Simulating an orange filter in Photoshop

Discussion in 'UK Photography' started by Al Treacher, Aug 22, 2003.

  1. Unfortunately, the colour image alone doesn't have all the information
    needed to do this. The original light from the subject consists of a
    complete spectrum of light wavelengths at each pixel. The colour image
    converts that spectrum into just three values: RGB.

    In the original, there might have been yellow that is pure spectral
    yellow, and other yellow that is a mixture of red and green (with no
    yellow). An orange filter can separate the two. For that matter, B&W
    film might separate the two with no filter at all because it's more
    sensitive to red than green, unlike the eye that is more sensitive to
    green than red.

    If your camera captured 10 or 30 or 60 different wavelengths at each
    pixel, you could also easily tell the two yellows apart, and you could
    simulate the effect of an orange filter, or the effect of a particular
    B&W film's sensitivity. But if you use a RGB camera designed to have
    the same colour vision as the human eye, the two colours end up with the
    same RGB value, and there's no way to separate them later.

    The fundamental problem is that your eye performs a many-to-one mapping:
    many different spectra all appear as exactly the same colour. Colour
    film and digital cameras perform as close as possible to the same
    mapping, so they reproduce colours the way your eye sees. This mapping
    cannot be reversed to later distinguish between the spectra. But a
    filter applied before the sensor, while the image is still composed of
    spectra rather than RGB values, can make the two spectra appear

    Dave Martindale, Aug 28, 2003
    1. Advertisements

  2. OK, so let us suppose that our 'perfect' digital camera has three RGB
    sensors that happen to have sensitivity ranges identical to those in my
    eye. (I realise that each person's equipment may differ from others).
    Then the effect of the filter in restricting light intensity falling on
    each sensor is the same as in my eye. (Now that too may be different
    from the equivalent sensitivity of pan film, but ignore that for the
    time being.)
    So in my hypothetical case, the digicam will do the same as the orange
    filter does to my eye. The bit I don't understand with all this arguing,
    is why no one has pointed out that the film sensitivity doesn't match
    the eye sensitivity anyway, and so the use of an orange filter is highly
    *artifical* anyway.
    True. Once information is lost, it is lost. What we are trying to
    simulate is the effect that the orange filter loses in the pan film
    You omitted to say that B&W film doesn't; having continuous mapping
    across the wavelength which is curtailed by the filter.

    Incidentally, since streetlight sodium vapour lighting (still used in
    some parts of the world) is effectively monochromatic, what does that
    look like to colour film and digicams?

    Must look some out and try.

    [The reply-to address is valid for 30 days from this posting]
    Michael J Davis
    Some newsgroup contributors appear to have confused
    the meaning of "discussion" with "digression".
    Michael J Davis, Aug 28, 2003
    1. Advertisements

  3. Al Treacher

    Peter Ashby Guest

    What you're overlooking here is that once you postulate this magic
    mapping function which is *potentially different for every pixel* then
    you're talking about something which has the complexity of just
    reproducing the picture you want regardless of the digital image.[/QUOTE]

    That is one extreme of the complexity curve yes. But in this thread we
    haven't established that this is the case, especially if the process is
    two step with a colour to B&W transition in the middle. As others have
    said in trying to interpose between the extremes, it is a simulation
    that is required, not the real thing.

    Peter Ashby, Aug 28, 2003
  4. That's right. A filter in front of a camera behaves about the same as
    putting the same filter in front of your eye.

    The fact that pan film doesn't have the same response as the eye
    probably means that you can't accurately simulate pan film with a colour
    digital camera, even when no filter is involved, since the colour
    camera's response is designed to match the eye as much as possible.
    The colour digicam looking through an orange filter will capture about
    the same image as your eye sees when looking through the orange
    filter. The argument has been about whether you can get the same
    effect with a colour digicam using no filter, followed by some set of
    operations in Photoshop.

    Someone using an orange filter is probably using it because they want
    artificial-looking (non realistic) recording of the scene. The question
    is, can we get an equivalent effect without using any filter?
    Sorry, I can't figure out what that last sentence means. Can you
    I can't figure out what you mean here either. B&W film does have
    continuous sensitivity across the visible light range, though it's
    different from that of the eye. A filter changes the shape of the
    sensitivity curve, sometimes drastically. But what are you getting at?
    Probably some yellowish colour. The red and green-sensitive CCD cells
    or film layers will be exposed, with little or no response in blue.
    What the film or camera records is really a monochrome image, though,
    representing reflectivity at that one yellow sodium wavelength.
    (Actually two spectral lines, but too close together to distinguish).

    By the way, there are two rather different types of sodium lighting.
    Low Pressure Sodium (LPS) lighting really is nearly monochromatic.
    High Pressure Sodium (HPS) has a much broader spectrum that does show
    colour in objects, though it's still far from realistic. HPS is much
    more common anywhere I've been in Canada.

    Dave Martindale, Aug 28, 2003
  5. Al Treacher

    Tony Spadaro Guest

    Tony Spadaro, Aug 28, 2003
  6. Al Treacher

    Neil Barker Guest

    Well, there's one thing that's good about this incredibly tedious and
    torturous thread - sure as hell cures insomnia....
    Neil Barker, Aug 28, 2003
  7. My apologies, really, I was doing some thinking aloud, so hadn't really
    spelt it out. Nothing to disagree with. I was thinking of the mapping of
    the sensitivity of pan film to a 3-colour gamut. (After all pan film was
    designed to match the sensitivity of the eye.) The orange filter trims
    that and changes the RGB intensities.
    I'm still puzzling over the reason why the mapping of the pan film with
    and without the orange filter, cannot be represented by a relative
    reduction in the RGB components. After all, the digicam can adjust for
    colour temperature, surely an orange filter is a more extreme
    (I did know that actually, I remember looking at the spectrogram in a
    school physics lesson in the 1950s.)
    Yes, I also know that, although I wasn't sure if the LPS is still in use

    It's been a fascinating discussion and I shall, when work pressures
    reduce somewhat, set up some tests to see if the differences can be

    Michael J Davis
    Personal email replies may be made to
    I have a photographic memory;
    but then I forget to load the film
    Michael J Davis, Aug 28, 2003
  8. Al Treacher

    Tony Spadaro Guest

    I've been taking photographs for 38 years and have used many many filters.
    I've been using Photoshop for about 5 or 6 years now and have taken the time
    to learn how to use it. I've never at any time said that a polarizor can be
    simulated in Photoshop. Don't put words in my mouth or claim to know my
    history. It makes you look like a real jerk.

    home of The Camera-ist's Manifesto
    The Improved Links Pages are at
    New email - Contact on the Menyou page.
    Tony Spadaro, Aug 28, 2003
  9. It comes out a lot redder on TV cameras and digital still cameras than
    it looks to the eye.

    Roderick Stewart, Aug 29, 2003
  10. It's been explained over and over again here how a 3 colur representation
    of a continuous spectrum lacks sufficient information to simulate all
    possible spectral inputs, particularly artificial ones such as filters.

    Colour temperature adjustment in an electronic camera (TV or stills) is
    doen by adjusting the relative gains of the R G and B cnannels. If there
    is sufficient range, it will be possible to make the white point any
    colour that can be produced by mixing the 3 primaries, and the perceived
    colour of the orange filter may well fall in this range. This will enable
    you to simulate the perceived colour of a white object seen through the
    filter. However, a real scene will contain real objects which can have
    spectral characteristics that rise and fall at *any* wavelengths, not
    necessarily coinciding with the RGB colour separation filters in the
    camera (or film), so the filter will affect the proportions of R G and B
    produced by these objects differently depending on their spectra. If you
    adjust R G and B in the camera, the proportions of R G and B will all be
    adjusted in the *same* ratio for *all* objects in the shot, which is not
    the same effect.

    Roderick Stewart, Aug 29, 2003
  11. If I recall correctly, that is my experience, and nicely demonstrates
    the discussions here. Thanks.

    [The reply-to address is valid for 30 days from this posting]
    Michael J Davis
    Some newsgroup contributors appear to have confused
    the meaning of "discussion" with "digression".
    Michael J Davis, Aug 29, 2003
  12. Al Treacher

    Lionel Guest

    that was not the point I read Lionel as making. It was one of
    discrimination. If our eyes were as good as a spectrophotometer then why
    did we invent spectrophotometers? The human eye is good but it is
    subject to too much variation. For eg we know that the colour of
    daylight varies during the day, but we didn't know that before we could
    measure it, either with a camera or a more sensitive instrument. Our
    eyes percieve colour brightness on a relative scale, not an absolute
    one, a spectrophotometer works on an absolute scale. That is my reading
    of Lionel's point.[/QUOTE]

    That's kind of the corner I was arguing from, but upon reflection, I've
    realised that Dave & Rod's points are correct. The annoying thing about
    my misconception is that I should know better, as I'm used to applying
    similar techniques in electronics. (My excuse is that mixing light of
    different wavelengths doesn't work the same way as mixing electrical
    signals of different wavelengths.)
    Lionel, Aug 29, 2003
  13. Al Treacher

    Martin Brown Guest

    It also depends somewhat on the choice of emulsion. On some fast slide
    films like Fuji 1600ASA it comes out almost lemon yellow. But most slide
    films render it somewhere on the orange side of the true colour.

    It is harder to judge with print film since auto colour balance on
    machine printing attempts (and fails) to remove any obvious colour

    My faithful old Kodak Dc120 renders the colour of sodium light pretty
    accurately. And I expect some other models do as well.

    Martin Brown, Aug 29, 2003
  14. Actually it does behave the same way, pretty much. The big difference
    is that the only sensors we have for measuring light measure power, not
    amplitude, and we usually lose any phase information. But the mixing
    of different frequencies is the same - they are independent in most

    Dave Martindale, Aug 29, 2003
  15. Al Treacher

    Frank ess Guest

    May I digress?

    I have had prints from Kodacolor negatives made in a fluorescent-lighted
    office returned with an overall green cast. I told the subjects that the
    prints were realistic, that their eye saw the green but their mind
    eliminated it from their perception. Was I wrong?

    Frank ess
    Frank ess, Aug 29, 2003
  16. Al Treacher

    Charlie D Guest

    No. You were correct, but your prints were still ugly.
    Charlie D, Aug 29, 2003
  17. You can't.
    Michael Scarpitti, Aug 29, 2003
  18. Yes, the sensitivities of R G ad B sensors in colour photographic systems
    are *designed* to match the human eyes closely as possible, with the
    intention that photographed objects or scenes match what the eye would see.
    It's one reason, but not the one we were talking about. Even with a colour
    photographic system, a filter in front of the camera, or in *front* of the
    separation filters that separate the light into R G and B components as
    part of the photographic process, can produce effects that are not
    reproducible by manipulation of the R G and B components *after*

    A monochrome photographic system, particularly if it uses film, will
    probably have an overall spectral response that is nothing like the eye,
    and the very act of representing the visual world in shades of grey is a
    departure from reality anyway, so talking about realism here would be
    academic. A filter in front of the lens here would also produce effects
    that could be different from those attainable by manipulating R G and B in
    a colour photo and then converting to monochrome, but since the result
    bears no relation to reality anyway, the acceptability of an effect is
    purely a matter of your opinion, which can be anything you want it to be.

    Roderick Stewart, Aug 30, 2003
  19. Probably. Eyes are different form cameras because they are attached to
    brains, and brains have previous experience and expectations. If you look
    at familiar objects with your eyes, you will see what you expect those
    objects to look like, no matter what they actually look like. It's a bit
    glib to say "the camera cannot lie", but probably more accurate to say
    "the camera will take what it sees literally".

    Roderick Stewart, Aug 30, 2003
  20. Al Treacher

    Frank ess Guest

    That's one for and one against.

    Anyone else?
    Frank ess, Aug 30, 2003
    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.