Camera Color-Space Limits - A Possible Fix?

Discussion in 'Digital Photography' started by JeffTaite, Sep 13, 2004.

  1. JeffTaite

    JeffTaite Guest

    Last winter when I was in the Everglades I took many macro photos of
    the insect life there. One subject always presented a challenge. The
    Zebra Long-wing butterfly. In that it has velvety black wings with
    bright light-yellow stripes. The yellow-colored stripes were always
    very vivid when looking at them with the naked eye. Yet when taking
    a digital photo of it the yellow always washed out to white or
    light-grays no matter the exposure level. Later reviews on the
    computer monitor and print-outs were always a big disappointment.

    I suspect this happens to be one of those odd hues that fall outside
    of the colors that the Bayer filter can represent. I could sometimes
    compensate for this before the final print by attempting some manual
    color replacement, but I couldn't take it too far before it would
    ruin the overall color cast of the blacks in the wings, turning them
    into a muddy brown. (This was before I learned some other editing
    methods using masks and luminance selections, etc.)

    Then just tonight something dawned on me.

    While I was further north last year I was playing around with some
    (now rarely used) film color-correction filters to see if they would
    enhance the fall colors, or whatever they might do. I found it
    interesting that no matter how deeply tinted some of those filters
    were that the automatic white-balance in the camera would attempt to
    compensate for them and still take a properly white-balanced
    exposure. (One old filter I found in a second-hand shop worked
    surprisingly well to enhance the fall colors, a Vivitar VMC 85B.)

    So this led me to think: would it be possible to bring those few
    unique hues that are outside of the camera's color-space back within
    range of its abilities to represent colors accurately, by putting
    different color filters in front of the lens? Is there a filter that
    would bring that bright light-yellow of the Zebra Long-wing
    butterfly back into the photos after the auto white-balance had done
    its thing? If so, which direction would I go (what filter hue) to
    best compensate for this problem with this color? Could this also
    work (using a different color filter) for those that report the
    violet/blue problem that some people encounter with certain species
    of flowers on Bayer filtered CCDs? I've not run into that particular
    violet/blue problem on my camera, but I do have this rarely
    encountered light-yellow problem.

    Has anyone tested this or thought of this before? Or is this a new
    (but untested) approach to circumvent one of the few limitations
    when going digital.

    Comments from the techno-gallery are most welcome. (They'll probably
    know any answers to this just from their experience with
    color-spaces, auto white-balance methods, Bayer arrays, and
    transmission wavelengths of filters.)
     
    JeffTaite, Sep 13, 2004
    #1
    1. Advertisements

  2. JeffTaite

    Colin D Guest

    If your camera auto white balance cannot be set for either
    color-temperature or preset for various color balances, then filters are
    always going to be unpredictable, since the camera will always try to
    undo what the filter is doing.
    Perhaps you could tell us a bit more about your camera and its
    capabilities.

    I wonder about your diagnosis of your problem being a bayer color-space
    limitation. I would think that if that was indeed the cause, then the
    particular color would also be outside the color-space of your monitor
    as well, in whch case you have a double problem.

    In my experience, color-space limitations affect the hue rather than the
    intensity of a particular color. In the case of your butterfly wing,
    the yellow would be registered mainly by the red and green filter
    response of the bayer array. If one or both primaries could not match
    the corresponding component in the wing, the result would be a color
    shift, rather than a washed-out gray/white, since each primary would be
    at its maximum response.

    It seems to me that the root of the problem is exposure-related, with
    the predominant black depressing the exposure to where the yellow is
    overexposed. If you have Photoshop, or a similar program that can
    analyze spot colors, run the eyedropper over the problem area of the
    image and have a look at the RGB levels. If one or more is sitting at
    255 (presuming a 24-bit image) then it is overexposed. If all RGB
    densities are less then 255, then you do have a problem, perhaps with
    color balance.

    If you could post an image somewhere and give us the url, some of us
    could take a look and maybe do some analyzing for you.

    Colin D.
     
    Colin D, Sep 14, 2004
    #2
    1. Advertisements

  3. JeffTaite

    EdO Guest

    Jeff - were the images taken in bright sun and or with Flash?

    I have foundt hat some insects and flowers the wings and petals are made
    up of thousands and thousands of very minute hairs and the act just like
    a mirror and reflect the light back with great intensity and cause
    overexposure.

    Don't know if this is the case but thought would throw my experience in
    the mix. My solution is shade.

    Ed Oliver
     
    EdO, Sep 14, 2004
    #3
  4. JeffTaite

    JeffTaite Guest

    Thanks for your reply.

    I'll dig through my archives later and try to find some
    representative photos of this. I can post one or two on one of the
    binary newsgroups for you if that'll be okay.

    I wasn't the only one to notice this on all photos of this
    butterfly. A friend was also taking photos of this butterfly with an
    identical camera model, and he noticed the same problem with this
    species' colors (as well as only one other species at a commercial
    Butterfly-House, also with a yellow pattern). The problem only
    showed up on this particular hue of yellow, not all yellows. We were
    both trying to figure out what was causing it. When I figured it
    must be a bayer filter (color space) limitation. Even though red and
    green dyes and phosphors are being used to represent yellows it
    doesn't mean that they will have the scope of wavelengths to
    represent all ranges of yellows. This is why I think that using a
    colored filter to "bend" the overall colors before being received by
    the CCD might allow the auto white-balance to recreate this
    particular hue.

    Or, if on the outside chance that it is what you say, a luminosity
    problem, then this method might fix the problem for those reporting
    it with the blue/violet misrepresentation from bayer arrays. That
    problem is not related to luminosity at all, so I don't see why you
    would quickly presume this lost-yellow is only due to luminosity.
    Those blues and violets that are in error are, after all, being
    analyzed and represented by red and blue dyes and phosphors. This is
    no different than yellows being perceived and recreated by red and
    green ones. I find it interesting that the problem I experienced is
    at the exact opposite in hue from the blue/violet problem. This too
    is what led me to think it is what I claim, a bayer filter (camera's
    color-space) limitation.
     
    JeffTaite, Sep 14, 2004
    #4
  5. JeffTaite

    Jeremy Nixon Guest

    I can think of two other explanations, both of which I think are more
    likely.

    The first is simply that one or more color channels is clipping in that
    area due to overexposure. With the rest of the picture properly exposed,
    the yellow area was too bright, possibly due to high reflectance. If you
    can determine what channels are clipping, it's actually possible for a
    color conversion filter to help, by forcing the camera's white balance
    to boost the blue channel to compensate, leading red to be less likely
    to clip.

    The second is that this area of this insect reflects a lot of infrared
    light and your camera is overly sensitive to it. This is also commonly
    a problem with flowers; insects see into infrared, and flowers reflect
    a lot of it. I don't know much about insects, but this is what leaps
    to mind. If this is the case, a hot mirror filter can help. What
    camera in particular are you using, anyway?
     
    Jeremy Nixon, Sep 14, 2004
    #5
  6. JeffTaite

    JeffTaite Guest

    I uploaded 6 photos to "alt.binaries.comp-graphics" under the topic
    heading of "The case of the disappearing yellows." Below is a copy
    of the post I made to that newsgroup with the photos attached. (In
    case it would be easier when replying to use this copy for quoting.)

    ---------------------------


    Here are some sample photos to better illustrate the problem I've
    noticed on a particular shade of yellow in nature. So far it's only
    shown up on certain butterfly species. I'm unaware of this lack of
    accurate color representation showing up in flowers or other
    subjects.

    All photos were taken with a Sony F717 and resized for upload and
    download time savings (to 640x480 resolution). EXIF info from the
    original photos should be intact.


    "lost_yellow_01.jpg" -- the original photo. I purposely chose one of
    the very few underexposed ones that had any yellow at all showing up
    in it, just to see if I could over-saturate the yellow to bring it
    back to how I remembered seeing it. The vast majority of the photos
    of this species had the intense yellow showing up as nothing but
    values of grays and whites.

    "lost_yellow_01_remembered_color.jpg" -- this is what I remember it
    being more like, however even this 100% saturation correction on the
    yellow of the wings does not adequately bring back the rich yellow
    on this butterfly. Note that the yellow of the flower is no problem
    at all and I had no reason to retouch its saturation.

    "no_problem_species.jpg" -- a different species also with yellow on
    black that doesn't exhibit this problem at all. Its yellows having
    more red (vs. green) components.

    "now_you_see_it.jpg" -- one of 2 unique shots that I didn't notice I
    had until going through my archives to find examples for this post.
    This and the next photo were taken moments apart. A slight change in
    the camera's auto exposure settings resulted in ...

    "now_you_dont.jpg" -- you'll see that although the exposures are
    nearly identical (f/3.2 vs. f/3.5), the yellow in this one is almost
    completely gone. Much of it due brightness limits, but this doesn't
    account for the yellow completely disappearing except in the very
    small margins right up against the blacks. There should have been at
    least some graduation of yellow in some parts of it, the whites are
    not pure whites. I assure you, I did nothing to these photos except
    resize them for uploading.

    familiar_one.jpg -- to better show this problem, here is a species
    that most everyone has seen in real life, a Tiger Swallow-tail. Note
    how pale the yellow appears. It was nothing like this when the photo
    was taken. This species' yellows can be easily repaired with
    saturation enhancement.

    So ... from these photos, I am somewhat concluding that this is a
    color-space problem in this camera, (which, in reference to the
    discussion about this, I think can be circumvented by using colored
    filters to shift the color-space in the camera and let the auto
    white-balance compensate for it). My guess is that this one unique
    shade of yellow (towards the greener side of the spectrum, rather
    than redder) makes it extremely difficult for the bayer filter and
    software to adequately interpret it.

    Note however, that with this particular camera make and model that I
    have never seemed to experience the problem many others (with other
    makes and models) have reported in the misinterpretted violets to
    blues and vice-versa. Out of thousands and thousands of photos I've
    taken it is only this particular shade of yellow on these few unique
    butterfly species where this even was apparent enough to draw my
    attention.
     
    JeffTaite, Sep 14, 2004
    #6
  7. JeffTaite

    JeffTaite Guest

    I'm glad you mentioned this. It could be a part of the problem that
    I didn't think to consider. If all 3 colors on the sensor are being
    evenly hit with a large amount of IR, then it would indeed change
    the colors to values of grays and whites.

    And since it is appearing on only certain species of butterfly
    wings, that too might lean towards IR explaining the largest part of
    the problem. I now wish I had taken some photos of these butterflies
    in IR only, just to see what if any type of IR was being reflected.

    As mentioned in the post for the sample shots that I uploaded, it's
    a Sony F717, well known for it's IR capabilities. But in general the
    built-in filter compensates for excess IR (as far as I know) when
    not using its Nightshot modes. These few butterfly subjects might be
    proving to be unique cases though.
     
    JeffTaite, Sep 14, 2004
    #7
  8. JeffTaite

    Jeremy Nixon Guest

    Or causing it to overexpose, or color shift.

    It could be ultraviolet, too, of course, but I thought infrared more likely
    because digital cameras aren't very sensitive to UV, so UV usually isn't an
    issue with them. They are very sensitive to IR, though; they all have a
    hot mirror built in, but some are more effective than others, which is why
    I asked which camera you're using. But I don't know about the Sony, and an
    on-lens hot mirror filter is an expensive thing to buy on a hunch.
     
    Jeremy Nixon, Sep 14, 2004
    #8
  9. JeffTaite

    andrew29 Guest

    No, that's not it. Bayer filters have an intrinsically wide gamut,
    wider than most working spaces. However, when the information from a
    sensor array is converted to JPEG, the colour space is transformed
    from that of the sensor to a standard space. This is usually sRGB,
    which is deficient in a number of ways. For example, saturated bright
    yellow is outside the sRGB gamut. (The Adobe RGB space is slightly
    better.) The colour space transformation from your camera's native
    space to sRGB is preserving luminance (or tying to) by reducing
    saturation.

    There's a simple solution to this: shoot in RAW, and convert later to
    a wide colour space. However, that doesn't guarantee you'll be able
    to display that yellow on a monitor, or to be able to print it!

    Andrew.
     
    andrew29, Oct 1, 2004
    #9
    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.