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

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

  1. The missing lines in your crop is a reflection of your flash, fool.
    Aluminum is covered with tiny alumina crystals that act as a diffraction
    grating. You're shining a point source of light at it so you're seeing
    a color pattern back.

    The only aliasing I see in the full-size photo is in the horizontal
    metal strips on the right side of the photo. The blur at the center of
    the test patterns is also slightly abnormal. Overall it's an
    insignificant problem. There are patterns in the finest details but not
    in the lower frquencies. The only way to fix the aliasing in the tiny
    details would be to eliminate the tiny details. You're much better off
    the way the photo is.
    Kevin McMurtrie, Nov 29, 2003
    1. Advertisements

  2. There are a few Sigma lenses that are sharp as well, although the quality
    spread is large, the must do aweful things to the images.

    Bart van der Wolf, Nov 29, 2003
    1. Advertisements

  3. Better yet, look at his 10D image of the same chart... oh, wait, no, he
    refuses to post it because the 10D Is so blurry you can hardly see than it
    has poor B&W resolution and absolutely abysmal color resolution. Here's
    what you could to see expect if he had any pride in his choice of cameras:

    It really takes a shameless person to keep posting here after refusing to
    post their own images, based on a "My camera is too bad" excuse.
    George Preddy, Nov 30, 2003
  4. K2

    Azzz1588 Guest

    You describe yourself well with this statement.

    You are already a proven liar, and dishonest through and through...

    "Only a Gentleman can insult me, and a true Gentleman never will..."
    Azzz1588, Nov 30, 2003
  5. I find the jagged unpredictable rainbows too distracting, though the detail
    level is better than other Bayers. It provides solid evidence for the need
    to heavily blur Bayer images. It would probably be competitive with the SD9
    if it were downsized to a non-interpolated 3.43MP, the number of RGB sets
    present on the 14n's sensor--same as the SD9.
    It uses a digital, not physical, blur filter. Still has lots of rainbow
    artifacts though, in detail areas where the digital filter cannot
    effectively tell the difference between actual optical detail and
    moire/noise/artifacts. Eyes are way ahead of computers in this respect.
    With 25% color information per pixel, that is interpolated the rest of the
    way. This is identical to the SD9 starting out with 100% color information
    per pixel, and interpolating another 75% of pixels as inbetweens. In fact,
    the SD9's interpolation method is much more accurate, because the baseline
    upon which estimates are built is known.
    George Preddy, Nov 30, 2003
  6. Actually Bayer does require a blur filter because unlike Foveon and film, it
    is not a full colro capture system, so you get henious artifacts without
    blurring the image rather badly. A good explanation with examples, here:

    But you are right, with a true full color capture sytem like film/Foveon,
    there is no need for a blur filter.
    George Preddy, Nov 30, 2003
  7. Film doesn't need a blur filter becasue it is a full color capture system,
    same as Foveon.

    And lets see, I'm up to about 12,000 digital pics with a about a hundred GBs
    of video, how cheap is film again?
    George Preddy, Nov 30, 2003
  8. Then you'll absolutely love this full sized 10D image, you can hardly see
    any diagonal lines at all...
    George Preddy, Nov 30, 2003
  9. K2

    Mxsmanic Guest

    This blurring, FWIW, is also partially the source of the "grainless"
    quality of digital images.
    Eyes have several fixes for this. The resolution is very high, which
    limits the visibility of aliasing. The distribution of photosites is
    randomized, which limits aliasing of regular patterns (the chances of a
    random pattern hitting the retina in a way that produces visible
    aliasing are very low). Finally, there's a lot of signal processing of
    the raw image to eliminate aliasing and other artifacts, such as
    chromatic aberration, which is pretty bad in the human eye (but it is
    predictable and can be removed digitally).

    Anyway, I'd personally like to see the Foveon perfected to replace
    matrix filters, because I'd very much like to see a full three colors
    per pixel in digital photography. One of my personal barriers to going
    digital is the current loss of color (and overall) resolution due to
    matrix filters, not to mention the loss of light sensitivity (diminished
    by a factor of three by the matrix filter).
    Mxsmanic, Nov 30, 2003
  10. K2

    Mxsmanic Guest

    I have scans from Kodak Portra 800 that show less noise (and better
    resolution) than that image. Is it really from a digital camera??
    Mxsmanic, Nov 30, 2003
  11. K2

    Nils Rostedt Guest

    The spiky histograms of this dawn shot, and the prominent dust spots,
    indicate that the photographer missed the exposure and had to do a
    significant levels adjustment. Look at the previous shot in the same
    gallery, taken a few minutes earlier. Its noise level is very low for such a
    grey and foggy subject, and the histograms are also much smoother. These
    image selections from GP are designed to discredit the 10D, remember.

    The noisy shot is still less noisy at ISO 400 than many of my ISO 100
    negatives, where just a slight underexposure is enough to make the scans
    Nils Rostedt, Nov 30, 2003
  12. You do not lose any color fidelity using Bayer.
    You might get colored artefacts. But the overall
    color is just as good as a theoretical RGB
    sensor camera.
    You do lose some resolution. But compared to what?
    A 6 Mpixel Bayer camera is compared to what? Nothing
    really. So - as long as no one makes a true RGB sensor
    camera, the resolution of Bayer is as good as it get.
    (NOTE - The Foven is not a true RGB camera - and it
    has no anti alias filter - both shortcomings making
    it not a highest quality imager.)
    Nope - you do not lose any sensitivity by having fewer
    sensors of the same size. You lose resolution. And
    in the Bayer case it is more complex than that, you lose
    no sensitivity and you lose very little resolution.
    The Bayer matrix filter is rather clever constructed.

    The reason why you don't want to use Bayer cameras
    is on the psychological plane.

    Roland Karlsson, Nov 30, 2003
  13. Incorrect.

    Film does not need anti alias filter, because it is not sampling.

    Foveon needs an anti alias filter, because it is sampling.

    I take by far more pictures now. Last I checked I had
    passed the cost of my camera in film only. Now, I am not
    really taking as many pictures as you - so if I buy
    an expensive DSLR with some lenses it may take some
    time to motivate it with economy :) Have to use the fun
    factor then. Harder to motivate to my wife though :)

    Roland Karlsson, Nov 30, 2003
  14. K2

    Mxsmanic Guest

    You lose a great deal of fidelity, unless you are shooting scenes that
    consist entirely of a single color.
    That would also be true with just three pixels, instead of three
    million. But it wouldn't be very useful for taking pictures.
    Compared to a sensor system that has all three colors for each pixel.
    A three-sensor system, or any system that captures all three colors for
    each pixel.
    The Foveon sensor qualifies, as do tri-CCD systems and scanning backs.
    Why not?
    The matrix filter removes an average of 2/3 of the light reaching each
    photosite, so you lost more than 1.5 stops due to the matrix filter
    There's nothing clever about it at all. How _else_ would you construct
    a matrix filter? There aren't a million different ways to do it.
    Mxsmanic, Nov 30, 2003
  15. K2

    Rafe B. Guest

    If you say so. And what are the comparable statistics in
    a Foveon sensor for light reaching the respective red,
    green and blue sensing layers?

    rafe b.
    Rafe B., Dec 1, 2003
  16. K2

    K2 Guest

    I think a lot of your understanding is technically correct but
    pragmatically not as important as you make it to be. I may not be
    quite the technician that some of you are, but I don't think lack of
    anti-aliasing is a way of "cheating" to make a sharp photo. It just
    requires a new way of seeing things. I would rather work toward
    smaller Foveon pixel sizes than larger sensors and the same old
    The finer the resolution of a monitor, the less of a problem that
    becomes. I'm not saying it's non-existent altogether, just that it's
    tolerable most of the time, especially in nature shots that don't have
    hard, contrasty straight lines.

    K2, Dec 1, 2003
  17. K2

    K2 Guest

    Are you reading from a script or something? Above I am only asking you
    how you run your monitor - so I can see how exaggerated the jaggies
    might look to you. For someone who claims to be reasonable you need to
    be less shrill.

    K2, Dec 1, 2003
  18. K2

    K2 Guest

    For the last time, I'm not George.

    K2, Dec 1, 2003
  19. K2

    K2 Guest

    My vision is just fine. The biggest difference is that you people are
    bashing a new technology that could easily supercede Bayer with more
    refinement. I prefer to see its good points and not get hung up on
    Some of them show it more than others, but I'm willing to accept that
    rather than having the whole thing blurred artificially. I consider
    anti-aliasing even more of an artifact than lack of it. I could always
    anti-alias a Foveon shot later if I wanted.
    There is clearly something "off" with that shot and you must know it's
    far harsher than a typical Foveon image. It looks like it was
    artificially sharpened after the fact. If the poster of the image is
    reading this, please explain. Until it fully downloaded, I almost
    thought someone had pulled that double-size scam like they did with
    the cat. Trying to make it seem more pixelized than it really was.

    Here is suspicious EXIF data from that image:

    Software - Adobe Photoshop CS Windows
    DateTimeOriginal - 2003:10:15 18:10:15
    DateTimeDigitized - 2003:11:27 19:47:07
    CustomRendered - Custom process
    I'll take some artifacts (not the hugely exaggerated ones above) over
    mosaic color interpolation and blurring. If I wanted an image to be
    softer I could just anti-alias it after the fact. Not an earthshaking
    problem. Why not talk about the amazing feat of getting full color at
    each pixel instead of falling back on old Bayer limitations? Make a
    Foveon pixel small enough and any harsh edges will all but vanish.
    It's a work in progress.

    K2, Dec 1, 2003
  20. K2

    Rafe B. Guest

    Your concern about monitors is utterly irrelevant
    to the issue of aliasing.

    My film scans of 645 film are typically around
    8500 x 6500 pixels. My screen is set to 1600 x 1200.

    rafe b.
    Rafe B., Dec 1, 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.