Sigma Digital SLR Cameras.

Discussion in 'Digital Photography' started by Biff Mullins, Jun 18, 2004.

  1. Biff Mullins

    Marc 182 Guest

    No argument there, except that it is also artificial in the analog
    domain.
    This part isn't correct. Changing the ISO setting on a digital camera
    changes the gain on the readout amp for the CCD, which is even before
    the A/D converters get the signal, much less the camera's CPU. This not
    processing, it is a physical change to how the CCD is readout. The
    effect is to increase the sensitivity of the CCD.
    No. This would imply that you could shoot a low-light scene at the ISO
    100 setting and then use Photoshop to get the same result as shooting at
    the ISO 1600 setting; however, this is simply not the case. The ISO 1600
    setting will record details in the shadows (also with a lot of noise of
    course) that no amount of photoshopping will be able to pull out.
    The sensitivity of CCDs are adjustable within limits.

    Marc
     
    Marc 182, Jul 26, 2004
    1. Advertisements

  2. Film could also be said to be the same, except that there are special
    For Canon that limit is 50 - 3200 on their CMOS sensors
    For Nikon's CCD's it is 200-1600 with up to a 2 stop boost at 1600
    Kodak has a CMOS sensor that goes from 160 - 1600

    The bottom line is that you can take a waterfall shot on a tripod and high
    speed action shot a few seconds later.
     
    Mark Kovalcson, Jul 26, 2004
    1. Advertisements

  3. Biff Mullins

    Crownfield Guest

    only is the 'direct sensor' is a good sensor.
    foveon sensors are a great idea that works poorly.

    that sums up the sigma cameras.
    they work poorly.
    thus george again picks a loser.
    say 'not a clue' preddiot.
    foveon is 2250 x 1500 pixels.
    that is only 3 mp.

    it is not 6, not 8, not 12, not 13,
    but 3.

    Hasselblad gave up on foveon
    once they saw how bad it was compared to better sensors.
     
    Crownfield, Jul 26, 2004
  4. A 2230 x 1500 image requires exactly 2250 x 1500 x 3 sensors.
    Obviously.

    Don't you realize there are 3 color channels?
    Exactly why the Canon 10D is only 1.5MP, optical. It only has 6M
    monochrome sensors, that is way, way, way too small to produce a 6M
    pixel 3 color channel image.

    A 6M pixel image absolutely requires 18M sensors.
    Hassy couldn't afford Foveon after they had already choosen Foveon
    over Leaf on color accuracy and resolution. Common knowledge, it's
    all over the web. Real shame too, the prototype was said by those who
    used it be the greatest digital still camera ever produced.
     
    Georgette Preddy, Jul 27, 2004
  5. Biff Mullins

    Andy Fraser Guest

    In rec.photo.digital, Georgette Preddy uttered these immortal words:
    I'm hoping someone can help me understand this better. I went to the Sigma
    website and looked at the specs of the SD10 and sure enough they say this
    camera produces images that are 2268 x 1512 x 3 layers.

    Here's where I get confused, please bare with me. If I take a picture taken
    with my Kodak 5MP camera (I'm not about to attempt to compare these 2
    cameras BTW) and save it as a bitmap I get a file that's 14,953,734 bytes
    in size. That's what I'd expect given the resolution of 2580 x 1932 x 3
    bytes per pixel, 1 each for the red, green and blue values of each pixel.
    The fact that that picture was arrived at by applying an algorithm to the
    data from the Bayer mosaic filter CCD jobbie is irrelevant to what I'm
    getting at. AIUI my camera produces a photograph with nearly 5 million true
    colour pixels in it while the Foveon X3 method used by the SD10 produces a
    3.4MP true colour image.

    I can see where Sigma get their 10.29MP value from. They seem to be saying
    that they have 10.29 million sensors. Ok but surely that's just a different
    (and in theory better. I don't know about in practice) way to get the RGB
    values for each pixel?

    Which ever way I look at it the Sigma SD10 still only produces 3.4MP 24-bit
    photos. I'd be pretty pissed off if I bought a camera claiming to be a 10MP
    camera but only got 3.4MP images. Are any of my assumptions correct? How
    can Sigma get away with saying their camera is a 10MP device? I, like a lot
    of other less knowledgable digital camera users, always assumed that the
    cameras MP value described the size of the images it outputs and not the
    number of sensors in the CCD.

    NB I'm not talking about the /quality/ of the output at all here, just the
    method that produces it, and I'm not trying to say that my Kodak consumer
    compact camera in any way compares to a DSLR. :)
     
    Andy Fraser, Jul 27, 2004
  6. That's correct.
    Yes. It has the advantage (in theory) that the colour-difference
    information is present at the full resolution of the sensor. The colour
    image from a 5 MP Bayer sensor can be thought of as a 5 MP B&W image
    with 1.25 MP worth of colour information added, while the Foveon is 3.4
    MP of both B&W and colour information. However, the human eye is much
    more sensitive to fine luminance detail (10 times better resolution!),
    so in practice the lower colour resolution given by the Bayer method
    doesn't matter in photography.

    Also, the Foveon sensor doesn't perform as well in practice as theory
    would suggest.
    Advertising sometimes lies. This is a good example of it. There is no
    reasonable definition of "pixel" that allows calling them 10 MP
    cameras.

    Dave
     
    Dave Martindale, Jul 27, 2004
  7. Biff Mullins

    Andy Fraser Guest

    In rec.photo.digital, Dave Martindale uttered these immortal words:
    That's how I understood it. Thanks for confirming it.
    That's a shame because it sounds like a good idea on paper.
    And don't we all know it. :)
    The "Total Digital Photography" buyer's guide lists the SD10 as 3.42MP x 3
    which is a bit less misleading than the Sigma advert in the same issue that
    proclaims "10.2 million pixels of true colour reproduction". To me that
    says "This camera will output an image with 10.2 million pixels in it, each
    represented by 24-bits of colour information". This should be illegal IMHO.

    If I was in the market for a DSLR and given the prices in the above
    mentioned buyer's guide I'd be more tempted to save myself £200 - £300 and
    checkout the Nikon D70 or Canon EOS 300D instead anyway if only because
    they're names I know.
     
    Andy Fraser, Jul 27, 2004
  8. IMO additional color information is advantageous in
    low contrast regions of normal contrast photographs.

    YMMV.

    Richard Ballard MSEE CNA4 KD0AZ
     
    Richard Ballard, Jul 27, 2004
  9. Biff Mullins

    Crownfield Guest

    and a 12 mp image absolutely requires 4,000 x 3,000 sensors.
    tiny 2250 x 1500 sensors do not cut it in the real world.
    they have a back now that works. actually two of them.
    One of them was described as the best color ever seen.

    I'll let you know when and which I get.

    22mp, each and every pixel: full color exposure.
     
    Crownfield, Jul 28, 2004
  10. If you drew a 5M pixel image by hand with Microsoft Paint, then saved
    it as a TIF it would be the same size too. That doesn't make it an
    optical photograph. All that matters is the optical quality of the
    optical data, the number of pixels are irrelevant, any image can be
    upscaled to consume any number of pixels.

    A 6MP Bayer has 6M monochrome sensors, a 3.4MP (output is 13.72MP)
    Foveon has 3.4M full color sensors. It would take a 13.72MP Bayer to
    equaly the quality of Foveon data, if such an enormous sensor existed.

    I suggest reading Phil Askey on the subject...

    "The Foveon X3 sensor captures full color for each pixel location and
    thus requires no demosaic processing, a single individual pixel has
    its own distinct color without influence from neighbouring pixels.
    This leads me to the term 'Single Pixel Resolution', the X3 sensor's
    striking ability to capture image detail down to a single pixel level,
    such as a wire which in the image is just one pixel across.

    You can of course produce similar results from a Bayer sensor camera
    by downsampling an image by 50% [meaning 25% of the original image
    area], this is (approximately) the same as combining two green, one
    red and one blue pixel together. To achieve this and be left with an
    image of the same size as the SD10 you would need 4536 x 3024 (13.7
    megapixel) input image."
     
    Georgette Preddy, Jul 30, 2004
  11. SNIP
    Nonsense, as usual. Sensors are not monochrome, nor are they full
    color.

    Photo sensors (used in imaging) are sensitive to a wide range of
    wavelengths, ranging roughly from 350-1000 nm. The sensitivity is not
    the same across the spectrum. To produce RGB output, the sensor is
    usually filtered to record the energy of a spectral band. The
    filtration is either done by adding a filter layer in front of the
    sensor array, or by the absorption characteristics of silicon itself.

    The output of any sensor is then by definition monochrome, because it
    represents the average energy of all filtered wavelengths integrated
    over the exposure time weighted by the native spectral sensitivity of
    the sensor.

    Bart
     
    Bart van der Wolf, Jul 30, 2004
  12. Then why do you continue to crow about megapixel comparisons?

    BJJB
     
    BillyJoeJimBob, Jul 30, 2004
  13. Biff Mullins

    JPS Guest

    In message <>,
    He crows about whatever is most convenient for him to crow about, based
    on where his kaleidoscopic vision happens to be pointing, at any given
    time.
    --
     
    JPS, Jul 30, 2004
  14. Biff Mullins

    Crownfield Guest

    preddiot thinks that 3 mp from his sigma
    equals or exceeds the 22 mp exposure
    from phase one and imacon backs.

    only a fool would actually believe this,
    but thats the preddiot for you.
     
    Crownfield, Jul 31, 2004
  15. If you are the real GP welcome

    If not, sod off troll

    Please forgive me if it really is you George but I believe your identity has
    been hijacked on several occasions

    This may be one too

    Arts

     
    Arty Phacting, Jul 31, 2004
  16. All Bayer MP ratings are strictly monochrome. Color is digitally
    interpolated after the shutter closes at an expense of -75% of the
    monochrome resolution. When a Bayer DSLR maker says 6MP, they mean 6M
    monochrome sensors, not full color. That is why you'll never see a
    color resolution test image used by on a site that is sponsored with
    big money from Bayer camera manufactures, like dpreview.com and
    imaging-resource.com.
     
    Georgette Preddy, Aug 1, 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.