Samsung sensor patent application

Discussion in 'Samsung' started by RichA, Jun 4, 2008.

  1. Your examples have an error:
    The S/N of 19.53 photons should only be 4.4. Your example shot noise image
    has much higher S/N.

    For comparison, you should show the case for ISO 1600 too.
    Roger
     
    Roger N. Clark (change username to rnclark), Jun 15, 2008
    #41
    1. Advertisements

  2. Of course; that is as expected. The 19.53 photons is at the clipping
    point, so the sigma is naturally about 0.6x what it would be, inclipped!
    This implies no error in the projected results, because, as always, we'd
    be looking at some clipping. The low-frequency sinusoidal sections show
    what unclipped "midtones" would be like.
    Do you mean ISO 1600 read noise levels, or an ISO 1600 exposure index?

    I didn't upload samples of read noises of 3 or 4 elctrons, but I've
    looked at them, and they are a little better than the 26 ones at lower
    exposure indices, but they have the same blanketing effect at the higher
    EIs.

    One thing I do need to fix is to change the method of avoiding division
    by zero; I should just have a separate conditional case for 0 photons
    signal. When I get a chance, I'll redo the Mathcad pages to include both
    ends of the read noise range. maybe even include samples of real noise
    (which, unlike shot noise, can be had from a real camera).
     
    John P Sheehy, Jun 21, 2008
    #42
    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.