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

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

  1. The 14n has, interestingly, the ability to sense 3.43M full color pixels
    (the number of complete RGB sets). It is interpolated to 13.72MP, a lot
    like another camera we know.
    George Preddy, Nov 29, 2003
    1. Advertisements

  2. It's common knowledge for those who know what they're talking about. It is
    also easy to see in unfiltered images. Discrete sampling results in visible
    aliasing if significantly modulated frequencies exist above the Nyquist
    Aliasing cannot be removed without removing detail. The anti-aliasing you
    seem to be talking about is AA *created* by resampling for
    printing/displaying at lower frequencies than present in the image. That
    will only add aliasing, or at best not make it worse.

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

  3. I believe I tried that, and the (maximum lossless compression) PNG came out
    a little bigger, but anyway they're both tolerable sizes.

    Bart van der Wolf, Nov 29, 2003
  4. Jaggies is not a big deal if the pixels are small enough, but aliasing is.
    Look at this example:
    (Crop from: )

    From what I can tell the effects are about 80 pixels high. And they
    are visible for me until I's so far from the monitor that the two
    rightmost lines merges into one. In photoshop not even a 5 pixel
    gaussian blur will get rid of it.

    What kind of printer do you have?
    I got an i805 and actually did a A4 print of this picture (So I need a
    life, what else is new.) and at the distance where the aliasing
    "disappear" I am no longer able to see the circles... It's just a grey
    Asbjørn Bjørnstad, Nov 29, 2003
  5. K2

    JPS Guest

    In message <>,
    Nonsense. It will be worse printed, because ink on paper bleeds less
    than light on a CRT, and the "pixels" are squarer.
    JPS, Nov 29, 2003
  6. K2

    JPS Guest

    In message <bq9rtg$ngv$>,
    The 14n pictures are not perfect, but they look much better to me than
    the SD9 pictures. The aliasing is less severe in the 14n, because each
    lens has to deliver more resolution to produce the same aliasing. I'm
    sure there are microlenses as well.

    The 14n images are not interpolated to 13.72mp; they are captured at
    13.72M pixels, with full color interpolated from partial color readings.
    JPS, Nov 29, 2003
  7. K2

    JPS Guest

    In message <>,
    He's talking about resolutions way beyond 3.43MP.
    I can't enjoy looking at photos where the mortar in a brick wall snaps
    to an obvious grid in some areas, and disappears completely in other
    areas. If that doesn't bother you, you are too visually obtuse to
    comment on image quality.

    The SD9 is a poor imager.
    JPS, Nov 29, 2003
  8. K2

    JPS Guest

    In message <>,
    No, printers don't blur that much on coated or photographic paper. You
    are fooling yourself.

    SD9 images look even worse in print.
    JPS, Nov 29, 2003
  9. K2

    Mxsmanic Guest

    Where's the problem?
    I had to look for it to see it, and even then, I'm not sure that the
    minor effects I'm seeing are what you have in mind.
    Mxsmanic, Nov 29, 2003
  10. K2

    Mxsmanic Guest

    Use an image-capture device that has higher resolution. Anti-aliasing,
    as I've said, is just the poor man's substitute for high resolution.
    Film doesn't require it because the resolution is so high (and because
    film grains are randomly positioned and slightly variable in character,
    which helps mask any aliasing even if it does occur).
    Mxsmanic, Nov 29, 2003
  11. K2

    JPS Guest

    In message <3fc88fa1$0$1501$>,
    I wonder what an SD9 image would look like if you could attach a Canon
    300mm f2.8L IS (sharpest EOS lens available). I have its little
    sibling, the 300mm f4L IS, which is supposed to be less sharp, and I get
    down to 52 pixels in your spoke-wheel test with it - get this - with
    both the Canon 1.4x and Tamron 2x converters *stacked*, with the main
    lens wide-open at f4. Imagine the aliasing possible on such a lens by
    JPS, Nov 29, 2003
  12. K2

    JPS Guest

    In message <>,
    In the pixel spacings of current sensors, anti-aliasing is totally
    necessary. There are lenses with 3x the resolution of current sensors.

    When we get to about 100mp, then we can start talking about leaving the
    filters out, assuming no revolution in lens design.
    JPS, Nov 29, 2003
  13. K2

    JPS Guest

    In message <>,
    No, that's your defense mechanism for dealing with the fact that many
    people are questioning your vision.

    Your vision sucks, and your comments about image quality are totally
    worthless. SD9 images have bogus horizontal and vertical pixel-edge
    emphasis that screams out, "I'm made of pixels, dammit!".

    Here's one taken with sharp optics:

    An otherwise nice picture that has gridding effects in the plants, where
    there is no horizontal and vertical detail in reality. On the fine
    shingle and brick patterns on the small building to the left, some areas
    snap the mortar and shingle shadows to pixel edges, giving perfect 0 and
    90 degree edges where they don't belong. In other areas, the mortar and
    shadows fall between pixels and the high-frequency content is
    attenuated, causing alternating patches of sharpness and blur.

    The SD9 (or foveon without proper filtering) is a poor imager.
    JPS, Nov 29, 2003
  14. Along the right edge of the metal trim there are two black lines that
    goes from very sharp to almost invisible and back to sharp and so
    on. It gives the impression of lousy workmanship on the part of the
    carpenter as it looks like the trim isn't flat against the wall.
    Minor or not, it can be avoided with an antialias filter. And I bet it
    pissses off the carpenter.
    Asbjørn Bjørnstad, Nov 29, 2003
  15. K2

    Rafe B. Guest

    That is absolutely false. Aliasing produces gross
    low-frequency artifacts -- no computer magic or
    sleight-of-hand will remove aliasing artifacts.

    You're dreaming K2. Quite out of your depth.
    Quit while you're ahead.

    rafe b.
    Rafe B., Nov 29, 2003
  16. K2

    Rafe B. Guest

    Film doesn't need it because it doesn't qualify as continuous
    sampling. And in fact film's resolution isn't that much better than
    the best digital capture. The advantage of film of course is that
    it's cheap and can be extended to large areas relatively easily.

    rafe b.
    Rafe B., Nov 29, 2003
  17. K2

    Rafe B. Guest

    Why be so stuck on the existence of gravity
    or electrical fields?

    Some things are arguable. Facts are not.

    rafe b
    Rafe B., Nov 29, 2003
  18. K2

    Rafe B. Guest

    Not conservative at all. We just happen to
    understand a few things that you refuse to.

    Your suggestion that Foveon "needs no
    antialiasing," or that aliasing can be removed
    by post-processing, or that the effects of
    aliasing can be ameliorated by using a
    better monitor or printer -- all bogus, and
    based on an utter lack of understanding.

    rafe b.
    Rafe B., Nov 29, 2003
  19. K2

    Rafe B. Guest

    Why should we forgive you for repeating a lie,
    again and again?

    rafe b.
    Rafe B., Nov 29, 2003
  20. K2

    Azzz1588 Guest

    Because he is relentless.......

    George dosnt know the truth between lying, and telling the truth.
    Goes along with his not knowing what a good picture looks like
    either. Or good color, or good contrast, or, or...........

    But he sure likes looking at test charts, and pixels though !!

    "Only a Gentleman can insult me, and a true Gentleman never will..."
    Azzz1588, Nov 29, 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.