Latest Kodak sensors have superior Dynamic Range! (on the paper)

Discussion in 'Kodak' started by RiceHigh, Mar 2, 2006.

  1. The charge, Q, integrates on a capacitor, the charge well, C to produce
    a voltage.

    V = Q / C

    So the voltage is linear... until it saturates.
     
    Kennedy McEwen, Mar 4, 2006
    #21
    1. Advertisements

  2. RiceHigh

    Bryan Olson Guest

    I'm missing something. Power increases as the square of
    Voltage. If the output Voltage is proportional to the
    charge, which is proportional to the light power, then
    the power coming out in the Voltage and current increases
    as the square of the power going in as light. You have the
    charge inducing the Voltage without another power source.
     
    Bryan Olson, Mar 4, 2006
    #22
    1. Advertisements

  3. Power is power, the output of one thing is merely the input of
    another.

    I assure you that input power signals *do* have "a current factor".
    It's a simple matter of what the input impedance of a device happens
    to be...
    That is confused. What "input power" are you talking about?
    Photons? Electrons? A CCD is a *sensor*, it does not have an
    electrical input signal. It has an electrical output signal,
    which is what we are talking about.
    It's all the same thing, and anyone who understands one
    understands the other, and can interchange them in appropriate
    places. When discussing a CCD alone, absent the rest of the
    processing system, a raw ratio is perhaps more useful. But when
    the rest of the system is part of the discussion, dB makes a
    great deal more sense; and that is particularly true when
    comparing fstops of exposure with the dynamic range of the CCD.
    That is simply false, and demonstrating it is easy. If a given
    camera has a CCD with an S/N in the range of 30-40 dB as you are
    claiming, the range of fstops that could be recorded in the
    images produced would be 5 to 7 (30-42 dB of dynamic range).
    Yet not only do *all* of the major camera manufacturers list the
    dynamic range of the CCD's they use as in the 60-75 dB range,
    the also can be *measured* to record images that record a range
    of illumination from more than 7 to more than 10 fstops.

    What that says is simple: They all have CCD's with a dynamic
    range that is more than you are claiming the best of them would
    actually be, using your formula.

    Clearly the correct formula is 20 log (S/N), and when it is used
    *all* of the calculations come out correct.
    There is only one definition, but they did give different ways
    to calculate it with *different* *parameters*. With
    "intensity", "voltage", "pressure" the correct multiplier is 20,
    and with power it is 10. That does *not* generate a different
    dB, it merely says that voltage and power are not the same
    thing.

    Indeed E = IR and P = IE demonstrate exactly what that
    difference is.
     
    Floyd L. Davidson, Mar 4, 2006
    #23
  4. RiceHigh

    Scott W Guest

    Bryan Olson wrote:

    Photocells are call square law detectors for just this reason. The
    power source is the charging voltage that recharges the capacitance of
    the photo sites.

    Scott
     
    Scott W, Mar 4, 2006
    #24
  5. RiceHigh

    bjw Guest

    Floyd. I think we're talking past each other. A CCD has an
    input signal from the light, and it's perfectly reasonable to
    talk about this as a power - the light energy incident in a
    given exposure time. I assumed that this is what
    people like to refer to as the dynamic range of the sensor -
    the range of input values that it can sense.

    Apparently the dynamic range spec is the dynamic range
    calculated from the output voltage. In that case, I give up,
    using 20 log (voltage max / voltage min) is correct and
    the dynamic range of the output power is 75 dB.

    I'm only complaining about this because I think people
    like to use dynamic range to talk about the range of
    input luminances the CCD can sense, and that range is
    not 75 dB in _input_.
    What you're showing here is that the manufacturer's
    spec sheets are all consistent in using 20 log (S/N),
    not whether that formula means input or output. I have
    no disagreement with the fact that the CCD under
    discussion has a dynamic range of 100,000/18 - I may
    have said that in my original post.
     
    bjw, Mar 4, 2006
    #25
  6. The dynamic range of the sensor, which is what it can sense, is
    a measure of the *output* electrical signal.
    Nothing else makes sense.
    Absolutely true.
    That is *NOT* true. The sensor *output* is the only way to
    determine what input range can be sensed. The input exposure
    range can of course be varied from far less than is sensed to
    far more than is sensed... but the definition of "sensed" is
    that a change in input results in a change in the output.
    It clearly shows that the sensed range matches the range they
    are calculating using 20 log (S/N).
    What you have been claiming is that a dynamic range indicated by
    100000/18 is 37.5 dB. That is not true. It is clearly twice
    that, just as everyone says. If it were 37.5 dB the camera
    would not be able to record an image with more than

    37.5
    ---- = 6.2+ fstops
    6

    And in fact there is no question that typically digital cameras
    are recording between 8 and 11 fstops and are being described
    as having a dynamic range of between 60 and 75 dB using the
    *correct* formula.

    Your formula simply is not the right one.
     
    Floyd L. Davidson, Mar 4, 2006
    #26
  7. RiceHigh

    JPS Guest

    In message <>,
    Not necessarily. The highest exposure level on the sensor before
    saturation is recorded at the lowest ISO, whereas the lowest usable
    sensor signal (whatever your standard may be) may only be obtained with
    amplification; i.e., a higher ISO.
    --
     
    JPS, Mar 4, 2006
    #27
  8. Yes, necessarily. (Anything else is *impossible*.)
    It is *still* being determined from the output voltage.

    I'm not sure your statement means what you think it does anyway.
    Keep in mind that "obtained with amplification" amplifies
    *everything* the same -- highs, lows, and noise -- it does not
    change the dynamic range of the sensor. It might change the
    operating point of the codec that digitizes the output though,
    which still doesn't change the dynamic range of either the
    sensor or the codec, but certainly can result in different
    results. (Granted that if the operating point for one is moved
    off the upper or lower end of the other, the dynamic range of
    the combination is compressed.)
     
    Floyd L. Davidson, Mar 4, 2006
    #28
  9. RiceHigh

    JPS Guest

    In message <>,
    They are on a completely different scale, though, and may not both be
    possible at one level of amplification, given the limitations of
    digitization hardware..
    It changes the DR of the *output*, though, in a single capture.
    In other words, you had to make it look like I was wrong, even though I
    wasn't. That's pretty much what I expected, when I saw your name.
    --
     
    JPS, Mar 4, 2006
    #29
  10. Your statement "not necessarily" apparently applied to something
    that had not been stated, as does your following discussion.

    It is irrelevant to the discussion you inserted it into.
    Post sensor amplification cannot change the the dynamic range of
    the sensor output.
    So if you aren't able to make a technical point you'll try
    gratuitous insults?

    Compressing the image (which is not what was being discussed) is
    a very different thing than reducing the dynamic range of the
    sensor (which is what was being discussed).
     
    Floyd L. Davidson, Mar 5, 2006
    #30
  11. More pointless detail that does *not* support the false claim
    that you made.

    And please note that any manufacturer who wants to go to 14 or
    even 16 bit digitization hardware can (and does) at any time.
    It does *not* change the dynamic range of the *sensor output*,
    which is what we were talking about and which you claimed was
    true.
    You were wrong then, you are wrong now, and I can't see any
    indication that you are going to change that. The fact that
    some unrelated factoid is true does not change what you said to
    something that is right. Even if you did see my name, and
    shivered.
     
    Floyd L. Davidson, Mar 5, 2006
    #31
  12. RiceHigh

    Bryan Olson Guest

    Ah, thanks.
     
    Bryan Olson, Mar 5, 2006
    #32
    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.