JPEG 2000: which camera will support it? When?

Discussion in 'Digital Photography' started by John Faughnan, Jan 28, 2004.

  1. John Faughnan

    Lionel Guest

    You're one one of the "other loons" that I referred to a different post
    on the same subject.

    HTH, etc.
    Lionel, Feb 5, 2004
    1. Advertisements

  2. John Faughnan

    Alfred Molon Guest

    You just made it into my killfile.
    Alfred Molon, Feb 5, 2004
    1. Advertisements

  3. John Faughnan

    Martin Brown Guest

    I hope you are joking.

    This will only show you the reflectivity of the ink system used by your
    printer when it tries to emulate the visual appearance of a solar
    spectrum. And it won't even do that particularly well.

    If you want to characterise a camera's filters the simplest test at a
    consumer level is to photograph the reflection of a solar spectrum off a
    silver shovelware CD. You will likely be very surprised by the result.

    Cokin do a 2 point star diffraction grating that also does a reasonable
    spectral dispersion for empirical testing of filters.

    Martin Brown, Feb 5, 2004
  4. Martin,

    thanks a lot! As usual, things are more complex than one first

    Hans-Georg Michna, Feb 5, 2004
  5. Thanks for the good information! Interesting.

    Hans-Georg Michna, Feb 5, 2004
  6. John Faughnan

    Lionel Guest

    If preddy isn't already in there, that's just fine with me.
    Lionel, Feb 5, 2004
  7. John Faughnan

    jpc Guest

    Working out a way to measure the Bayer filter curves is something I've
    been thinking about doing for a year or so. I was just curious how it
    could be done cheaply and easily. And if you don't mind me asking why
    are you interested?

    jpc, Feb 6, 2004
  8. Of course: The Bayer RAW is missing 2/3s of the natural color spectrum
    *which the human eye is sensitive for*!

    And tell that to the 'experts' here which always like to speak about
    sampling and aliasing artifacts of the Foveon sensor:
    The art of Bayer sensing is a *major* violation of basic sampling
    assumptions, rendering all further results useless.
    The Foveon sensor gives a *much* better approximation to native

    Guido Vollbeding, Feb 6, 2004
  9. John Faughnan

    Lionel Guest

    Do you realise that your response related in almost no way to Michael's
    comment? (Quite apart from it being untrue, of course.)
    Which ones? - Please quote the actual rules concerned, eg; Nyquist, etc.
    Then why does their output look so good?
    Except that we don't /need/ a "much better approximation to native
    photographs", because mosaic sensors pick up all the information that
    the eye needs it to pick up. Sure, it'd be nice to have three colours at
    every sensor pixel, & there are lots of nice things we could use that
    extra data for, but we don't actually /need/ it. And if it comes at the
    expense of a clunky camera, buggy colour & sub-average lenses, then no
    thanks, I'll stick with something that I know can produce the results I
    Lionel, Feb 6, 2004
  10. The basic assumption of sampling theory is that every single sample
    sufficiently covers the range of susceptible values.
    The analog-digital conversion must result in quantized values which
    cover the susceptible range.
    This digital quantization of values is, beside resolution (digital
    quantization in space), one of the two basic sampling parameters.

    If the quantization range of single values is as limited as in the
    Bayer case, the other parameter, resolution in space of multiple
    values, does not make sense. That's why the 'MP' rating of Bayer
    sensors is wrong and misleading.
    Nice for you, but not for me...

    Guido Vollbeding, Feb 6, 2004
  11. John Faughnan

    Ron Hunter Guest

    Now you are revealing your ignorance. The Bayer system is modelled
    after the way in which primate vision works. Plonk.
    Ron Hunter, Feb 6, 2004
  12. jpc,

    pure curiosity. I'm wondering how much emphasis they put on high
    color resolution and precision.

    The human eye has two different kind of interspersed sensors. It
    uses high resolution color sensitive cells and a sprinkling of
    black and white sensitive cells, which are much more sensitive
    and are useful mainly at night. Interestingly we have no black &
    white cells in the focus spot, so we can't focus very well on
    anything at night. In fact, you may have to look elsewhere to
    see something more clearly at night.

    Maybe one day we will have digital cameras with two kinds of
    interspersed sensors, one bunch for high quality, high light
    situations and a thinner grid of high sensitivity, lower quality
    sensors for low light situations. I would appreciate this and
    gladly dispense with the flash.

    I think there is at least one camera that already works that
    way, but I have no details. I also remember that at least one
    Sony camera has a special low light setting, but I don't know
    whether it actually has different kinds of sensors.

    Hans-Georg Michna, Feb 6, 2004
  13. John Faughnan

    jpc Guest

    Not at all. Since the original poster had asked a legitimate question
    and all he was getting in the post at the time was a continuation of
    one of the sillier flame-wars I've seen I attempted to answer his main
    question--do the transmittance of Bayer filters overlap. Which they
    As a rough and ready appoximation this method worked reasonably well
    And yes I hadn't thought of using a cd as a diffraction grating and
    yes your method worked even better. With it I discovered the blue
    filter does block the middle red completely instead of passing a few
    percent like I stated earlier. It also passes the deep red and
    infrared, something I'd already known from my infrared photography.

    The big surprise what how much blue the green filer passes,
    Both the red and blue filters look like they are single layer
    interference filters. The green filter, on the other hand looks like
    it is a multilayer since it transmittance peaks in the green than
    drops about 30 % and stays pretty constant throughout the blue.

    Why I don't know. The camera--an oly 3020Z --produces accurate colors
    so it must be deliberate. If anyone knows something about how the
    bayer calculations work jump in and educate us.

    jpc, Feb 6, 2004
  14. John Faughnan

    jpc Guest

    My guess is that the bayer calculations won't work unless there is
    some overlap
    You have to trade resolution for S/N and dynamic range. And since it
    is a square root function the resolution goes bad fast.
    Fugi came out with a fancy sensor last summer. No one in this news
    group seemed that impressed. The one time I played with it I thought
    it handled nicely but I never had a chance to take any images.

    I also remember that at least one
    With the better Sony cameras and camcorders you can flip the ir filter
    out of the light path. The sensor stays the same.

    jpc, Feb 6, 2004
  15. SNIP
    Mostly yes, but it (and the G and B filters) are not strict band pass
    filters, there is some overlap in both G and B. The amount is determined by
    the filters themselves, so only the manufacturer knows.

    Bart van der Wolf, Feb 6, 2004
  16. A 'monochrome' sensor does that, so your assumption needs some refining...

    Bart van der Wolf, Feb 6, 2004
  17. Another mis-conception seems to be here. Overlap is not a bad thing. It
    makes for a more accurate reading as long as the overlapping portions are
    not linearly related.
    Gherry Bender, Feb 7, 2004
  18. Yes, a 'monochrome' sensor does that, but in a Bayer sensor the color
    *prevent* that. Thus the Bayer system is a false sampling system.
    It is to notice that the 'experts' speaking about sampling theory terms
    like aliasing and Nyquist don't see that the Bayer system violates the
    basic sampling assumptions and thus only produces artifacts.
    Their theory makes them blind for the most obvious facts.

    Guido Vollbeding, Feb 7, 2004
  19. John Faughnan

    Lionel Guest

    That comment is so vague as to be nonsensical. What kind of sample? What
    range? And what the hell is a 'susceptible value', exactly?
    None of the above gibberish makes any sense. If you're going to try to
    sound scientific, please try to use the correct terminology, rather than
    making up your own.
    Lionel, Feb 8, 2004
  20. John Faughnan

    Lionel Guest

    You still haven't explained this strange theory of yours in actual
    scientific terms yet. So far, you haven't made any sense.
    Lionel, Feb 8, 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.