SIGMA VIDEO CAMERA W/ FOVEON X75v....Tomorrow the official announcement will be their on D P Reviews

Discussion in 'Digital Photography' started by Jorge Prediguez, Jun 30, 2004.

  1. This new Foveon sensor is based on X3 technology and records 3 levels
    of colour like movie film. 3 chips will not be needed to give the same
    quality as a 3 chip high end Sony. Look for the announcement tomorrow.
    They will announce a new Sigma video camera and the new Foveon X75v
    sensor.
     
    Jorge Prediguez, Jun 30, 2004
    #1
    1. Advertisements

  2. Jorge Prediguez

    Mark B. Guest

    Gee George, does that mean you'll be spamming the NG with more garbage under
    yet another screen name?
     
    Mark B., Jun 30, 2004
    #2
    1. Advertisements

  3. Jorge Prediguez

    Ron Guest

    And just like the other Smegma crap, something to avoid.
     
    Ron, Jun 30, 2004
    #3
  4. Jorge Prediguez

    Sander Vesik Guest

    ^^^^^

    ah. the oall-telling poster name
    unlike 4 different colours like say our beloved Fuji print films?
    It would be so very nice if you stopped posting to rec.photo.equipment.35mm

    At least until such time when foveon starts making 35mm film.
     
    Sander Vesik, Jun 30, 2004
    #4
  5. (Jorge Prediguez) wrote in
    So ... now you have learned the skill of cross posting.
    How very clever of you Mr Preddy.

    BTW - movie film does not have three layers with colocated
    color sensors. It is more like a Bayer one with the sensing
    elements (grains) not colocated. And ... sometimes it is
    more than three layers. And the layers do record rather clean
    colors, not like the Foveon crap where all layers are a
    variant of monochrome layers, slightly filtered.


    /Roland
     
    Roland Karlsson, Jun 30, 2004
    #5
  6. Jorge Prediguez

    JPS Guest

    In message <>,
    I wouldn't be surprised if the current 3.43MP X3 records 3 levels of
    blue and green, blurs the hue, and gets the luminance directly from the
    red channel, because that's what the images look like.
    --
     
    JPS, Jul 2, 2004
    #6
  7. The precise spectral response of the 10M is published all over the
    web, same for the 5M.

    http://www.hanvision.com/contents/products/HVDUO3-10M_June_20_2004_HVC.pdf

    That is exactly why the Foveon Pro 10M used in the SIgma SD9 and SD10
    are also used for exacting scientific applications like the above,
    while Bayers never are as the data they present is 75% made up during
    interpolatative color processing. That's right, every pixel color in
    every Bayer image is 75% guessed, while 100% of every pixel color in a
    Foveon image was measured with perfect quantum accuracy precisely as
    presented, much like the time keeping accuracy of an atomic
    clock--absolutely perfect, all the time.
     
    George Preddy, Jul 2, 2004
    #7
  8. How do you keep a straight face with this bullshit?
     
    Randall Ainsworth, Jul 2, 2004
    #8
  9. SNIP
    Which indicates that e.g. green (550 nm wavelength) is seen as 49% blue,
    63%green, and 45% red. That means the color is VERY desaturated and massive
    postprocessing is needed to get a usable color (but with lots of noise).
    The above implementation is also only 30-bits digital color output!

    Bart
     
    Bart van der Wolf, Jul 2, 2004
    #9
  10. Jorge Prediguez

    Sander Vesik Guest

    LOL. 9-bit colour :cool: a whopping 128 different colours! excellent match to
    the colour screen of your first-gen colour screen mobile phone.
     
    Sander Vesik, Jul 2, 2004
    #10
  11. Not really. Those are "normalized" curves. I suspect they mean that
    each has been scaled to make the integral under the curve 1. It would
    be shocking if the blue channel had a higher peak response than red.
    "Massive"? "Lots"? Let's say "more noise in saturated colors".
    Yup. Read the fine print. "30-bit RGB output interfaces to parallel
    LVDS framegrabbers supporting four 8-bit input channels."

    Apparently they were working to an interface that was only capable of
    32 bits. That doesn't tell the limit of the sensor.

    The Foveon sensor has advantages and disadvantages, as with most
    technology. My impression is that the sensor in this camera (and the
    Sigmas) is competitive with 6MP color-filter-array sensors.
     
    Stephen H. Westin, Jul 2, 2004
    #11
  12. The ideal spectral response curves overlay the Foveon curves almost
    precisely (which is intentional of course), that is why Foveon is the
    only option for scientitfically perfect work. Colors are quantum
    accurate, all Bayer colors are guessed--100% of the output pixles are
    totally wrong, thus all Bayer images are of exactly zero scientific
    value.
     
    George Preddy, Jul 2, 2004
    #12
  13. Jorge Prediguez

    FLY135 Guest

    They can't be totally wrong and I can prove it. Just take a picture and
    make them more wrong, which wouldn't be possible if they were totally wrong.
    Then the picture will become unrecognizable.
     
    FLY135, Jul 2, 2004
    #13
  14. [Can you folks please stop cross posting?]

    (George Preddy) wrote in
    No, they don't. They are extremely far from perfect.
    Nothing intentional at all - you cannot modify how silicon absorbs
    different wavelengths of light with any precision.
    Is that why (almost?) no one uses Foveon for scientific works?
    In quantum physics you are not really talking about
    accuracy, rather you are talking about probability.
    Rather good guessing from the ones doing Bayer cameras. That
    it guesses that the sky is blue and the grass green I can understand.
    But I don't get it that it can guess that a certain car is red.
    You should see noise or maybe some other picture then. Imagine
    taking a picture of a dog and out comes a picture of a cat.
    That would be neat! I like cats better. They are cuter on photos.
    This is a bold statement!

    "All Bayer images are of no value."

    Let that be our mantra!


    /Roland
     
    Roland Karlsson, Jul 2, 2004
    #14
  15. Jorge Prediguez

    JPS Guest

    In message <020720040637144949%>,
    He actually believes it. SteveGeorge believes that if you arrange words
    in a certain way, you can conjure up any reality you want.
    --
     
    JPS, Jul 2, 2004
    #15
  16. Jorge Prediguez

    Crownfield Guest

    then he is probably a lawyer in real life.
     
    Crownfield, Jul 3, 2004
    #16
  17. I also happen to know that JohnMartha just bought an SD10 in real
    life. Welcome to the club.
    What's wrong with that?
     
    George Preddy, Jul 3, 2004
    #17
  18. 75% guessed pixels using values that are interpolated smack dab
    between their two neighbors isn't unrecognizable, it is just horribly
    low optical resolution.

    You don't understand any of this.
     
    George Preddy, Jul 3, 2004
    #18
  19. You can't do any better than quantum mechanical precision when
    measuring colors exactly every time. There is a small probability
    that your camera will go up instead of down when dropped, how many
    times has that happened to you?

    Canon uses tiny, manmade plastic color filters glued on top of a
    monochrome sensor, precise color accuracy could not be worse.
    Actually you don't get it at all. Guessing that a pixel is the same
    color as an its optically determined neighboring pixel doesn't make
    either color "wrong", what it does is flush optical resolution down
    the toilet.

    Bayer interpolation = simple upscaling. There is no optical
    difference except that Photoshop does a much better algorithmic job.
     
    George Preddy, Jul 3, 2004
    #19
  20. Jorge Prediguez

    JPS Guest

    In message <>,
    You're fabricating the technology, and you're bringing in all kinds of
    irrelevant connotations. Fact is, you have no point about anything,
    because Sigma DSLR color is inferior to Canon, Pentax, and Nikon DSLR
    color. The one thing it's supposed to be best at, it is worst at. A
    total sham.
    --
     
    JPS, Jul 3, 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.