Re: 25 Reasons to Aviod the SD-10 (was 15 Reasons to Aviod the SD-10)

Discussion in 'Digital Cameras' started by Paul Howland, May 9, 2004.

  1. Paul Howland

    Lionel Guest

    Kibo informs me that (George Preddy) stated
    that:
    ....a figment of your imagination, Preddiot.

    Why can't you produce these 'professional' photos you claim to have
    taken with your imaginary SD9?
     
    Lionel, May 22, 2004
    #61
    1. Advertisements

  2. Actually, they don't. You get a 3.42 meg file if you select Same Size
    rather than Double Size or Half Size in the raw file converter. The
    labelling on the camera is misleading and the image should really be
    judged as a 3.4 meg file. The fact that it is the best possible 3.4 meg
    file (nothing apart from a 3-shot camera could match it) makes a
    passable upscaled interpolation possible, but this is just what it is -
    an interpolated resizing of a clean, small file.

    David
     
    David Kilpatrick, May 22, 2004
    #62
    1. Advertisements

  3. Paul Howland

    Bill Funk Guest

    Absolutely not.
    The current CCD/CMOS sensors are absolute proof of that.
    Except for physics, you mean.
    A 3.42MP sensor will natively output a 3.42MP image. Simple fact. Any
    extra pixels are interpolated, both light & color data.
    A 6MP sensor puts out a native 6MP image; again, simple fact. The
    *color* is interpolated, but the pixels themselves are not.

    So it's interpolated *pixels*, or interpolated *color*; either way,
    there's interpolation involved.
    All your blather won't change that.
    No, for the Bayers, all the native pixels are output directly by the
    sensor; the *color* is interpolated, but the pixel itself is not.
    As opposed to the current SDx cameras, which interpolate *all* pixels
    over the native 3.42M.

    I'm not sure where you get this "optical" crap; if it applies at all,
    it applies to both; which puts the Foveon at a severe disadvantage,
    simply because it has fewer pixels to start with.
     
    Bill Funk, May 22, 2004
    #63
  4. Paul Howland

    Bill Funk Guest

    I thought that was what I said.
    Selecting another size resulkts in an interpolated image, no matter
    where the resizing is done: in the camera, or in the computer.
    End result remains the same: the SDx cameras are capable of a *native*
    3.42MP image; any other output is interpolated.
     
    Bill Funk, May 22, 2004
    #64
  5. Paul Howland

    Bill Funk Guest

    Ah, so you admit that the SDx cameras are actually 3.42MP cameras.
    Thanks.
    But those sensors are still represented in the native output file; a
    6MP camera will put out a 6MP image, with interpolated color.
    Which, BTW, seems to work very well.
     
    Bill Funk, May 22, 2004
    #65
  6. Paul Howland

    Lionel Guest

    Kibo informs me that (George Preddy) stated
    that:
    Good boy, George, now you're getting it.

    Now, stop tryig to change the subject, & show us these photos you claim
    to have sold.
     
    Lionel, May 22, 2004
    #66
  7. Paul Howland

    E. Magnuson Guest

    Yes, he's conflating the Sigma camera (which has a 1/6000th of a second
    shutter speed) with the sensor. Even here he has to exaggerate:
    http://www.foveon.net/prod_f7.html still says 4.4 FPS. That's in "rolling
    shutter mode" which is incompatible with the 1/6000th of a second
    (or any other speed) mechanical shutter anyway.
     
    E. Magnuson, May 22, 2004
    #67
  8. Paul Howland

    JPS Guest

    In message <>,
    It's not interpolated spatially, but the RGB is interpolated from
    PseudoRGB.
    --
     
    JPS, May 22, 2004
    #68
  9. (Laurence Matson) wrote in
    The first time I heard about the Bayer mosaic filters,
    I said - this is a bluff! They say 1 Mpixel, but it is not!
    But, Bayer interpolation has shown to be rather good.
    It was bad lenses and maybe bad interpolation algorihms
    that made the first cameras really lousy. OK - Bayer needs
    interpolation - now I said that twice :)

    OK - no one relly knows how the Foveon RAW to RGB converter
    works. But the tries I have seen to emulate it, they interpolate.
    Without interpolation it seems to be very hard to get rid of the
    excessive noise produced by the Foveon sensor. You can view
    this interpolation as smoothing or noise reduction if you wish,
    but it is still a kind of interpolation.


    /Roland
     
    Roland Karlsson, May 22, 2004
    #69
  10. Paul Howland

    JPS Guest

    In message <>,
    That must explain why the color is so much better with Bayer DSLRs.
    --
     
    JPS, May 22, 2004
    #70
  11. Paul Howland

    JPS Guest

    In message <>,
    There will always be optically naive people, who think that a black
    pixel right next to a white pixel equals image quality.
    --
     
    JPS, May 22, 2004
    #71
  12. Paul Howland

    Skip M Guest

    Laurence, now you're seeing why George irritates so many, so much. His
    numbers are not only wrong, they are intentionally so. He has been quoting
    these same figures over and over, in the face of innumerable people telling
    him he's wrong.
     
    Skip M, May 22, 2004
    #72
  13. (Laurence Matson) wrote in
    muhahah ... you ARE George!

    Nice try :)


    /Roland
     
    Roland Karlsson, May 22, 2004
    #73
  14. (Laurence Matson) wrote in
    If you have a sharp lens the is well focussed, then the entire
    picture will be full of aliasing errors. It has nothing to do
    with strange situation. We are not talking about Moire patterns,
    where it is most obvious, we are talking about pixel and detail
    misalignment.
    Sampling at the right position you mean, I assume?

    Thats the clue to the problem. The pixels will not move to the
    correct postion, therefore it will sample an incorrect value.
    As I have said otherwise - yes. If you don't like the "blur" you get
    with an anti alias filter and really wants pixel sharpness, you shall
    omit an anti alias filter. You will then get a sharper picture, full
    of faulty details. But if you like that, you do.
    The 14n does not have any anti alias filter, so it is bound to have
    problems. This picture looks weird though. The problem with those
    bright colors is hardly an aliasing problem. Something looks to be
    broken.
    No they don't. They call it anti alias filter.


    /Roland
     
    Roland Karlsson, May 22, 2004
    #74
  15. Paul Howland

    E. Magnuson Guest

    Been there, done that. He's been simply corrected many dozens of
    time. (Have you bothered to lookup Preddy's posting history yet?)
    Hang around a bit more, and we'll see how long you can remain
    polite.
     
    E. Magnuson, May 22, 2004
    #75
  16. Paul Howland

    E. Magnuson Guest

    To paraphrase a previous posting of yours Laurence, you're better than
    this. Did someone hijaak your account? First off, Canon has 3
    different CMOS sensors 2 of which have significantly higher luma
    resolution than the SD9/10. So let's assume you are talking about the
    just about 10D/300D sensor.
    The in-camera processes are essentially identical between these
    two. Please, if you are going to say such things, at least provide a
    cite. It is precisely this type of behavior that reminds us of
    he-who-shall-not-be-named. Remember the high ground.

    I have thousands of them. And so have many others. How much time have
    *you* spent with them? Stick to what you know. Sing the praises of the
    SD9/10. You'll even get some "amens" or at least a few people opening
    their minds to new possibilities. But you should know better than to
    denigrate the competition without a lot more proof in hand.
    No, they don't. Perhaps what you say might be true if you stick only
    to Canon FVU or in-camera JPEG for processing and do no other
    post-processing. But using either C1 or ACR can give distinctly
    better results and more choice between lowest noise and fine
    details. While there are some corner cases for both cameras, in the
    end it comes down to a very personal "true detail" vs. "aliasing"
    judgement call. Almost every review and side-by-side comparison has
    come to this same conclusion. Perhaps you are right and they are all
    wrong. But we'll need more proof.

    Smaller that what? The SD9/10? The pixel pitch is smaller but what
    about the fill factor? Cites please that show the actual sensor sites
    are smaller than the SD9/10. You can't win this argument on
    numbers. Either the images are better or they are not. (And even then
    you have to define "better")
    Finally, something most of us can agree on. In such a competitive
    field almost nothing is "unanimous".
     
    E. Magnuson, May 22, 2004
    #76
  17. No, it does not. Outside of Foveon and a few people who believe their
    marketing literature, and perhaps yourself, nobody believes it is 10.2
    MP because it has only 3.4 million light-sensing positions.
    Yes, but that's not what matters. A pixel is a picture element,
    spatially separated from its neighbours. The X3 sensor has only 3.4
    million of them.
    No, the Bayer cameras in question sample the image at 6 million points,
    not 2 million points, and so produce 6 megapixel images.

    There is a difference in the number of *bytes* per pixel for these two
    different sensor designs, but bytes are not pixels. And the luminance
    resolution, perhaps the single most important performance figure for a
    sensor, is determined by the number of measuring locations, not the
    number of colours measured at each site or the number of bytes of output
    data.

    And, in fact, the 6 MP Bayer cameras resolve more real detail than the
    3.4 MP Sigma/Foveon, even though the Sigma camera omits an anti-aliasing
    filter.
    Luminance resolution correlates pretty well with pixel count, as long as
    you count pixels properly (number of measurement sites). Both the Sigma
    cameras and the Canon 6 MP cameras resolve detail accurately up to about
    0.35-0.4 cycles/pixel. Beyond that, the Sigma generates mostly
    sharp-looking but incorrect (aliased) detail, while the Canon cameras
    produce mostly grey. The latter is a more correct representation of the
    scene contents, but some people may prefer the former. (It depends on
    the subject too).

    Dave
     
    Dave Martindale, May 22, 2004
    #77
  18. Who calls it a blur filter, other than Foveon?

    You *can* do antialiasing with a filter that simply blurs the image a
    certain amount. But in order to get the response near zero at the
    Nyquist frequency, you'd lose quite a bit of lower-frequency contrast
    that *is* representable without aliasing.

    It seems that, instead of simply blurring the image, camera
    manufacturers usually use a birefringent material that gives a
    controlled-spacing double image (really quad image, since it's done in
    both directions). This also prevents aliasing at the Nyquist frequency,
    but its rolloff is more abrupt, which retains more useful image
    information. But it would be a mistake to call this a "blur filter"
    since it does not blur the image.

    Dave
     
    Dave Martindale, May 22, 2004
    #78
  19. Paul Howland

    E. Magnuson Guest

    Boy (err, it's Girl today then?) nothing gets past you does it?

    Well, while listening to some old Tom Lehrer music, when he said
    it was "National Make Fun of the Handicapped Week" I just took it
    to heart.
     
    E. Magnuson, May 22, 2004
    #79
  20. Paul Howland

    Lionel Guest

    Kibo informs me that (Georgette Preddy)
    stated that:
    Nobody's interested in your lies, Preddiot. Show us these photos of
    yours.
     
    Lionel, May 22, 2004
    #80
    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.