one
    supposed to correct an artifact that is entirely dependent on the
    various levels of colors presented in the picture aka dependent on the
    picture taken ? build in an Artificial Intelligence in the camera that
    makes a "best fit" evaluation ? What if it it fails ? ruined picture ?
    imbsysop, Nov 29, 2004
  9. Roger

    Mike Guest

    What a program can do is simply replace the red/purple/blue
    CA with a neutral gray color. This is NOT correction, but an
    attempt to hide the problem and make it less noticeable.
    Detail lost to CA cannot be magically recreated.
    Mike, Nov 29, 2004
  10. ... and barrel and pincushion distortion could be corrected in firmware,
    too, given enough CPU power in the camera...

    David J Taylor, Nov 29, 2004
  11. Roger

    Pete S. Guest

    And strangely enough, we use the same spectrum in every picture. Some
    colours more, some colours less. But always the same spectrum.
    It's not dependent in the levels of colours presented. Distortion on
    red is always the same, blue is always the same, green is always the
    same. Just work out what the lense does to each of the RGB channels
    and resize each of the channels to compensate. This is how presets for
    photoshop plugins (panotools) can be "preset" for a given lens.
    It doesn't need to be "best fit". The RGB abberation is a set of
    constants for each lens, according to the material the lens is made

    If the above still cause you some confusion, some very basic physics
    using refraction / diffraction should sort you out.

    Pete S.
    Pete S.
    Pete S., Nov 29, 2004
  12. Roger

    Martin Brown Guest

    Lateral colour which is the most common aberration in a normal
    achromatic compound lens can be corrected. The image is in good focus
    but with a magnification that depends on the wavelength of light. This
    leads to radial colour fringing well off axis.

    The longitudinal chromatic aberration in a simple lens is not easily
    correctable. The image is in proper focus at only one wavelength
    typically green and blurred at the others.
    They are way more sophisticated than that. You can do it manually by
    hand using split to RGB and rescale the red & blue images to match.

    CPU and power requirements for a portable device.
    Some of it isn't lost - just in the wrong place.

    Martin Brown
    Martin Brown, Nov 29, 2004
  13. Roger

    imbsysop Guest

    that's why I wrote "The RESULT of lens CA ..."

    You assume just one single wavelength for each color ... there is no
    such thing as "blue" or "green", it's a bunch of wavelengths, there is
    no RGB in nature, only in "digital", think backwards :)
    imbsysop, Nov 29, 2004
  14. David,

    of course. I'm wondering why they haven't done this.

    The preceding two messages are wrong. The information is not
    lost, only distorted according to color. All that needs to be
    done is enlarge the red picture a little and make the blue one a
    little smaller, so they all match the green one. Not difficult,
    if you know the parameters of the lens, and probably even
    possible without, by using mathematical techniques like

    Hans-Georg Michna, Nov 29, 2004
  15. Roger

    Roger Guest

    My only experience with CA reduction is recently with PSP9. It may be
    possible to do in camera reduction but the current techniques as
    implemented the PSP9 tools may not be adequate for in-camera work.
    Currently the user has to identify the area to be corrected. If the
    user identifies an area (with rectangular box) that is too broad (or
    that does not contain CA), the correction is too aggressive (modifying
    colors other than the "purple" colors). The degree of correction also
    seems to depend on the "intensity" of the CA. IOW identifying only one
    area may leave residual fringing in other areas, so multiple areas
    must be defined.

    Using these techniques would require in-camera software that searched
    the edges in a photo on the high-contrast boundaries and applying
    corrections (possibly iteratively) and testing adjacent boundaries to
    determine the adequacy of correction and making second order changes.
    This seems a bit cumbersome, there must be other ways. I'm just trying
    to describe the process with one of the current tools (that does a
    decent job albeit sometimes tedious).

    Once I began to identify the CA in my photos, it becomes more
    important to me to remove it. By comparison to others, I thought the
    CA in my Canon S60 was reduced in CA effects. Now that I'm worrying
    about the effects on more critical "prints", I find it interesting on
    how much more I'm aware of the CA fringing and sensor blooming in
    night photos.

    I'm pleased to have the tools in PSP9 to begin to deal with the
    problem, because I don't think I'll be able to replace the camera very
    soon. The troubling part is that I do a lot of night time photography
    due to my business travel schedule. The sensor blooming becomes a real
    problem and the PSP9 tools can handle them to some degree - you still
    have the over-exposure portion, but you can reduce the purple halo's.

    Roger, Nov 29, 2004
  16. Likely CPU power - it would add substantially to the processing time of
    each shot, and will loose some resolution in the process.
    In addition to the CPU power required, part of the issue may be that the
    problem isn't actually chromatic aberration at all - but light incident
    onto the sensor at other than 90-degree "normal" incidence. The
    correction probably depends on which adjacent pixel is getting the
    spurious illumination. A tractable problem, though but, like geometric
    distortion, possibly different for each zoom setting.

    David J Taylor, Nov 29, 2004
  17. Roger

    Big Bill Guest

    How would software know what's an aberration, instead of what's in the
    subject itself?
    *We* know, because we can recognize the subject. While there is
    software that can recognise *some* subject matter, I seriously doubt
    you'd want to pay for it, much less such software that would be
    intelligent enough to recognize and correct all subject matter.
    Big Bill, Nov 29, 2004
  18. Roger

    Annika1980 Guest

    Just another good reason to shoot in RAW mode and use ACR's CA removal tool
    when converting.
    Annika1980, Nov 29, 2004
  19. Roger

    Pete S. Guest

    Cor, do you have a digital camera with an infinite number of colour
    channels? Sorry, that would be an analogue camera.

    I, and the rest of nature may well be analogue, but the cameras, and
    the processing software, are not. They use RGB. Or CMYK. So in a
    digital cameras world there really is only a single wavelength for
    each colour. The "inbetween" colours are merely combinations of the
    digital channels.

    Pete S.

    Pete S.
    Pete S., Nov 29, 2004
  20. Roger

    Pete S. Guest

    Because abberation is a function of the lens, not the subject. Nikon
    produce ED lenses for this very reason.

    It is constant, but shows up more on certain subjects, say where you
    have a sharp edged black object and a light background. A power cable
    across a sky, for example, that cuts across a corner of a picture.

    Pete S.
    Pete S., Nov 29, 2004
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.