Bow before the raw power of FOVEON

Discussion in 'Darkroom Developing and Printing' started by cameraguy, Jan 19, 2004.

  1. cameraguy

    cameraguy Guest

    cameraguy, Jan 19, 2004
    1. Advertisements

  2. cameraguy

    DJ Guest

    From that site:

    "In many cases, the difference in sharpness and detail is compounded by the use
    of blur filters in mosaic-based digital cameras. The blur filters are intended
    to minimize luminance and color artifacts. The artifacts are unpredictable
    byproducts of the complex processing required to interpolate the information
    mosaic image sensors miss."


    The need for AA filters is a direct consequence of sampling theorem, which
    applies to ANY imaging device with a regular array of pixels. The *amount* of
    lowpass filtering required may arguable vary between Foveon and Bayer, but the
    need is still there.

    A pity really. Potentially Foveon is a superior concept to Bayer (although by
    all accounts current implementations are still not delivering fully on their
    potential). But for the Foveon people to put out such misinformation seriously
    damages their credibility. If I were a major camera maker I would be reluctant
    to entertain technology from a dishonest company. Foveon are shooting themselves
    in the foot.
    DJ, Jan 20, 2004
    1. Advertisements

  3. cameraguy

    Tom Phillips Guest

    Correct. Foveon typically is not honest in it's representation of
    technical issues. Their engineers don't even return phone calls when
    technical queries are made. Nyquist is something they ignore. In fact,
    the 3D chip stacking method Foveon employs tends to require *larger*
    pixels (i.e., a greater pixel pitch) than 2D sensors. And since the
    photodetector sites have to be larger, it's a further limitation on the
    sampling rate. When imaging subjects of high frequency detail, the
    signal frequency still has to be dumbed down to prevent artifacting.
    That is the major inherent cause of digital imaging artifacting.
    However, not as much interpolating should be required given that each 3D
    photodector site collects full color data, i.e., you'd get more "real"
    pixels vs. more interpolated pixels.
    The pity is ignorant geeks who know nothing about the realities of
    imaging physics and the differences between digital and photochemical
    abilities are effortlessly brainwashed by the marketing B.S. of
    manufacturers whose only goal is to sell products and sustain market
    share -- usually by either misrepresenting their product or
    misrepresenting in comparison with film imaging.
    Tom Phillips, Jan 20, 2004
  4. There's a thing that records images very well.It's called the FILM.Well
    done, Foveon;you discovered the wheel all over again.
    Dimitris Tzortzakakis, Jan 20, 2004
  5. cameraguy

    JPS Guest

    In message <>,
    That would be one of the bigger benefits of full-RGB photosites, if it
    were done with a color-precise technology (which the Foveon isn't).
    There would be less battle with noise vs sharpening, as sharpening
    strength could be a little weaker at sub-pixel radii to get the same end

    Of course, by omitting AA filters completely (and microlenses, in the
    SD9), the Sigma DSLRs go too far, and end up with nonsense data.
    JPS, Jan 24, 2004
  6. cameraguy

    Tom Phillips Guest

    He's referring to the sampling limitations (Nyquist Theorem) inherent with *all*
    silicon sensors. The RGB filtering system makes no difference.
    Tom Phillips, Jan 25, 2004
  7. cameraguy

    Tom Phillips Guest

    It doesn't affect *NYQUIST*
    Tom Phillips, Jan 25, 2004
  8. cameraguy

    JPS Guest

    In message <>,
    It most certainy does. The Bayer CFA pattern requires more aggressive
    filtering than a stacked-RGB sensor or a monochrome sensor, to avoid
    color moire.
    JPS, Jan 25, 2004
  9. cameraguy

    DM Guest

    DM, Feb 18, 2004
  10. More like "the ignorant one."

    Fred A. Miller, Feb 18, 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.