A fun game to play with George Preddy

Discussion in 'Digital Photography' started by Lionel, Jan 19, 2004.

  1. No pixel has known color. Not for any camera.
    It is always an aproximation.


    /Roland
     
    Roland Karlsson, Jan 21, 2004
    1. Advertisements

  2. You will never get George to give a straight answer to the question
    why a Foveon sensor do not need any anti alias filter.

    The reason for this might be -
    1. he is just trolling - and not answering questions
    is a well known method for keeping a thread alive.
    2. he knows he is wrong, but he don't want to admit it.
    3. he has some kind of mental blocking; some questions
    he cannot simply answer.
    4. he do not understand what we mean with alaising, so
    he cannot really answer - for he have no answer.
    5. he really believes that because Foveon is measuring
    and not sampling color, it does not need any.
    6. something else.

    question for you George. This camera
    http://www.steves-digicams.com/dcs620.html
    is monochrome and it has an anti alias filter.

    So ... if you take three pictures with a R, G and B
    filter respectively. Would it then still need the
    anti alias filter?


    /Roland
     
    Roland Karlsson, Jan 21, 2004
    1. Advertisements

  3. Soooo ... now I understand. This is what it boils down to.
    Foveon is made by master minds so it must be fantastic.
    Then everything that anyone else says must be wrong. Hmmm
    .... I think I see a superiority syndrome here.


    /Roland
     
    Roland Karlsson, Jan 21, 2004
  4. We all do :)

    I don't know why - but this Preddy is entertainment.


    /Roland
     
    Roland Karlsson, Jan 21, 2004
  5. In some sense he is right. It takes 10 Million luminance
    samples. Unfortunately (for spatial resolution) the samples
    comes in bunches of 3 at the same location. Therefore it
    is 3.4 Mpixels.

    In some other posts he claims that there are 10 Million
    spatially separated sensors. In some sense he is right
    here also. Unfortunetely (for spatial resolution) he then
    (probably) refere to 3D spatial separation. But the only
    thing interesting here is 2D spatial separation. And then,
    we only have 3.4 million 2D spatially separated sensors,
    i.e 3.4 Mpixels.

    So ... why can't he se this? Or do he see it and just is
    playing games? I don't know.


    /Roland
     
    Roland Karlsson, Jan 21, 2004
  6. OK - show us - give us the link to a 3 sensor
    digital video camera that claims the they have
    3 times more pixels than resolution.


    /Roland
     
    Roland Karlsson, Jan 21, 2004
  7. Lionel

    C.G. Guest

    STOP PRESS! GUY WITH SMALL DICK SAYS SIZE DOESN'T MATTER!!

    --
    Colm



    : The Foveon sensor has 10.3M discrete loactions because each channel samples
    : 3.43M discrete locations spread evenly over the entire sensor area.
     
    C.G., Jan 21, 2004
  8. Lionel

    Bill M Guest

    Each pixel starts with a known R, G or B component and luma...pixels
    are completed by examining at least 8 neighboring pixels. Starts with
    6MP, ends with 6MP...
     
    Bill M, Jan 21, 2004
  9. I think all 6 ;-)

    Bart
     
    Bart van der Wolf, Jan 21, 2004
  10. So?

    Bart
     
    Bart van der Wolf, Jan 21, 2004
  11. Any links to their statement?

    Bart
     
    Bart van der Wolf, Jan 21, 2004
  12. Lionel

    Lionel Guest

    ROTFL! Good call!
     
    Lionel, Jan 22, 2004
  13. So? Obviously his statements don't reflect the
    realities of imaging sensor technology, and the
    schematic drawings of the Foveon Chip try to
    represent the Foveon chip for something it isn't.

    At least the drwaings on this website are in
    striking contrast to the real nature of the
    chip as documented in the relevant patent at
    http://www.uspto.gov.

    This looks like either complete idiots have put
    this website together, or it is attempted (or real?)
    fraud just like those guys at Silicon Film did.

    If Carver Mead knows what is on this website or
    has even authorized this website, then he is a
    crook. A big crook.
     
    Michael Quack, Jan 22, 2004
  14. Since the Foveon chip samples in a substandard way
    and uses a prehistoric interpolation algorithm, it
    matters as much as a fart in the gas chamber.
     
    Michael Quack, Jan 22, 2004
  15. No, they are not. Look at their patent,
    and you know why the colors are so bad and
    unacceptable, and why there is so much
    more aliasing.

    Also, Foveons websites are not honest, they are
    posting schematic drawings that do not show
    their chips as they are. This is misleading
    if not outright fraudulent.
     
    Michael Quack, Jan 22, 2004
  16. If any manufacturer would indeed claim such a stupid
    thing - it wouldn't change anything, since the video
    signal is standardized. Whatever extra information could
    be gathered if we were stupid enough to believe you
    had a clue - it would not make it into the video signal.
     
    Michael Quack, Jan 22, 2004
  17. And many others.
    The leading crook is by far the better description.

    The website is deceptive, factually lying.
    If this was done on purpose, it is fraud.
     
    Michael Quack, Jan 22, 2004
  18. Michael Quack, Jan 22, 2004
  19. Look at the patent, there *is* lateral offset.
    Which means that our little Gizmo-Preddy is in some
    way right. But after getting wet, he fails to
    comprehend that due to their way of colorseparation
    not by precisely defined primary extraction filters
    but by absorption depending on layer thickness
    (which is out of control in their manufaction)
    and by inferior interpolation procedures using only
    3 instead of 9 values Bayers use per output pixel
    in combination with the lack of AA filter the
    Foveon/Sigma renders dramatically inferior colors
    and massive artefacts in comparison to Bayers.
     
    Michael Quack, Jan 22, 2004
  20. Too bad they don't master real interpolation
    algortihms for their inferior chip.

    Taking only 3 values with lots of sampling error
    instead of 9 values of known precision for a
    single output pixel is indeed declaring intellectual
    and technical bancruptcy.
    Already the 3.4 are yucky interpolation.
     
    Michael Quack, Jan 22, 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.