Are there any 3 CCD Digital Still Image Cameras .....

Discussion in 'Digital Photography' started by adminforto, Feb 22, 2007.

  1. adminforto

    adminforto Guest

    .... if so, please advice.
     
    adminforto, Feb 22, 2007
    #1
    1. Advertisements

  2. adminforto

    Rich Guest

    No. The reason for this method is to do tri-colour processing which
    is the most accurate method of recording colour and preserves
    resolution. Scientific CCD cameras use this, but they use a filter
    wheel/bank and take three exposures. Principally used in astronomy
    where most subjects are static.
    But there is nothing stopping someone from constructing a still image
    camera using three monochrome sensors and the proper dichroic filters.
    The images can then be combined to form a high resolution, accurate
    colour image.
     
    Rich, Feb 22, 2007
    #2
    1. Advertisements

  3. NO subjects in astronomy are static!
     
    Fred Garvin, Male Prostitute, Feb 23, 2007
    #3
  4. adminforto

    peter Guest

    This begs the question: Why do some consumer/prosumer camcorders use this
    kind of sensor but no still photo cameras do?
     
    peter, Feb 23, 2007
    #4
  5. Oh come on Fred, you know what was meant by that. Is your truss on a
    little too tight today? It can happen to the best of us.

    I remember the ol' college days of hanging a large heavy photocell
    vacuum tube-type apparatus at the end of an unbalanceable telescope,
    firing it up and pointing at our assigned target star, and sliding the
    U, B, and V filters in the optical path to get readings, then trying to
    find a patch of nearby "dark" sky to get more readings, then doing both
    again and again until we went mad or fell asleep. You young people today
    have it too EASY, with your computerized whatchamacallits and yer CCD
    gizmos.

    Sincerely, Abe Simpson, Cranky Astronomer
     
    Unclaimed Mysteries, Feb 23, 2007
    #5
  6. adminforto

    Paul Rubin Guest

    I think multiple ccd's is becoming obsolete in camcorders too. They
    were used in older camcorders because it was difficult to make large
    CCD's at that time, but it's gotten easier now. A prosumer 3ccd
    camcorder (Sony VX2100, about $2000) has three 1/3" CCD's for about 65
    sq mm of sensor area. Since it has no Bayer filter (the light is
    separated by a lossless dichroic prism) it has maybe as much photon
    collecting ability as a 130 sq mm Bayer sensor. But it's 4x as
    expensive as a Nikon D40 which has a 330 mm^2 sensor. And if you want
    a 3ccd camera with more sensor area (like 1/2" ccd's) you have to
    spend $5k or more. Basically as the ccd's get larger, the perfectly
    aligned mechanics and optics needed for a multiple ccd system
    apparently get extremely expensive. The latest ultra-high-end video
    cameras (see www.red.com) apparently now use single large ccd's
    instead of multiple ccd's.
     
    Paul Rubin, Feb 23, 2007
    #6
  7. The much higher resolution of the still cameras makes it much harder to
    adequately align the three CCDs and the beamsplitters. Enough harder
    that it's not worth it.
     
    David Dyer-Bennet, Feb 23, 2007
    #7
  8. adminforto

    Bill Tuthill Guest

    The old Minolta DSLR did. It was called the RD-175. 1997.
    http://www.sds.com/mug/rd175.html
     
    Bill Tuthill, Feb 23, 2007
    #8
  9. [A complimentary Cc of this posting was sent to
    Rich
    My understanding of 3 CCD sensors is that they use dichroic filters,
    which have a color-passthrough curve with very steep sides. AFAIU
    color rendition, this provides very bad match to the eye color
    sensitivity curves; this should imply that color accuracy of 3 CCD
    sensors should be horrible.

    One indirect confirmation of this theory is the tons of makeup one
    *must* take to participate in TV shows. Why this would be needed if
    the color rendition was "natural"? [One explanation I've heard was
    the heat; but this was from an anchor, who was not very technically
    savvy...]

    So: can one pinpoint holes in this theory?

    Thanks,
    Ilya
     
    Ilya Zakharevich, Feb 23, 2007
    #9
  10. adminforto

    Ben Brugman Guest

    That is because of wrong use of the word 'natural'; most people
    on tv look far better on tv than in the 'natural' world.

    In the 'natural' world most things can be improved with extra lights
    (because it's dark inside), with reflectors, because the sun is to
    harsh, with makeup to hide imperfections or to avoid shinis, with a
    hairdo etc. etc.

    This goes for Film, CCD and also for 3-CCD camera's.

    If the 3-CCD camera realy has small sensitivity in the three
    area's a lot of real colors would not become visible.
    (Orange for example would not show up).
    Furthermore the sensitivity of the CCD would be severely limited
    if the bandwith of the color becomes small.

    Hope this is helpfull.

    Ben.


     
    Ben Brugman, Feb 24, 2007
    #10
  11. [A complimentary Cc of this posting was sent to
    Ben Brugman
    Do not follow what you say here.

    My argument is: to get correct color rendition, the eye color
    sensitivity curves should be well-approximatable by linear combination
    of the sensor's 3 color sensitivity curves. AFAIU, the curve for
    dicroic mirror has large "flat" parts, and quite steep slopes between
    these parts. Therefore, the same would hold for any linear
    combination; therefore, one woun't be able to approximate eye's
    curves.

    Of course, this relies on my (almost missing) knowledge on how well
    one can "tune up" a pass-through-curve of a dicroic mirror...

    Yours,
    Ilya
     
    Ilya Zakharevich, Feb 24, 2007
    #11
  12. adminforto

    ben brugman Guest

    First my principal reaction was to the "tons of makeup".

    These tons of makeup are used with any camera technologie, so here the
    3 CCD camera is not distinct from other camera's.


    Because you said that the curve is very steep, I 'also' assumed that the
    curve would be very smal. So limited in sensitivity to different colors.
    (This limits the total sensivity as well.).
    This assumptions was not correct. I would think that dichroid mirrors
    are very color specific (smal in curve), but this is not based on anything.
    So a steep curve can still be wide, I did not realise that.

    Ben.
     
    ben brugman, Feb 26, 2007
    #12
  13. [A complimentary Cc of this posting was sent to
    ben brugman
    I was not speaking about *using* makeup. I spoke about *needing*
    makeup. I do not need makeup when I shoot people with my film cameras.
    I do not need makeup when I shoot people with my digital cameras.

    But what I heard, is that with "real" TV cameras, without makeup one
    gets unnatural face colors.
    My impression is exactlty the opposite. If my "theory" is correct,
    the *only* advantage of 3CCD cameras is that they (potentially)
    capture ALL photons coming through the lens (as opposed to RGB Bayer
    sensor, where about 2/3 of photons are lost in the filters).

    The price is (conjectural) bad color rendition of 3CCD...
    My impression was that dichroic mirror have narrow region where they
    *switch* from "pass-through" to "reflect" mode. Outside of these
    regions, the spectral curve will be in 0-or-1 mode (with steep
    transition "switch" regions in between). But I have no data on actual
    widths of these region.

    Hope this helps,
    Ilya
     
    Ilya Zakharevich, Feb 27, 2007
    #13
  14. adminforto

    Ben Brugman Guest

    Makeup and television that's a long history. I do not think 3 CCD camera's
    changed all that much.

    If all photons get or transmitted or reflected and if in the end they hit
    a sensor, yes you would have a more sensitive setup. You are probably
    right in that the total amount of photon's collected will be more.
    (I was just incorrectly thinking about narrow pass filters. Not about
    splitting
    the photons withouw loosing them.).
    Thanks for clearing that up.

    ben
     
    Ben Brugman, Feb 27, 2007
    #14
  15. adminforto

    J. Clarke Guest

    The need for makeup in television production has nothing to do with
    the camera (the details might, but not the overall need). Makeup has
    been needed by performers just about as long as there have _been_
    performers. Even neolithic hunter-gatherers use it.
     
    J. Clarke, Feb 27, 2007
    #15
  16. [A complimentary Cc of this posting was sent to
    J. Clarke

    If the meaning of this is "makeup was needed with pre-CCD cameras
    too", then I do not see how it relates to the question at hadn:
    AFAIK, pre-CCD cameras were using dicroic splitters as well.
    Let me repeat my question again: what the anchor said was

    "face colors are wrong without makeup".

    I do not know how savvy this person is in technical details. So,
    essentially, my question was: is it true, and does not it reflect bad
    color reproduction?

    And I do not think your answer is helping...

    Hope this helps,
    Ilya
     
    Ilya Zakharevich, Feb 28, 2007
    #16
  17. adminforto

    J. Clarke Guest

    Watch live coverage on TV news and you'll know the answer to your
    question. For that matter, watch live coverage of sporting
    events--you think the TV crew goes out and dabs makeup on the
    basketball players or swimmers?

    In a studio the camera is very, very unforgiving, whether it's film or
    video. Apply something to kill the reflection off of Katie Couric's
    nose and what you used for that won't match the rest of her face, so
    you have to blend it in. You want to do all that stuff properly
    beforehand, not try to whip it up during a 30 second commercial break.
    And moving a light or telling her to hold her head a certain way isn't
    an option like it is in portrait photography where you take the shot
    and you're done instead of rolling camera for 20 minutes.

    But even in high level still photography makeup is used--you really
    think that fashion models and centerfolds don't get a thorough
    workover by professional makeup artists with touchup during the shoot?
    If it wasn't considered _necessary_ then the producers would not be
    paying for it to be done.
     
    J. Clarke, Feb 28, 2007
    #17
  18. [A complimentary Cc of this posting was sent to
    J. Clarke
    Thanks a lot; this "reflection" argument is a killer.

    So my "makeup needed" evidence is a bogus. But this still lives the
    initial question intact: what is the ACTUAL spectral curve of dicroic
    mirrors used for 3CCD (as opposed to PopSci one, which has steep
    slopes, and is otherwise flat)?

    Thanks,
    Ilya
     
    Ilya Zakharevich, Mar 6, 2007
    #18
  19. adminforto

    Ben Brugman Guest


    If only mirror are used to split the original 'beam' of light for the three
    different sensors. (So no additional filtering is done, and the sensors
    are not special in a way so fairly spectral neutral).

    THEN, it doesn't matter to much how steep or what the shape of the curve
    is because the curves together for the three sensors must make a 100 %
    flat line over the complete spectrum. (limited by the sides offcourse).

    With traditional filtering (and our eyes) there are three curves (bel
    shaped)
    overlapping, added together you get something like a camel with three humps,
    but not a 100 % value and not a flat shape.

    So my conclusion is that mirrors allone can not give a curve resembling the
    eye. Most people won't notice this in the end result, because we are very
    capable of addapting our eyes if the light changes, this adaption also
    occurs
    when the sensitivity changes because of the three CCD. But if you compare
    color cards, there should be very noticible difference. (If the color cards
    are made
    with real colors is).

    With the 3 humps there are some colors which we are less sensitive to, the
    colors
    in the lower parts. With the 'flat' curve these colors 'produce' more light.

    Ben





    ben
     
    Ben Brugman, Mar 7, 2007
    #19
  20. [A complimentary Cc of this posting was sent to
    Ben Brugman
    Bulls. To get reasonable colors, you need to able to approximate
    spectral sensitivity curves of cones with linear combinations the
    effective spectral sensitivity curves of 3 sensor.
    Sorry, but I see nothing *in your post* to lead to such a conclusion.
    [But as I said, the tiny pieces of information I have ALSO lead to
    such a conclusion ;-]

    Hope this helps,
    Ilya
     
    Ilya Zakharevich, Mar 7, 2007
    #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.