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

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

  1. SNIP
    Yes, the more I think of it, the more I value the Bayer CFA concept. The
    extreme opposites of the spectrum are not only unlikely (if ever) to occur
    side by side in a natural setting that requires luminance resolution, they
    are also *extremely* unlikely to form a metameric color.
    This coupled to the similarity with human vision which also has a lower
    sensitivity for color resolution, makes it a very efficient method to
    capture single shot tri-chromatic color.
    Of course everybody would like to use an even better method, but not at the
    cost of poor/incorrect color separation.
    Whatever provokes reactions, makes him feel as if he's an important pivot
    However, I realize that by keeping him busy, we're actually doing his local
    community a service, and it also offers a platform to correct some of the
    fundamental misunderstandings about digital imaging.
    It's a pitty the entire Foveon concept is placed in such a bad light by
    Preddy's examples, but I guess that's the price of progress.

    Bart van der Wolf, Dec 2, 2003
    1. Advertisements

  2. K2

    imbsysop Guest

    ... but if one leaves out the GP BS then this thread is in fact highly
    informative and of educational power on the why & how's of digital
    photography :) no doubt about that ..
    imbsysop, Dec 2, 2003
    1. Advertisements

  3. K2

    Tony Spadaro Guest

    Tony Spadaro, Dec 2, 2003
  4. If someone tries the same fraud over and over again with people
    that tell him that it is a fraud, what would that then make him?

    Roland Karlsson, Dec 2, 2003
  5. Yes, it's a loss for those that have kill-filed everything with certain
    keywords, although I understand it.

    Bart van der Wolf, Dec 2, 2003
  6. I checked. I couldn't actually find a test target for the SD9. Could you
    point me to one on DPrevew. They did have one for the 10d with a resolution
    of a little over 1600 lines HZ. Since the SD9 only has a little over 1500
    horizontal lines in the sensor, it is hard to see how it would legitamately
    get over 1500.
    Gherry Bender, Dec 3, 2003
  7. K2

    imbsysop Guest

    well killing anything that has GP in the "From" field nicely filters
    out the "noise" :)
    imbsysop, Dec 3, 2003
  8. Only about 98% ;-)

    Bart van der Wolf, Dec 3, 2003
  9. K2

    pehache Guest

    After having read many posts on this topic, I'd like to add my 2

    The truth is that the Foveon sensor is by nature *much less* sensitive
    to aliasing than sensor based on a Bayer matrix.

    For, the elementary photosites are not points, they have a spatial
    dimension. This corresponds to a low-pass filter, which is
    anti-aliasing. It is not a perfect low-pass, all the more than the
    actual size of the photosites is smaller than the distance between 2
    of them, so it leaves a bit of residual aliased high frequencies. In
    practice most of the aliasing is removed, even if some residual
    aliasing can be observed on a few pictures.

    The same comments apply to Bayer-based sensors. However, the fact that
    each individual color is undersampled requires a more aggressive
    anti-aliasing filter than the natural one based only on the photosite
    size. This explains why most of Bayer-based sensors have an
    additionnal anti-aliasing filter.
    pehache, Dec 3, 2003
  10. K2

    Guest Guest

    You are clueless. Sampled systems require an anti-alias filter. Foveon needs
    one. Bayer needs one more because the color samples are spatially offset.

    They still both need one though.
    Guest, Dec 3, 2003
  11. K2

    Guest Guest

    You should be annoyed with George now then - he has been proved to be
    modifying his so called out of camera originals.
    Guest, Dec 3, 2003
  12. K2

    pehache Guest

    A pure sampling system, yes.

    But the fact is that a CCD sensor is far from a pure sampling system.
    There is an integration over the photosite's surface.
    Not to the same extent. A Bayer sensor is closer (even if still very far
    in reality) to a pure sampling system than Foveon, and therefore requires
    more anti-aliasing filtering.

    ne pas répondre à cette adresse / do not reply-to this adress

    Posté depuis
    pehache, Dec 3, 2003
  13. K2

    Guest Guest

    You're such a low life bottom feeder!

    There is only one reason to post full size images on the web, and that is so
    idiots like you can measurebate over them. 800x600 is a good size for web
    Guest, Dec 3, 2003
  14. K2

    Guest Guest

    And I am going to trust what they tell me over reputable magazines, web
    sites AND MY OWN EYES for exactly what reason?
    Guest, Dec 3, 2003
  15. K2

    Guest Guest

    I think the human visual system that has evolved over millions of years (or
    created by the supremem being if that's your belief) know how to see
    perfectly well - it doen't need your "new way of seeing things".
    Guest, Dec 3, 2003
  16. K2

    Guest Guest

    It doesn't matter what you monitor setting are you fool.

    You just zoom in to 100% in the image editing app of your choice.
    Guest, Dec 3, 2003
  17. K2

    Guest Guest

    It doesn't win in resolution. It quite clearly resolves less detail than a
    most 6MP bayer cams.

    It also very clearly suffers from alias artefacts ( 9 <> 5 lines)
    Guest, Dec 3, 2003
  18. K2

    Guest Guest

    No they don't.

    The camera which is the closest for the price match to the Sd9 is the 300D.
    The 300D kicks the SD9 into next week for quality of images. It kick it into
    next century on sales.

    Bullshit all you like about sensor counts. The pictures are what matters,
    and yours are rubbish.
    Guest, Dec 3, 2003
  19. K2

    Guest Guest

    No George,

    We knew about Foveon before the Sd9 came out. We were waiting with baited
    breath to see its results.

    It's results are poor. I hope the SD10 is better, but its regrettable that
    they still have no AA filter so I still don't want one.

    Now go on, blither on about pre-production versions I dare you. I will tell
    you that now FOR FREE that the images you post look terrible. Is your camera

    They are the facts and it is only YOU who is too ignorant to accept them.
    Guest, Dec 3, 2003
  20. Of course you can, they are from 3 mutually exclusive populations. The SD9
    takes 10.3M spatially discrete samples.
    George Preddy, Dec 3, 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.