Dpreveiw.com settles the MP debate for good

Discussion in 'Digital Cameras' started by George Preddy, Jan 10, 2004.

  1. Phil Askey has been kind enough to settle this debate as a matter of record.
    According to dpreview.com:

    The SD9 has a 3.43MP sensor with 10.3M photosites.
    The Canon 10D has a 1.58MP sensor with 6.3M phototsites.

    Nice to see common sense has finally prevailed. Ref:

    "Phil: As I expected this news article has stirred up a whole lot of further
    debate. The confusion here comes about from existing manufacturers use of
    the word 'pixel'. Our understanding of a pixel is a single dot in the final
    output image, however we all know that for a mosaic sensor that single
    output pixel has been made up from several input photosites. Thus although
    we call an exisiting CCD '3 million pixels' it (I suppose) be called '3
    million photosites'. In which case the X3 sensor used in the SD9 would be
    called '10.3 million photosites'.
    When debating this it would probably make more sense to refer to the pixel
    as what we see in the output image and the photosite as a single captured
    colour value from the sensor (input)."
    http://www.dpreview.com/news/0203/02030602foveonx3notation.asp
     
    George Preddy, Jan 10, 2004
    #1
    1. Advertisements

  2. George Preddy

    dylan Guest

    But who cares ?. This isn't going to interest people into buying a SD9.

    You are just filling up this NG with drival. You may love the Sigma/X3, and
    the sensor may be have some advantages when perfected and put in a good
    camera, but just sniping at other products isn't going to win you any
    friends.
     
    dylan, Jan 10, 2004
    #2
    1. Advertisements

  3. I don't think quoting dpreview is sniping. It's time the Sigma bashers
    learned about these cameras.
     
    George Preddy, Jan 10, 2004
    #3
  4. SNIP
    so following that e.g. 6MP for a D60/10D and 3.4 for an SD-9/10 ?
    so e.g. 6M photosites for a D60/10D and 10.3M photosites for an
    SD-9/10 ?

    I do believe that almost everybody, except you, already understood
    that. Also commonly understood is the fact that resolution is the
    result of sampling points in two (and not three) orthogonal sampling
    directions, IAW sample spacing on the input or pixel pitch on the
    output.

    So, now Phil Askey also confirms that, what's new?

    Bart
     
    Bart van der Wolf, Jan 10, 2004
    #4
  5. George Preddy

    Nils Rostedt Guest

    For some unknown purpose other than a fair comparison of cameras,

    Wrong. According to Phil's definition,

    The SD9 has a 3.43MP sensor with 10.3M photosites.
    The Canon 10D has a 6.3MP sensor with 6.3M photosites.

    Read it.
     
    Nils Rostedt, Jan 10, 2004
    #5
  6. George Preddy

    mhlife Guest

    ---


    It's a legitimate comment about cameras

    It's firmly on-topic

    He's perfectly entitled to post it

    End of story.
     
    mhlife, Jan 10, 2004
    #6
  7. George Preddy

    hljgli Guest

    I never said he couldn't post it, just gave my opinion of it.
     
    hljgli, Jan 10, 2004
    #7
  8. Your reply was intentional gibberish. Phil says the Canon 10D is a properly
    classified as a 1.5MP camera because it takes four photosites to make a
    pixel. Read the link this time, before you reply...

    "Our understanding of a pixel is a single dot in the final output image,
    however we all know that for a mosaic sensor that single output pixel has
    been made up from several input photosites. Thus although we call an
    exisiting CCD '3 million pixels' it (I suppose) be called '3 million
    photosites'. "

    http://www.dpreview.com/news/0203/02030602foveonx3notation.asp

    Note what he said there, "we all know." That means you know it too, so why
    are you pretending not to know this?
     
    George Preddy, Jan 10, 2004
    #8
  9. George Preddy

    B.o.w.s.e.r Guest

    Trying to carry on a rational discussion with Preddy, who I've killfiled,
    but the replies still "get him in," is like mud-wrestling with a pig. After
    a while, you realize the pig likes it.

    Sigma=really sucky color.
     
    B.o.w.s.e.r, Jan 10, 2004
    #9
  10. I did read it, but where does he say that ?????????????
    By the way, the article you linked was from the 6th of March, 2002.
    How could he mention the 10D since it didn't even exist ???????????
    You are lying again, pathetic.

    SNIP
    You are not making any sense, time for your medication again I guess.

    Bart
     
    Bart van der Wolf, Jan 10, 2004
    #10
  11. George Preddy

    Mark Johnson Guest

    I wonder if they should just have used "recording", rather than,
    "recorded" - number of recording pixels, to emphasize the action of
    those. And instead of "output", maybe just simply - number of image
    pixels.
     
    Mark Johnson, Jan 10, 2004
    #11
  12. George Preddy

    dslr Guest

    It's not, because he's wrong, has been told so on countless occasions
    and will not accept it.
    It's not, because he's wrong, has been told so on countless occasions
    and will not accept it.
    As are we to killfile him - I just wish everyone else would.
    Unfortunately, I doubt it :-(
     
    dslr, Jan 10, 2004
    #12
  13. George Preddy

    JPS Guest

    In message <btomqa$66b$>,
    I know all I need to know. They suck at recording accurate color (even
    if they are good at recording relative changes of color at high
    frequencies), and they sample too sharply, resulting in non-contiguous
    images. The output of the SD9 is like the output of a properly-filtered
    23MP image reduced to 3.43MP with the "nearest neighbor" algorithm.
    --
     
    JPS, Jan 10, 2004
    #13
  14. George Preddy

    JPS Guest

    In message <btp9bl$9pbl8$-berlin.de>,
    It's just further proof that SteveGeorge is fixated in a state of
    childhood "word magic". He actually believes that being able to call
    the SD9 "10.3MP" somehow elevates it above what it really is. If the
    standards committee in a drunken stupor decided tomorrow that the SD9
    was 10.3MP, the fact still remains that the SD9 only samples data at
    3.43M locations, and the "6MP bayers" at 6M+ locations. We'd just have
    to find another term instead of "pixel" to express this fact.

    You see this theme throughout all of his argument; if A then B, if B
    then C, etc, with semantic bait and switch occuring at ever step. He is
    trying to build a Steve-supporting world of illusion through cheap
    language tricks.
    --
     
    JPS, Jan 10, 2004
    #14
  15. He's wrong. A photosite is a single *location*, that's why it is called
    a "site". The Foveon X3 sensor has 3 colour sensors at each site.

    So: the SD9/SD10 have 3.4M photosites containing 10.3M sensors, which
    produces an image with 3.4M pixels.

    The 10D has 6M photosites and 6M sensors, producing an image with 6M
    pixels.

    Foveon may *want* to count each colour sensor as a separate "site", and
    they may even have convinced Phil, but it's still wrong. After all, if
    you own a 3-storey apartment building with 3 families living in it, you
    might count it as 3 homes, but they all occupy the same site (place).

    Dave
     
    Dave Martindale, Jan 10, 2004
    #15
  16. George Preddy

    Dan Sullivan Guest

    Would it be fair to say that the Foveon sensor has 3.43MP sensors sites for
    each of red, blue, and green, totalling 10.3M individual photosites?

    And that the 10D has 1.58MP sensors for each of red, blue, and green,
    totalling 6.3M individual photosites?

    Dan Sullivan
     
    Dan Sullivan, Jan 10, 2004
    #16
  17. George Preddy

    Nils Rostedt Guest

    No, that wording fails to define the number of light sensing locations in
    the image plane (which gives the number of pixels) , as well as the number
    of light sensing locations in the optical axis plane, in a way meaningful
    for comparison.
     
    Nils Rostedt, Jan 10, 2004
    #17
  18. George Preddy

    Crownfield Guest

    and if the sd-11 has 12 million photosites,
    and they all are colocated at one x-y site,
    will it make an even better picture?

    how many pixels will you be abl to interpolate
    out of that mega super pixel?

    hmmm...
     
    Crownfield, Jan 10, 2004
    #18
  19. George Preddy

    steve Guest

    Nyquist is rolling in his grave, and a undred years of sampling theorem
    are screaming 'bull$shit'

    Regards,

    steve
     
    steve, Jan 10, 2004
    #19
  20. George Preddy

    steve Guest

    Nyquist is rolling in his grave, and a hundred years of sampling theorem
    are screaming 'bull$shit'

    Regards,

    steve
     
    steve, Jan 10, 2004
    #20
    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.