Dark Frame Subtraction - any experts out there?

Discussion in 'Digital Photography' started by mark.thomas.7@gmail.com, Jan 22, 2007.

  1. Guest

    Having just collected a few Comet McNaught pics on a digital that does
    not include DSF (Fuji S9500), I'm interested in having a go. The 9500
    did very well on all of the images that were <15 seconds, but there are
    a couple that I went for longer exposures to try to get the best
    possible result in showing the faint comet tail, and they have suffered
    as a result.

    So before I start playing with my images (tifs and raw) and subtracting
    the empty frame I took... Are there any special/easy techniques that
    folk have discovered that might help, or is it just a case of.. well,
    just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    works best? I did a quick bit of research, and noticed a few
    references to using LAB mode somehow, but it was unclear what they did
    or how that helped.

    (And yes, I've had a quick try with noise reduction software, but to
    get a decent result there is too much compromise - I lose some of the
    faint stars and traces of the comet.)

    Any help would be appreciated, so I can avoid my least favorite
    saying...

    "Don't want to re-invent the wheel"


    Thanks,

    mark t
     
    , Jan 22, 2007
    #1
    1. Advertising

  2. Charles Guest

    On 22 Jan 2007 01:27:10 -0800, wrote:

    >Having just collected a few Comet McNaught pics on a digital that does
    >not include DSF (Fuji S9500), I'm interested in having a go. The 9500
    >did very well on all of the images that were <15 seconds, but there are
    >a couple that I went for longer exposures to try to get the best
    >possible result in showing the faint comet tail, and they have suffered
    >as a result.
    >
    >So before I start playing with my images (tifs and raw) and subtracting
    >the empty frame I took... Are there any special/easy techniques that
    >folk have discovered that might help, or is it just a case of.. well,
    >just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    >works best? I did a quick bit of research, and noticed a few
    >references to using LAB mode somehow, but it was unclear what they did
    >or how that helped.
    >
    >(And yes, I've had a quick try with noise reduction software, but to
    >get a decent result there is too much compromise - I lose some of the
    >faint stars and traces of the comet.)
    >
    >Any help would be appreciated, so I can avoid my least favorite
    >saying...
    >
    >"Don't want to re-invent the wheel"
    >
    >
    >Thanks,
    >
    >mark t



    You might try posting the question in alt.binaries.pictures.astro
     
    Charles, Jan 22, 2007
    #2
    1. Advertising

  3. There are two important sources of noise in a digicam, fixed pattern
    noise and Johnson noise. Noise reduction software is designed to only
    take out the Johnson noise, which varies from image to image.

    While the fixed pattern noise (due to different sensitivities of
    various pixels) is pretty low in modern digicams, it is not zero, and
    can be a problem in astronomical photography. Take a picture of a
    PERFECTLY uniform background (not the easiest to find). Photoshop has
    masking functions which can be used to subtract one scene from another,
    but it requires some math and is not the easiest process for the
    inexperienced. PSP has a similar function.


    wrote:
    > Having just collected a few Comet McNaught pics on a digital that does
    > not include DSF (Fuji S9500), I'm interested in having a go. The 9500
    > did very well on all of the images that were <15 seconds, but there are
    > a couple that I went for longer exposures to try to get the best
    > possible result in showing the faint comet tail, and they have suffered
    > as a result.
    >
    > So before I start playing with my images (tifs and raw) and subtracting
    > the empty frame I took... Are there any special/easy techniques that
    > folk have discovered that might help, or is it just a case of.. well,
    > just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    > works best? I did a quick bit of research, and noticed a few
    > references to using LAB mode somehow, but it was unclear what they did
    > or how that helped.
    >
    > (And yes, I've had a quick try with noise reduction software, but to
    > get a decent result there is too much compromise - I lose some of the
    > faint stars and traces of the comet.)
    >
    > Any help would be appreciated, so I can avoid my least favorite
    > saying...
    >
    > "Don't want to re-invent the wheel"
    >
    >
    > Thanks,
    >
    > mark t
     
    Don Stauffer in Minnesota, Jan 22, 2007
    #3
  4. JC Dill Guest

    On 22 Jan 2007 01:27:10 -0800, wrote:

    >with my images (tifs and raw) and subtracting
    >the empty frame I took... Are there any special/easy techniques that
    >folk have discovered that might help, or is it just a case of.. well,
    >just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    >works best?


    <http://mysite.verizon.net/~vze4r2c2/Astro/AstroDigiCamFAQ.html#processingsoftware>

    HTH

    jc

    --

    "The nice thing about a mare is you get to ride a lot
    of different horses without having to own that many."
    ~ Eileen Morgan of The Mare's Nest, PA
     
    JC Dill, Jan 23, 2007
    #4
  5. writes:

    > Having just collected a few Comet McNaught pics on a digital that does
    > not include DSF (Fuji S9500), I'm interested in having a go. The 9500
    > did very well on all of the images that were <15 seconds, but there are
    > a couple that I went for longer exposures to try to get the best
    > possible result in showing the faint comet tail, and they have suffered
    > as a result.


    > So before I start playing with my images (tifs and raw) and subtracting


    You have RAW, so you are in good shape. You need to work on the pixel data
    numbers, not on the `image'.

    There are two noise sources to correct out. Bias noise, and dark current.
    Bais frames you can get by shooting off a string of shots with the lens
    cap on to keep all the dark in ;) Then average them and subtract that
    from your DN data. Dark current depends on time, so shoot a set of darks
    at longer times. You can do sets to match the exposure time of your shots
    and use the correct set. Again, average each set and subtract that from
    your image.

    Oh, and you need to do this at the aprox temp of you night out for
    best results.

    > the empty frame I took... Are there any special/easy techniques that
    > folk have discovered that might help, or is it just a case of.. well,
    > just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    > works best? I did a quick bit of research, and noticed a few
    > references to using LAB mode somehow, but it was unclear what they did
    > or how that helped.


    > (And yes, I've had a quick try with noise reduction software, but to
    > get a decent result there is too much compromise - I lose some of the
    > faint stars and traces of the comet.)


    > Any help would be appreciated, so I can avoid my least favorite
    > saying...


    > "Don't want to re-invent the wheel"


    You may also want to do a flat field correction. This is a way to correct
    out any pixel to pixel gain variations, and and `help' the lens gives us
    in uneven field and vignetting. Good flats are VERY time consuming to do,
    and with most reasonable DSLR they have to be good to be worth it.

    Go to Roger Clarks site and read his stuff, or hunt around the astro groups.
    There are a pile of packages to do this that run on a cripplebox.
     
    Paul Repacholi, Jan 27, 2007
    #5
  6. Ron Hunter Guest

    Paul Repacholi wrote:
    > writes:
    >
    >> Having just collected a few Comet McNaught pics on a digital that does
    >> not include DSF (Fuji S9500), I'm interested in having a go. The 9500
    >> did very well on all of the images that were <15 seconds, but there are
    >> a couple that I went for longer exposures to try to get the best
    >> possible result in showing the faint comet tail, and they have suffered
    >> as a result.

    >
    >> So before I start playing with my images (tifs and raw) and subtracting

    >
    > You have RAW, so you are in good shape. You need to work on the pixel data
    > numbers, not on the `image'.
    >
    > There are two noise sources to correct out. Bias noise, and dark current.
    > Bais frames you can get by shooting off a string of shots with the lens
    > cap on to keep all the dark in ;) Then average them and subtract that
    > from your DN data. Dark current depends on time, so shoot a set of darks
    > at longer times. You can do sets to match the exposure time of your shots
    > and use the correct set. Again, average each set and subtract that from
    > your image.
    >
    > Oh, and you need to do this at the aprox temp of you night out for
    > best results.
    >


    Indeed! I have seen test shots where a minimal (10 degree F)
    temperature drop made dramatic difference in the noise.


    >> the empty frame I took... Are there any special/easy techniques that
    >> folk have discovered that might help, or is it just a case of.. well,
    >> just 'subtracting' it (ie an inverted layer I guess) to whatever degree
    >> works best? I did a quick bit of research, and noticed a few
    >> references to using LAB mode somehow, but it was unclear what they did
    >> or how that helped.

    >
    >> (And yes, I've had a quick try with noise reduction software, but to
    >> get a decent result there is too much compromise - I lose some of the
    >> faint stars and traces of the comet.)

    >
    >> Any help would be appreciated, so I can avoid my least favorite
    >> saying...

    >
    >> "Don't want to re-invent the wheel"

    >
    > You may also want to do a flat field correction. This is a way to correct
    > out any pixel to pixel gain variations, and and `help' the lens gives us
    > in uneven field and vignetting. Good flats are VERY time consuming to do,
    > and with most reasonable DSLR they have to be good to be worth it.
    >
    > Go to Roger Clarks site and read his stuff, or hunt around the astro groups.
    > There are a pile of packages to do this that run on a cripplebox.
     
    Ron Hunter, Jan 27, 2007
    #6
  7. Guest

    Thanks folks - Paul, that was very enlightening - can't think why I
    didn't think to check out Roger's site.

    I've done a few minor experiments, using very basic techniques, and
    have had some success. But I think I need to take several more dark
    frames and average them, or perhaps try to match the temperatures..!
    Given the varying results I got, it would appear that over the time I
    was shooting the comet the temperature dropped pretty significantly,
    so what worked quite well on early images, didn't work so well on
    later ones.

    If I finally get some decent results, I'll post them here.

    Thanks again!
     
    , Jan 28, 2007
    #7
  8. writes:

    > Thanks folks - Paul, that was very enlightening - can't think why I
    > didn't think to check out Roger's site.


    > I've done a few minor experiments, using very basic techniques, and
    > have had some success. But I think I need to take several more dark
    > frames and average them, or perhaps try to match the temperatures..!
    > Given the varying results I got, it would appear that over the time
    > I was shooting the comet the temperature dropped pretty
    > significantly, so what worked quite well on early images, didn't
    > work so well on later ones.


    > If I finally get some decent results, I'll post them here.


    Don't be discoraged. If you hunt around the HST site, you can find the daily
    summary reports, and they include the data volumes of Scientific Data, aka images,
    and the callibration data. The latter is generally MANY times larger that
    the Sci data.

    If Roger sees this, he may be able to give a handwave number for Cassini, and
    maybe the MRO ;)
     
    Paul Repacholi, Feb 9, 2007
    #8
    1. Advertising

Want to reply to this thread or ask your own question?

It takes just 2 minutes to sign up (and it's free!). Just click the sign up button to choose a username and then you can ask your own questions on the forum.
Similar Threads
  1. hermes

    Any EE experts out there?

    hermes, Nov 16, 2005, in forum: Wireless Networking
    Replies:
    20
    Views:
    1,438
    Michael A Cooper
    Dec 1, 2005
  2. Denise

    Any HP experts out there?

    Denise, Jan 24, 2004, in forum: Computer Support
    Replies:
    14
    Views:
    935
    Jan Ras
    Jan 27, 2004
  3. Sens Fan Happy In Ohio

    Any DOS Experts Out There??

    Sens Fan Happy In Ohio, May 2, 2004, in forum: Computer Support
    Replies:
    11
    Views:
    784
    Toolman Tim
    May 3, 2004
  4. CSB

    Any AOL AIM experts out there?

    CSB, Jan 5, 2005, in forum: Computer Support
    Replies:
    5
    Views:
    455
    Alexander Rogge
    Jan 5, 2005
  5. Toolman Tim

    Another screwup--any Dell experts out there??

    Toolman Tim, Sep 15, 2005, in forum: Computer Support
    Replies:
    3
    Views:
    1,930
    Toolman Tim
    Sep 16, 2005
Loading...

Share This Page