Got <140 bit DR Image?

Discussion in 'Australia Photography' started by BradGuth, Jan 28, 2009.

  1. BradGuth

    BradGuth Guest

    What else can your camera see if its FOV is centered upon a TIG
    welding arc?

    There’s truthworthy radar imaging of 4 to 36 looks per composite
    pixel, then always good old conventional Kodak film with its nonlinear
    spectrum and limited DR(dynamic range), and then there’s some rather
    nifty CCD/CMOS imaging alternatives that requires by far the most
    deductive observationology, especially if there are extreme DR and
    various colour/hue saturations that need to be taken into account,
    along with those unavoidable shadows and sometimes surface glare/
    mirror like reflective considerations.

    The finite number of pixels per given FOV isn’t the only truth worthy
    holy grail of quality imaging, especially if there are issues of
    dynamic range and color/hue saturation to consider. (usually there
    are)

    Instead of the Kodak/Fuji 10~11 db worth of DR, or the traditional <14
    bit of high end cameras, whereas we now have <140 db/bit of extended DR
    (dynamic range) via CMOS imaging that’s affordably off the shelf (so
    to speak), plus absolutely terrific optical options in bandpass
    filters, that'll look directly into a vibrant star (such as Sirius)
    without losing the crisp detailed surface, horizon and of whatever is
    in nearby orbit. Actually, nearly that kind of ccd/cmos pixel dynamic
    range has existed and been in our public funded DARPA and NASA use for
    quite some time, just not allowed to see the public light of day.

    http://www.ims.fraunhofer.de/uploads/media/cmos_image_sensor_arrays_en_07.pdf

    Camera A/D imaging converters are sort of where the technology buck
    stops (so to speak), although more powerful and parallel CPUs tend to
    make a 128 DB image doable, though a single scan 64 DB format might be
    a practical upper limit/scan that only supercomputers could manage,
    and perhaps 32 DB for stuff that’s taken on the fly (so to speak).
    However, even by using a 16 DB A/D with such an extended dynamic range
    capable CMOS imager is going to knock some eye-candy socks off, and if
    need be via taking multiple scans, one of each 16 DB portion of the
    available dynamic range should be more than good enough (4 scans could
    extract 64 DB worth of dynamic range), although you’d likely need a
    Blue Ray DVD for storing each image, especially if taken in full color
    format, and worse yet if that spectrum happens to include some UV and
    IR.

    Keeping this little Fraunhofer CMOS imager within its thermal specs
    (that’s wider than most as is), would be the only application
    requirement in need of R&D for its custom packaging.

    ~ BG
     
    BradGuth, Jan 28, 2009
    #1
    1. Advertisements

  2. BradGuth

    BradGuth Guest

    True, but at the 1/40,000th of a second scan and even with 4 bracket
    shots there's practically nothing else to be seen in that image.

    Take another look at the results of a true 140 bit CMOS imager.

    http://www.ims.fraunhofer.de/uploads/media/cmos_image_sensor_arrays_en_07.pdf

    ~ BG
     
    BradGuth, Jan 28, 2009
    #2
    1. Advertisements

  3.  
    John McWilliams, Jan 28, 2009
    #3
  4. BradGuth

    Mr.T Guest

    Better hope nothing moves during those 4 shots then. Not even the arc!


    Isn't that what the other 3 shots are for? Look up "HDR imaging".

    MrT.
     
    Mr.T, Jan 28, 2009
    #4
  5. BradGuth

    BradGuth Guest

    With the 140 db CMOS imager it's accomplished in one scan. In other
    words, looking directly at Sirius ABC is no longer an issue of not
    seing everything that's within the given FOV.

    If need be, there are even cameras that can average 2<256 scans per
    composite image, so even that's covered.

    ~ BG.
     
    BradGuth, Jan 28, 2009
    #5
  6. BradGuth

    Mr.T Guest

    Of course, just let me know when a similarly equipped camera is available to
    the public at a couple of thousand dollars though.
    Otherwise what's your point? :)

    BTW the incorrect/ad-hoc transposition of "bits" and "DB" makes the OP both
    wrong and unreadable in any case!

    MrT.
     
    Mr.T, Jan 28, 2009
    #6
  7. BradGuth

    J. Clarke Guest

    Then there's the "<". My old Coolpix 990 has considerably < 140 dB
    and why one would want a camera with < 140 bits is a mystery to me.
     
    J. Clarke, Jan 28, 2009
    #7
  8. BradGuth

    BradGuth Guest

    If you can't figure out the obvious point, then what's the point?
    My dyslexic mind isn't what it used to be, so sue me. In spite of
    myself, you know exactly what I'd intended to convey. The A/D bits of
    digital imaging was similar enough to the range of DR in db. Most
    scientific cameras are A/D limited as to creating 16 bit image data,
    although the 140 db worth of CMOS imager DR itself is simply pretty
    hard to ignore.

    BTW, bet there's a wrong word used somewhere in the constitution, or
    in the bible, so either of those are thereby "unreadable in any
    case!"?

    ~ BG
     
    BradGuth, Jan 28, 2009
    #8
  9. BradGuth

    BradGuth Guest

    What else can your camera see if its FOV is centered upon a TIG
    welding arc, or while peering through a powerful telescope at Sirius
    ABC? (140 db DR instead of 140 bit, via Mr.T)

    There’s always the most truthworthy 8 bit radar imaging of 4 to 36
    looks per composite pixel, then always we have good old conventional
    Kodak or Fuji film with its nonlinear spectrum and rather limited DR
    (dynamic range), and then there’s some rather nifty CCD/CMOS imaging
    alternatives that requires by far the utmost of our best deductive
    observationology, especially if there are extreme DR and various
    colour/hue saturations that need to be taken into account unless
    excluding suits your fancy, along with those unavoidable shadows and
    sometimes surface glare/mirror like reflective considerations.

    The finite number of pixels as per any given FOV isn’t the only
    truthworthy holy grail of quality and truthworthy imaging, especially
    if there are complex issues of dynamic range and color/hue saturation
    to consider(usually there are), meaning that it’s sometimes a whole
    lot better having extended DR rather than any given raw number of
    pixels.

    Instead of the Kodak/Fuji 10~11 db worth of DR, or the traditional <14
    bit of high end cameras, whereas we now have <140 db of extended DR
    (dynamic range) via CMOS imaging that’s affordably off the shelf (so
    to speak), plus absolutely terrific optical options in bandpass
    filters, that'll look directly into a vibrant nearby star (such as
    Sirius) without losing the crisp detailed surface, horizon and of
    whatever is in nearby orbit, such as Sirius C along with whatever
    moons. Actually, nearly that kind of ccd/cmos pixel dynamic range has
    existed and been in our public funded DARPA and NASA use for quite
    some time, just not ever allowed to see the public light of day.

    http://www.ims.fraunhofer.de/uploads/media/cmos_image_sensor_arrays_en_07.pdf

    Camera A/D imaging converters are sort of where the technology buck
    stops (so to speak), although more powerful and parallel CPUs tend to
    make a 128 bit image doable, though a single scan 64 bit format might
    be a practical upper limit/scan that only supercomputers could manage,
    and perhaps 32 bit for stuff that’s taken on the fly (so to speak).
    However, even by using a 16 bit A/D with such an extended dynamic
    range capable CMOS imager is going to knock some eye-candy socks off,
    and if need be via taking multiple scans, such as one of each 16 DB
    portion of the available dynamic range should be more than good
    enough (4 scans could extract 64 DB worth of dynamic range), although
    you’d likely need a Blue Ray DVD for storing each image, especially if
    taken in full color format, and worse yet if that spectrum happens to
    include some UV and IR.

    Keeping this little Fraunhofer CMOS imager within its thermal specs
    (that’s wider than most as is), would be the only application
    requirement in need of R&D for its custom packaging for also blocking
    as much other unwanted radiation as possible.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #9
  10. BradGuth

    Mr.T Guest

    Gee even the best SLR's only do ~14 bits per channel, or 42 bits per pixel
    atm.
    But there still seem to be a lot of people who do want (or will happily
    accept) "<140 bits DR".

    MrT.
     
    Mr.T, Jan 29, 2009
    #10
  11. BradGuth

    Mr.T Guest

    Not so, or we wouldn't be arguing where you are technically wrong and/or
    technically obtuse.
    Not as an equivalent figure they are not, 1 bit =~6dB and therefore dB is
    not the same as the number of stops either.
    There are sample shifting techniques that can use 16 bit A/D's to provide 32
    bit data, more than enough for 140dB sensors.

    MrT.
     
    Mr.T, Jan 29, 2009
    #11
  12. BradGuth

    Guest Guest

    i've seen it called sensel, for sensor element.
    is 22 bit enough? :)
    <http://www.pentax.jp/english/imaging/digital/slr/k10d/feature.html>

    however, phil is a bit skeptical (as are many others):
    <http://www.dpreview.com/reviews/pentaxk10d/>
     
    Guest, Jan 29, 2009
    #12
  13. BradGuth

    Mr.T Guest

    If it's just luminosity then it's often referred to as an "alpha channel",
    and yes 14 bits per single channel pixel is..... 14 bits.
    Can you explain this differentiation further, I'm not sure what you are
    getting at, but have no experience with astronomical imaging. Doesn't a
    sensor location map to a picture element location as with other cameras?
    Fortunately it's not necessary then.

    MrT.
     
    Mr.T, Jan 29, 2009
    #13
  14. BradGuth

    BradGuth Guest

    Thanks for that informative feedback, as I certainly didn't understand
    what it would take in order to fully realize the maximum 140 db
    imaging capability. This should mean our consumer video products
    should soon incorporate at least this much or better DR, and thus
    images of Sirius shouldn't be over saturated to the point of losing a
    good look-see at Sirius C, especially if certain optical bandpass
    elements are utilized.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #14
  15. BradGuth

    BradGuth Guest

    The matter of fact that anyone can systematically exclude as much DR
    (dynamic range) as you like, so as to minimize whatever contrast or
    color/hue saturation in order to exclude other information that’s
    above or below a specified range, isn’t new. It seems our NASA has
    for decades used this method whenever there were other secondary
    planets and/or major stars within a given FOV(frame of view) of
    whatever primary asteroid, planet or moon, such as their having
    excluded Venus, Mars, Jupiter and Saturn whenever found above the
    physically dark as coal Selene/moon horizon.

    Of course there are a few exceptions where they’d forgotten to do this
    prior to getting published, so as even that of the limited DR of film
    easily shows other planets plus a few of the brightest stars besides
    just depicting Earth outside of our physically dark moon. The fairly
    old (1994) Clementine mission offers a perfectly good DR example of
    that era CCD recording our moon, Venus and sun all in the same FOV
    that got published. Of course the public hardly ever gets to see but
    at most 0.1% of any given mission images, and even less of the other
    science, so it’s hard to tell how many original or better examples
    exist outside of their terrific mineralogy imaging that’s only now
    getting outdone by ISRO.

    Low and higher resolution versions of the moon, Venus and sun (notice
    the Earthshine illuminated moon, and the more proper color/hue
    saturation of Venus when viewed from outside our atmosphere):
    http://goes.gsfc.nasa.gov/pub/goes/clementine.venus.sun.emoon.jpeg
    http://www.nrl.navy.mil/NewsRoom/images/clem.jpg
    http://www.nrl.navy.mil/NewsRoom/images/clem.tif

    Now try to imagine a FOV of our moon plus Earth and Venus, as could
    easily have been obtained via Apollo missions (via orbiting or EVAs)
    A11, A14 and A16, as having easily excluded the sun and thereby having
    a full color saturation and terrific contrast of just Earth and Venus
    above the dark as coal lunar horizon. Oddly that never happened.

    Future missions that’ll have the 140 db or better scope of DR and
    faster scans should offer impressive results, especially of the
    complex mineralogy of our moon. Too bad the new and improved Hubble
    imagers are not this good, but other orbiting cameras should equal or
    likely exceed this level of DR, and subsequently getting terrific
    images of Sirius C along with whatever moons.

    According to MrT, “There are sample shifting techniques that can use
    16 bit A/D's to provide 32 bit data, more than enough for 140dB
    sensors.”

    ~ Brad Guth Brad_Guth Brad.Guth BradGuth BG / “Guth Usenet”
     
    BradGuth, Jan 29, 2009
    #15
  16. BradGuth

    BradGuth Guest

    True, as you'll most likely never have to make use of the full DR,
    although whenever the sun or that of a directed glare of the sun gets
    in the FOV, it certainly would be nice not having the image entirely
    saturated with solar photons, as well as peering into deep shadows at
    the same time.

    Your 6 bits per cmos imager DB even sounds terrific. I suppose what's
    next will be the 256 db imager and 21 bit A/D processor, along with a
    terabyte in media storage capacity, plus 100:1 zoom optics including
    many nifty bandpass filter options, all for under a few grand.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #16
  17. BradGuth

    BradGuth Guest

    Perhaps no breath holding is necessary.

    Using 2 or 4 quick scans per image could make the 16 bit ADC more
    viable for the 140 db cmos imager. There's at least one commercial/
    science camera that can selectively average up to 256 scans per output
    image.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #17
  18. BradGuth

    BradGuth Guest

    “To actually capture a 140 dB dynamic range requires a minimum of 23
    bits per sensor data sample. We don't want to hold our breath waiting
    for a 24 bit ADC?”

    Perhaps no breath holding is going to be necessary.

    Using 2 or 4 quick scans per image could make the existing 16 bit ADC
    more viable for utilizing the 140 db cmos imager. There's at least
    one commercial/science camera that can selectively average from 2 <
    256 scans per output image, and that’s purely software.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #18
  19. BradGuth

    BradGuth Guest

    “To actually capture a 140 dB dynamic range requires a minimum of 23
    bits per sensor data sample. We don't want to hold our breath waiting
    for a 24 bit ADC?”

    Perhaps no breath holding is going to be necessary.

    Using 2 or 4 quick scans per image could make the existing 16 bit ADC
    more viable for utilizing the 140 db cmos imager. There's at least
    one commercial/science camera that can selectively average from 2 <
    256 scans per output image, and I believe that’s purely software.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #19
  20. BradGuth

    BradGuth Guest

    At 6 db per bit, we may need a 23+ bit ADC, or perhaps not.
    “To actually capture a 140 dB dynamic range requires a minimum of 23
    bits per sensor data sample. We don't want to hold our breath waiting
    for a 24 bit ADC?”

    Perhaps no breath holding is going to be necessary.

    Using 2 or 4 quick scans of “sample shifting techniques” per composite
    image that could make the existing 16 bit ADC more viable for
    utilizing the 140 db cmos imager (a 24 bit composite needing as little
    as only two 12 bit scans). There's at least one off-the-shelf
    commercial/science camera that can selectively average from 2 < 256
    scans or frames per composite output image. I believe that’s
    accomplished either via BASLER proprietary firmware or purely
    software, and I’m certain by now there are much higher resolution
    cameras capable of the same.

    http://www.baslerweb.com/beitraege/news_en_77572.html

    http://www.photonicsonline.com/arti...ic-Features-And-A-Lot-More-0001?VNETCOOKIE=NO

    A 16 bit ADC using the dual scanned and sample shifted composite image
    as its output, as such could muster up enough collective processing in
    order to utilize an imager that’s offering <192 db of DR.

    ~ BG
     
    BradGuth, Jan 29, 2009
    #20
    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.