Bayer with NO anti-aliasing (Kodak Pro 14n)

Discussion in 'Digital Photography' started by K2, Nov 28, 2003.

  1. K2

    Rafe B. Guest


    Much of the so-called "detail" that you see with
    a non-aliased image capture is false detail. It
    simply wasn't in the original image.

    No, no, and again: no. Lack of aliasing
    introduces artifacts into the captured image.
    The monitor has nothing to do with it, nor
    the printer that might be used to print this
    image. The artifacts are in the image, and
    cannot be "taken out again" by any sort of
    magic.


    The only time that you can "get away" with no
    anti-aliasing is if the original image produced
    by the lens had no detail beyond the sensor's
    f/2 resolution limit. That can certainly happen --
    for example if the image wasn't properly
    focused in the first place, or there was simply
    no detail to capture (clear blue sky, ripples in
    smooth water, etc.)

    But anything that you'd associate with sharpness
    or detail is *exactly* what would cause aliasing
    artifacts.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 1, 2003
    #61
    1. Advertisements

  2. K2

    K2 Guest

    Actually, it seems to have been sharpened with software, probably from
    a blurred original. Keep reading...
    I don't like it when people use deception to make a "point,"
    especially after claiming MY vision sucks. Here's another photo taken
    with an SD9 that has similar architectural features to your sample.
    Plenty of places where brickwork could causes jaggies, but almost none
    to be seen at normal resolution. Why is that, do you suppose?

    http://img.dpreview.com/news/0209/sigmasd9_sunny/SD9IMG04810.jpg

    Explain why it shows nothing close to the jaggies in your so-called
    typical shot? I'll tell you why: because it wasn't sharpened after the
    fact. Yours is over twice the filesize (with the same pixel
    dimensions) and was probably run through an Adobe filter or two and
    saved at JPEG "100%," which creates an abnormally large file. It has
    the look of a camera-shake blurred image that was sharpened to
    compensate. Just a poor original made into a poorer sample. Not the
    fault of the sensor.

    If you're going to critique something, don't go out of your way to
    find fake or exaggerated samples of its flaws. In my other post, "Yes,
    Foveon pixels are a bit large" I show a legit example of jaggies and
    it's far milder than your claims.

    K2
     
    K2, Dec 1, 2003
    #62
    1. Advertisements

  3. K2

    Rafe B. Guest


    In time, if National Semiconductor can figure out how
    to get decent yields on Foveon sensors, in decent
    sized chips, with decent pixel counts, we may eventually
    concede your point. For now, Foveon is still an
    immature and single-sourced technology.

    The other issue is that Foveon attempts to solve a
    problem that's not much of a problem to begin with.
    Foveon does indeed improve spatial resolution of
    color information, but it just so happens that human
    vision sucks at exactly that. Human vision perceives
    detail based on grayscale (intensity) but our
    perception of color comes with very poor spatial
    detail.


    It is not artifical blurring. It is a "necessary evil" in
    any digital capture system. Precisely the same
    principles apply in digital audio, for example.


    Absolutely, totally, utterly incorrect. This statement
    simply shows that you do not understand what
    anti-aliasing is or why it's needed.

    Look, K2, you really owe it to yourself and to
    the rest of the folks on this forum to educate
    yourself a bit on the matter, before you continue
    to spread complete falsehoods on a topic that is
    clearly beyond your current comprehension.

    I'm not saying that this is an easy matter to
    explain -- it has to do with frequency domain,
    Fourier/Laplace transforms, and the like -- stuff
    that's generally taught to electrical engineers.

    But given that you're quite out of your league
    on this matter, you might at least have the sense
    and/or decency to back off and listen, rather than
    argue with ignorance and sophistry.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 1, 2003
    #63
  4. K2

    K2 Guest

    [quoting and correcting myself here]
    Actually, it was pointed out to me that the lake image with the
    motorboat was post-sharpened and the jaggies weren't the fault of the
    Foveon sensor. After examining it again it's pretty obvious why it has
    supernaturally sharp boundaries; stop doing that and calling them
    originals, folks! Here's the shot in question:

    http://www.pbase.com/image/22816462
    (no EXIF info, overly large filesize)

    I thought I was giving honest criticism but the sensor is proving to
    be even better than I suspected!

    K2
     
    K2, Dec 1, 2003
    #64
  5. K2

    Ray Fischer Guest

    Because everybody knows that 14 divided by 3 is really 3.43
     
    Ray Fischer, Dec 1, 2003
    #65
  6. K2

    Mxsmanic Guest

    I don't have data for the Foveon.
     
    Mxsmanic, Dec 1, 2003
    #66
  7. K2

    Rafe B. Guest


    Good of you to admit that, at least.

    In any case, if you're going to cite parameter
    'X' as an argument against technology A, don't you
    think it's meaningful to know the value of 'X' for
    technology B?


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 1, 2003
    #67
  8. K2

    K2 Guest

    Well said. It's too bad some people seem bent to tear it down before
    it's marketed in other cameras. For God's sake, it's only in the
    infant stages and it's still very good.

    I sure hope this newsgroup isn't browsed by major manufacturer reps as
    a place for educated opinions. The anti-Foveon noise is drowning out
    reasonable critiques at the moment. It's almost like doing a marketing
    survey based on AM radio caller opinions

    K2
     
    K2, Dec 1, 2003
    #68
  9. K2

    Mxsmanic Guest

    If you're going to attempt to discredit others by claiming that they
    don't have hard data, don't you think you should first provide some of
    your own, to avoid being grouped among them?
     
    Mxsmanic, Dec 1, 2003
    #69
  10. K2

    Ray Fischer Guest

    You do not understand aliasing. Monitor resolution cannot fix
    aliasing artifacts. Nothing can.
    Aliasing isn't just jaggies. It also introduces false colors and
    false detail.
     
    Ray Fischer, Dec 1, 2003
    #70
  11. K2

    Rafe B. Guest


    Nope. I wasn't the one making the original point.

    Just suspected that you were running half-cocked,
    as usual. And I was right.


    rafe b.
    http://www.terrapinphoto.com
     
    Rafe B., Dec 1, 2003
    #71
  12. K2

    K2 Guest

    So it's not perfect. But bear in mind that several samples presented
    as evidence of jaggies have been sharpened and are misleading. I have
    gone to lengths to illustrate that in other posts today. I'd like to
    see sample that you personally think has jaggies. You'll find that I'm
    seeing exactly what you see, assuming your eyes are good.
    But Foveon images look much sharper per the same screen real estate.
    That's what matters most to me. They look more "real," even with minor
    jaggies (that go away at high monitor resolutions; you keep ignoring
    that).
    It's only a necessary evil in a Bayer system because it was put there
    to counter the side effects of color interpolation. Notice the title
    of this thread? They didn't use it in that camera and it's quite
    acceptable (and much sharper per unit area) without it. I am not
    ignorant and I know where you're coming from but a few noticeable
    jaggies don't matter to me, especially when they all but vanish at
    higher monitor resolutions. The jaggies are a limitation of the
    monitor far more than the camera.
    Fine, I could "blur" it. But I don't like that soft look, and the
    minor jaggies in a Foveon shot bother me far less than the fuzziness
    and huge image dimensions needed to capture the same detail with
    Bayer. Surely you can relate to that one iota?
    I understand the concepts here well enough, so spare me the arrogance.
    The problem is that you insist digital cameras "must" work only one
    way; they must have anti-aliasing or the Bayer police will take them
    away in the night!

    To me, this is about keeping an open mind and losing dogma. That's how
    breakthroughs are made. If the Foveon bugs you so much, why not email
    its inventor(s) and tear into them? You might learn something yourself
    if you did that. Claiming absolute knowledge on any technical topic is
    a dead end.

    K2
     
    K2, Dec 1, 2003
    #72
  13. K2

    K2 Guest

    Not at all. Upping the monitor resolution effectively brings the
    pixels closer to infinitely small. If you don't think that matters
    you're not the sage you claim to be.
    What size and type is the monitor though? If it's a CRT of 19" or less
    you should be hard pressed to see any jaggies at normal size in any
    Foveon shot. If it's an LCD, they're known for being ruthless with
    stair-stepping.

    K2
     
    K2, Dec 1, 2003
    #73
  14. Huh!? I don't have a 10D. Send me one, and I'll take some snaps.
    Besides that, what is the point? We already know the answer. The 10D
    has higher luminance resolution and little artifacts, the SD-9 has
    higher color resolution and more artifacts. You and K2 prefer the
    latter, others prefer the former.
    As I've said before, my camera is excellent based on what I have seen
    others get out of it. My images is not worthy of publication, though.
     
    Asbjørn Bjørnstad, Dec 1, 2003
    #74
  15. K2

    Mxsmanic Guest

    Case closed, then.
     
    Mxsmanic, Dec 1, 2003
    #75
  16. I'm not talking about the flash reflection, The pattern goes all the
    way up to the top of the image. If it is some sort of diffraction, I'm
    amazed that George managed to tilt the camera just enough to move the
    pattern over one pixel for each time the pattern appears.
    That's where opinions differ, I guess.
     
    Asbjørn Bjørnstad, Dec 1, 2003
    #76
  17. Even infinitely small pixels do not eliminate aliasing. The fact that
    you think it does just means that you don't know what aliasing is.
    Aliasing is not jaggies, though jaggies are one particular manifestation
    of aliasing.
    Jaggies are not the only visible artifact of aliasing.

    Dave
     
    Dave Martindale, Dec 1, 2003
    #77
  18. K2

    Guest Guest

    You just do not listen do you?

    Most of the people who are being anti-foveon have said that they think
    foveon is a great idea but it isn't ready yet.

    Stop lying to suit your agenda.
     
    Guest, Dec 1, 2003
    #78
  19. K2

    Larry Lynch Guest

    You have used the correct word in your statement.

    "reasonable"

    When the Foveon is mature enough to take "reasonably" good pictures, ie
    good color and good resolution in ONE PICTURE, (I have seen both, but
    not in the same picture) people will argue "reasonable" points for and
    against it.

    As it stands, the Foveon is for people who want to be on the leading
    edge of the technology without caring if it works well.

    Foveon is an "INFANT" in the industry, and most photographers want a
    fully mature camera.

    When the developers get the "bugs" worked out so that flesh-tones shot
    in daylight dont look Jaundiced, or flesh-tones in shadow dont look like
    alien skin, they will have a camera I'm interested in looking at
    (again).

    You should take the time you waste being a Foveon "zealot" and spent it
    comunicating with Sigma, and tell them what needs to be fixed, instead
    of telling a bunch of people who know better, that the Foveon is GREAT!
     
    Larry Lynch, Dec 1, 2003
    #79
  20. K2

    JB Guest

    You're completely misunderstanding what aliasing is, and what the
    anti-aliasing filter is for. It's not about "jaggies", it's about
    information with a spatial frequency greater than half the sampling
    frequency being aliased to lower freqencies within the nyquist limit.

    This doesn't just lead to jaggies - it can lead to gross low-freqency
    artificial signals being introduced into the image which *cannot* be
    removed after without losing information (post-sampling, there is no
    way of knowing whether it's real or an aliasing artefact). That's why
    you need an anti-aliasing filter before sampling - to ensure that
    there is no spatial frequency content above half the sampling
    frequency present in the signal by the time it hits the sensor.

    This is true for *any* sampling system - bayer, foveon or anything
    else. It's well understood, mathematically proven, and not disputed
    by anybody with any understanding of digital signals and systems.

    /JB
     
    JB, Dec 1, 2003
    #80
    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.