Binary Pixel and High Dynamic Range Imaging

Discussion in 'Digital Cameras' started by Eric Stevens, Apr 15, 2013.

  1. Eric Stevens

    Eric Stevens Guest

    Eric Stevens, Apr 15, 2013
    #1
    1. Advertisements

  2. Eric Stevens

    Savageduck Guest

    Savageduck, Apr 15, 2013
    #2
    1. Advertisements

  3. Eric Stevens

    Eric Stevens Guest

    Eric Stevens, Apr 16, 2013
    #3
  4. Eric Stevens

    RichA Guest

    RichA, Apr 16, 2013
    #4
  5. Oh, they just let the pixels overflow, empty them then and
    count how often that happens. Probably using pixels that
    overflow whenever a photon or three are captured, compared to
    our many 10's of thousands sensors.

    I had similar ideas, (count the time to overflow, very bright
    pixels overflowing really fast), but (unfortunately) didn't
    get the step to 'multiple overflows', which solves certain
    problem with non-static scenes. Oh well, I'm not a professional
    sensor designer.

    -Wolfgang
     
    Wolfgang Weisselberg, Apr 17, 2013
    #5
  6. Eric Stevens

    Alfred Molon Guest

    I wonder how the noise performance is, because perhaps the step of
    discharging the pixel could introduce some inaccuracies or the threshold
    when the pixel is discharged perhaps is not so exactly measured. These
    inaccuracies would mean additional noise.
     
    Alfred Molon, Apr 18, 2013
    #6
  7. Eric Stevens

    David Taylor Guest

    If you have so many photons you're getting overflows, does noise matter?
    I suppose it depends what the overflow count is, though....
     
    David Taylor, Apr 18, 2013
    #7
  8. Eric Stevens

    Alfred Molon Guest

    If that is so simple, why don't all cameras use such binary pixel
    sensors?
     
    Alfred Molon, Apr 18, 2013
    #8
  9. Eric Stevens

    Eric Stevens Guest

    Because you have to do all the processing for each pixel in the brief
    period that the shutter is open. You have to provide each pixel with a
    device to detect when they are full, to then empty the pixel before
    the next photon arrives and to keep track of how many times this
    occurs. This may happen several times during the one exposure.

    There is the further complication that not all pixels will be
    over-filled in this fashion and it is necessary that they each be able
    to function properly in low light.

    From the name 'binary pixel' they may even be counting individual
    photons as they arrive.

    I suspect that the sensor will be multilayered with some kind of
    high-speed processor behind each pixel. Rambus is probably one of the
    better equipped companies to undertake this kind of work.
     
    Eric Stevens, Apr 19, 2013
    #9
  10. Same as A/D converting. There's also noise added and
    inaccuracies introduced.
    That only matters if that influences the end result more than
    tolerable as defined by the maker and sold to the customer
    (e.g. smartphone makers).
    On the other hand you are not limited at all by the limited
    well size.

    -Wolfgang
     
    Wolfgang Weisselberg, May 2, 2013
    #10
    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.