Nagging questions

Discussion in '35mm Cameras' started by Eric Miller, Dec 2, 2009.

  1. Eric Miller

    Eric Miller Guest

    Just some random, mostly resolution-related questions. I fully accept
    that the questions themselves may be misguided, so feel free to interpret.

    Chip size, noise and other stuff.

    All other things being equal, will a larger sensor have more noise? I
    ask this for two, interrelated reasons, the first of which is that I see
    discussions of the end of the Megapixel wars for APS-C sensor due to the
    loss of image quality with higher pixel pitch. The second is that I have
    recently seen some pretty good images from a Canon G10 which has a much
    higher pixel pitch albeit on a much smaller chip.

    Actual resolution increase with higher pixel pitch.

    It certainly seems that the G10 has much higher resolution than my 5D
    (but certainly not better image quality) considering the pixel
    pitch/sensor size. What actual resolution increase could I expect with
    that kind of pixel pitch on an APS-C or FF/35mm sensor? Is there some
    formula that describes the diminishing resolution gains due to added noise?

    Size of pixel/photosite where resolution increase would not equal
    additional pixels.

    Is there a certain way in which noise is related to the size of the
    individual photosite? Or is that dependent upon other factors that can
    be addressed through better engineering/manufacture of the sensor?

    Conversion of Bayer pattern only affect colors of image.

    Finally, do the algorithms used to convert the output of a Bayer-type
    sensor independently process luminance and color information? I guess
    that would be, does the algorithm produce color and luminance
    information independently of each other such that a grayscale value
    could be produced for each pixel without using information collected at
    more than one photosite?

    Eric Miller
    www.dyesscreek.com
     
    Eric Miller, Dec 2, 2009
    #1
    1. Advertisements

  2. Eric Miller

    Guest Guest

    no. for a given sensor technology and pixel count, a larger sensor will
    have larger pixels, and therefore have less noise.

    <http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
    ry/>
    resolution depends on a lot of factors, including the sensor, the lens,
    how accurately you focus, etc. you can have the best sensor in the
    world but if you slap a cheapo lens on the camera, it won't matter
    much.

    if you want to know how a g10 compares with a 5d, you need to test them
    both. dpreview has done that, and the g10 will have slightly higher
    resolution, but a lot more noise.
    the bigger the better, but better sensor technology counts too. a small
    sensor of today might be better than a large sensor of a few years ago,
    but you can always make a larger sensor with the latest technology so
    it's moot.
    there are many many ways to demosaic bayer images, and you will need
    more than one photosite no matter what you do.
     
    Guest, Dec 2, 2009
    #2
    1. Advertisements

  3. Eric Miller

    NameHere Guest

    I guess that's why this 1/2.5" sized sensor has more dynamic range than
    most APS-C sized DSLR sensors which only have 7-8 EV stops.

    http://farm4.static.flickr.com/3142/2861257547_9a7ceaf3a1_o.jpg

    You're a pretend-photographer troll, one that loves nothing better than to
    spout biased misinformation posted by another DSLR-Troll who posted tests
    on his website to justify why he wasted all that money on his DSLR gear.
    I.e. Clarkblindness.com
     
    NameHere, Dec 2, 2009
    #3
  4. Eric Miller

    Guest Guest

    most dslr sensors are capable of over 12 stops, limited by the d/a
    converter.
     
    Guest, Dec 2, 2009
    #4
  5. Eric Miller

    Bob Larter Guest

    [snip]

    Don't mind this loon. He has no idea what he's talking about, & delights
    in wasting everybody's time.
     
    Bob Larter, Dec 4, 2009
    #5
  6. Eric Miller

    Peter Chant Guest

    nospam wrote:

    Is this the limit? Then why did Fuji have their fancy "super" sensor with
    differing sizes of sites? I wonder why someone does not come out with a 16
    bit a/d chip?

    Pete
     
    Peter Chant, Dec 4, 2009
    #6

  7. Bob Larter's legal name: Lionel Lauer
    Home news-group, an actual group in the "troll-tracker" hierarchy:
    alt.kook.lionel-lauer (established on, or before, 2004)
    Registered Description: "the 'owner of several troll domains' needs a group where he'll stay on topic."

    <http://groups.google.com/groups/search?hl=en&num=10&as_ugroup=alt.kook.lionel-lauer>

    "Results 1 - 10 of about 2,170 for group:alt.kook.lionel-lauer."
     
    Bob Larter is Lionel Lauer - Look it up., Dec 4, 2009
    #7
  8. Eric Miller

    Guest Guest

    recent slrs use 14 bit d/a converters and some medium format backs have
    16 bit converters.
     
    Guest, Dec 4, 2009
    #8
  9. Eric Miller

    JimKramer Guest

    Reading all the way down to "RAW headroom" on the D3 link, 2nd paragraph:

    "As usual the default Adobe Camera RAW conversion delivers less dynamic
    range than JPEG from the camera (a more contrasty tone curve and very little
    noise reduction in shadows). Simply switching to 'Auto' in the ACR
    conversion dialog reaps huge rewards (we measured the result to have exactly
    12 stops of dynamic range), and in our tests with real world shots produced
    superb results with images that seemed to be over exposed beyond
    redemption."

    -Jim
     
    JimKramer, Dec 4, 2009
    #9
  10. Eric Miller

    Guest Guest

    dpreview tests dynamic range by comparing jpegs which is bogus, to say
    the least.
    scroll to the bottom where they measured 12 stops with raw.

    dxo labs measured it at 12.2 here:
    <http://www.dxomark.com/index.php/eng/Image-Quality-Database/Nikon/D3>

    and roger clark measured 13.7 here, but he measures the capability of
    the sensor itself, not the rest of the system.
    <http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
    ry/>
     
    Guest, Dec 4, 2009
    #10
  11. Eric Miller

    Eric Miller Guest

    Eric Miller, Dec 4, 2009
    #11
  12. Eric Miller

    Guest Guest

    if the pixels are the same size, the larger sensor will have more of
    them, which is also beneficial.
     
    Guest, Dec 4, 2009
    #12
  13. Eric Miller

    Ofnuts Guest

    Adding more bits makes sense only if you get meaningful data, i.e., if
    the low bits are still above noise level. IIRC each bit earned requires
    6dB more signal/noise power ratio. For 14 bits, the electronics should
    be as quiet as those of an audio device with 84db of signal/noise ratio,
    and for 16 bits if goes to 96dB. Not undoable, but not that easy.
     
    Ofnuts, Dec 5, 2009
    #13
  14. Eric Miller

    Guest Guest

    yes it is.

    <http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
    ry/>
     
    Guest, Dec 6, 2009
    #14
  15. Bzzzt.
    Sensor size *as such* has nothing to do with dynamic range.
    Pixel size, however ...
    Small sensors tend to have smaller pixels, that's all.
    http://clarkvision.com/articles/digital.sensor.performance.summary/index.html#data
    indicates that the sensors are perfectly capable of 12+ stops,
    based on their full well capacity and read noise, but excluding
    A/D noise (which is not really tied to the sensor, anyway).
    The rest of the page makes it clear that the camera performance
    is limited by
    a) 12 bit A/D converters,
    b) noise in A/D converters,
    c) *much* less improvement for A/D converters of 14 bit than would
    be assumed by the additional bits. High megaherz speeds seem
    to do that, at least for now.

    -Wolfgang
     
    Wolfgang Weisselberg, Dec 7, 2009
    #15
    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.