Loading ...
Sorry, an error occurred while loading the content.
 

L.P. gradients & pixel math

Expand Messages
  • Randy Nulman
    Besides flattening the backgrounds prior to an RGB combine, Ron had suggested using pixel math to bring the background count down to about 50 counts. (If your
    Message 1 of 7 , Aug 31, 2001
      Besides flattening the backgrounds prior to an RGB combine, Ron had
      suggested using pixel math to bring the background count down to
      about 50 counts. (If your background reading is 750..use pixel math
      to subtract 700 from the image..leaving 50)

      This has worked very effectively for me, but I'm just curious if this
      proceedure also deducts from the image (non background) areas
      and "clips" it in any way?

      Thanks in advance,
      Randy
    • Ron Wodaski
      No, you are not removing any signal. The high background can be thought of as a bias, offsetting the signal by that amount. Now you _did_ lose some signal in
      Message 2 of 7 , Aug 31, 2001
        No, you are not removing any signal. The high background can be thought of
        as a bias, offsetting the signal by that amount.

        Now you _did_ lose some signal in the sky brightness. so this is not the
        same as imaging from a dark site, but you aren't losing anything
        _additional_ when you perform the subtraction. <g>

        Ron Wodaski
        The New CCD Astronomy
        http://www.newastro.com

        -----Original Message-----
        From: Randy Nulman [mailto:rjnulman@...]
        Sent: Friday, August 31, 2001 8:11 AM
        To: ccd-newastro@yahoogroups.com
        Subject: [ccd-newastro] L.P. gradients & pixel math


        Besides flattening the backgrounds prior to an RGB combine, Ron had
        suggested using pixel math to bring the background count down to
        about 50 counts. (If your background reading is 750..use pixel math
        to subtract 700 from the image..leaving 50)

        This has worked very effectively for me, but I'm just curious if this
        proceedure also deducts from the image (non background) areas
        and "clips" it in any way?

        Thanks in advance,
        Randy



        To unsubscribe from this group, send an email to:
        ccd-newastro-unsubscribe@egroups.com



        Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
      • Tony Kriz
        Ron: This is good news as I was also concerned that making these adjustments might reduce total image signal. I m finding your suggestion of using the Hutech +
        Message 3 of 7 , Aug 31, 2001
          Ron:

          This is good news as I was also concerned that making these
          adjustments might reduce total image signal.

          I'm finding your suggestion of using the Hutech + pixel math +
          flattening the background is enough to virtually eliminate the light
          pollution gradient problems.

          I'm curious as to what exacting happens using pixel math that allows
          it to effect the background and not the image signal. This may be
          too complicated for my layman brain but inquiring minds want to know.

          Thanks.

          Tony
        • Ron Wodaski
          That solution (filter + pixel math + gradient removal) is pretty much the recipe for dealing with light-polluted skies. To that I would add: take the longest
          Message 4 of 7 , Aug 31, 2001
            That solution (filter + pixel math + gradient removal) is pretty much the
            recipe for dealing with light-polluted skies. To that I would add: take the
            longest possible exposures, and lots of them.

            Consider two pixels in your image. Once is in the background -- no stars, no
            nebula, just background. The other pixel is in, say, a galaxy. Let's see
            what happens to the photons from those two sky locations from start to
            finish.

            Let's say you take a 10-minute exposure. Let's assume that the noise level
            is +/-10 ADU. The value of the hypothetical background pixel is made up of:

            * 1,000 ADU (Analog to Digital Units, the measure of brightness) because of
            sky glow

            * 242 ADU from light pollution

            * 100 ADU because of a pedestal supplied by the camera driver (SBIG only)

            * -8 ADU because of noise

            So the background pixel, after a dark frame and flat field, has a value of
            1,336 ADU.

            The value of the galaxy pixel is made up of:

            * 422 ADU due to photons from the galaxy

            * 1,000 ADU because of sky glow

            * 330 ADU due to a light pollution gradient (the galaxy is closer to the
            light pollution source than the background pixel was, just for the sake of
            example)

            * 100 ADU pedestal

            * +2 because of noise (noise is random, so it could be anything between -10
            and +10)

            So the galaxy pixel, likewise after reduction, has a value of 1854. The
            galaxy pixel is brighter than the background pixel by (1854-1336), or 518
            ADU. Some of this is due to the light pollution gradient, which has not yet
            been corrected.

            If I subtract 1286 from both pixels, then their values are:

            background: 50 ADU
            galaxy pixel: 568 ADU

            The galaxy pixel is still 518 ADU brighter, but if I have taken several
            images and want to combine them, they will all have the same background
            level, and thus any averaging or other calculations on the image pixels will
            be more valid because the brightness values are being measured on the same
            scale from the same starting point. That is the advantage you gain by
            "normalizing" the images (that is, giving them a uniform background level).

            If you now do a proper gradient removal, the pixels will reach have a more
            accurate brightness relationship to one another. The background pixel has
            been set to 50, so it is the reference point. This is, in effect, "black."
            Since one always measures the darkest part of the background when
            normalizing, this background pixel was chosen because it is the darkest part
            of the gradient. The difference between the galaxy pixel gradient
            contribution (330) and the background (242) is 88 ADU. After normalization,
            that is how much the galaxy pixel will be darkened by gradient removal. So
            the pixel values after gradient removal are:

            background: 50 ADU (no gradient to remove)
            galaxy pixel: 480 ADU

            So the "true" difference in brightness is 430 ADU. Normalization removes sky
            glow; gradient removal removes the light pollution contribution, and
            afterwards we are left with the more-or-less accurate brightness value for
            each pixel.

            You will add a small amount of noise by doing this processing (we are
            assuming things about how even the gradient is, for example), but the net
            gain exceeds the cost in noise increase, so it is still worth doing.

            Ron Wodaski
            The New CCD Astronomy
            http://www.newastro.com

            -----Original Message-----
            From: Tony Kriz [mailto:tkriz1027@...]
            Sent: Friday, August 31, 2001 11:33 AM
            To: ccd-newastro@yahoogroups.com
            Subject: [ccd-newastro] Re: L.P. gradients & pixel math


            Ron:

            This is good news as I was also concerned that making these
            adjustments might reduce total image signal.

            I'm finding your suggestion of using the Hutech + pixel math +
            flattening the background is enough to virtually eliminate the light
            pollution gradient problems.

            I'm curious as to what exacting happens using pixel math that allows
            it to effect the background and not the image signal. This may be
            too complicated for my layman brain but inquiring minds want to know.

            Thanks.

            Tony



            To unsubscribe from this group, send an email to:
            ccd-newastro-unsubscribe@egroups.com



            Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
          • Randy Nulman
            Ron, Thanks for this information..it s very helpful. If I had a number of red, green, blue exposure, I would first combine them and then perform pixel math on
            Message 5 of 7 , Sep 1, 2001
              Ron,

              Thanks for this information..it's very helpful.
              If I had a number of red, green, blue exposure, I would first combine
              them and then perform pixel math on each combined result (prior to
              the RGB combine routine).

              If I understand your response correctly, this is incorrect. I
              believe you're saying to do the pixel math on each individ. exposure
              prior to any combine...then combine...then do the RGB combine. Is
              this correct? If so, is there any validity of doing a "final" pixel
              math correction to the combined result of each channel prior to the
              RGB combine (or is this overkill?)

              Thanks,
              Randy

              --- In ccd-newastro@y..., "Ron Wodaski" <ronw@n...> wrote:
              > That solution (filter + pixel math + gradient removal) is pretty
              much the
              > recipe for dealing with light-polluted skies. To that I would add:
              take the
              > longest possible exposures, and lots of them.
              >
              > Consider two pixels in your image. Once is in the background -- no
              stars, no
              > nebula, just background. The other pixel is in, say, a galaxy.
              Let's see
              > what happens to the photons from those two sky locations from start
              to
              > finish.
              >
              > Let's say you take a 10-minute exposure. Let's assume that the
              noise level
              > is +/-10 ADU. The value of the hypothetical background pixel is
              made up of:
              >
              > * 1,000 ADU (Analog to Digital Units, the measure of brightness)
              because of
              > sky glow
              >
              > * 242 ADU from light pollution
              >
              > * 100 ADU because of a pedestal supplied by the camera driver (SBIG
              only)
              >
              > * -8 ADU because of noise
              >
              > So the background pixel, after a dark frame and flat field, has a
              value of
              > 1,336 ADU.
              >
              > The value of the galaxy pixel is made up of:
              >
              > * 422 ADU due to photons from the galaxy
              >
              > * 1,000 ADU because of sky glow
              >
              > * 330 ADU due to a light pollution gradient (the galaxy is closer
              to the
              > light pollution source than the background pixel was, just for the
              sake of
              > example)
              >
              > * 100 ADU pedestal
              >
              > * +2 because of noise (noise is random, so it could be anything
              between -10
              > and +10)
              >
              > So the galaxy pixel, likewise after reduction, has a value of 1854.
              The
              > galaxy pixel is brighter than the background pixel by (1854-1336),
              or 518
              > ADU. Some of this is due to the light pollution gradient, which has
              not yet
              > been corrected.
              >
              > If I subtract 1286 from both pixels, then their values are:
              >
              > background: 50 ADU
              > galaxy pixel: 568 ADU
              >
              > The galaxy pixel is still 518 ADU brighter, but if I have taken
              several
              > images and want to combine them, they will all have the same
              background
              > level, and thus any averaging or other calculations on the image
              pixels will
              > be more valid because the brightness values are being measured on
              the same
              > scale from the same starting point. That is the advantage you gain
              by
              > "normalizing" the images (that is, giving them a uniform background
              level).
              >
              > If you now do a proper gradient removal, the pixels will reach have
              a more
              > accurate brightness relationship to one another. The background
              pixel has
              > been set to 50, so it is the reference point. This is, in
              effect, "black."
              > Since one always measures the darkest part of the background when
              > normalizing, this background pixel was chosen because it is the
              darkest part
              > of the gradient. The difference between the galaxy pixel gradient
              > contribution (330) and the background (242) is 88 ADU. After
              normalization,
              > that is how much the galaxy pixel will be darkened by gradient
              removal. So
              > the pixel values after gradient removal are:
              >
              > background: 50 ADU (no gradient to remove)
              > galaxy pixel: 480 ADU
              >
              > So the "true" difference in brightness is 430 ADU. Normalization
              removes sky
              > glow; gradient removal removes the light pollution contribution, and
              > afterwards we are left with the more-or-less accurate brightness
              value for
              > each pixel.
              >
              > You will add a small amount of noise by doing this processing (we
              are
              > assuming things about how even the gradient is, for example), but
              the net
              > gain exceeds the cost in noise increase, so it is still worth doing.
              >
              > Ron Wodaski
              > The New CCD Astronomy
              > http://www.newastro.com
              >
              > -----Original Message-----
              > From: Tony Kriz [mailto:tkriz1027@a...]
              > Sent: Friday, August 31, 2001 11:33 AM
              > To: ccd-newastro@y...
              > Subject: [ccd-newastro] Re: L.P. gradients & pixel math
              >
              >
              > Ron:
              >
              > This is good news as I was also concerned that making these
              > adjustments might reduce total image signal.
              >
              > I'm finding your suggestion of using the Hutech + pixel math +
              > flattening the background is enough to virtually eliminate the light
              > pollution gradient problems.
              >
              > I'm curious as to what exacting happens using pixel math that allows
              > it to effect the background and not the image signal. This may be
              > too complicated for my layman brain but inquiring minds want to
              know.
              >
              > Thanks.
              >
              > Tony
              >
              >
              >
              > To unsubscribe from this group, send an email to:
              > ccd-newastro-unsubscribe@egroups.com
              >
              >
              >
              > Your use of Yahoo! Groups is subject to
              http://docs.yahoo.com/info/terms/
            • Tony Kriz
              Ron: Great explaination, even I can understand! BTW, I played around with your tips on halo removal using the rubber stamp tool and it works pretty well.
              Message 6 of 7 , Sep 1, 2001
                Ron:

                Great explaination, even I can understand!

                BTW, I played around with your tips on halo removal using the rubber
                stamp tool and it works pretty well. After a few attempts you gain a
                steady hand and can move quickly from star to star.

                I tried to use the polygonal lasso tool technique but I couldn't
                figure out how to select both the inner star area and the outer halo
                at the same time. Is there a special trick to selecting both areas
                so you can adjust down the levels or curves in the halo area?

                Thanks.

                Tony


                --- In ccd-newastro@y..., "Ron Wodaski" <ronw@n...> wrote:
                > That solution (filter + pixel math + gradient removal) is pretty
                much the
                > recipe for dealing with light-polluted skies. To that I would add:
                take the
                > longest possible exposures, and lots of them.
                >
                > Consider two pixels in your image. Once is in the background -- no
                stars, no
                > nebula, just background. The other pixel is in, say, a galaxy.
                Let's see
                > what happens to the photons from those two sky locations from start
                to
                > finish.
                >
                > Let's say you take a 10-minute exposure. Let's assume that the
                noise level
                > is +/-10 ADU. The value of the hypothetical background pixel is
                made up of:
                >
                > * 1,000 ADU (Analog to Digital Units, the measure of brightness)
                because of
                > sky glow
                >
                > * 242 ADU from light pollution
                >
                > * 100 ADU because of a pedestal supplied by the camera driver (SBIG
                only)
                >
                > * -8 ADU because of noise
                >
                > So the background pixel, after a dark frame and flat field, has a
                value of
                > 1,336 ADU.
                >
                > The value of the galaxy pixel is made up of:
                >
                > * 422 ADU due to photons from the galaxy
                >
                > * 1,000 ADU because of sky glow
                >
                > * 330 ADU due to a light pollution gradient (the galaxy is closer
                to the
                > light pollution source than the background pixel was, just for the
                sake of
                > example)
                >
                > * 100 ADU pedestal
                >
                > * +2 because of noise (noise is random, so it could be anything
                between -10
                > and +10)
                >
                > So the galaxy pixel, likewise after reduction, has a value of 1854.
                The
                > galaxy pixel is brighter than the background pixel by (1854-1336),
                or 518
                > ADU. Some of this is due to the light pollution gradient, which has
                not yet
                > been corrected.
                >
                > If I subtract 1286 from both pixels, then their values are:
                >
                > background: 50 ADU
                > galaxy pixel: 568 ADU
                >
                > The galaxy pixel is still 518 ADU brighter, but if I have taken
                several
                > images and want to combine them, they will all have the same
                background
                > level, and thus any averaging or other calculations on the image
                pixels will
                > be more valid because the brightness values are being measured on
                the same
                > scale from the same starting point. That is the advantage you gain
                by
                > "normalizing" the images (that is, giving them a uniform background
                level).
                >
                > If you now do a proper gradient removal, the pixels will reach have
                a more
                > accurate brightness relationship to one another. The background
                pixel has
                > been set to 50, so it is the reference point. This is, in
                effect, "black."
                > Since one always measures the darkest part of the background when
                > normalizing, this background pixel was chosen because it is the
                darkest part
                > of the gradient. The difference between the galaxy pixel gradient
                > contribution (330) and the background (242) is 88 ADU. After
                normalization,
                > that is how much the galaxy pixel will be darkened by gradient
                removal. So
                > the pixel values after gradient removal are:
                >
                > background: 50 ADU (no gradient to remove)
                > galaxy pixel: 480 ADU
                >
                > So the "true" difference in brightness is 430 ADU. Normalization
                removes sky
                > glow; gradient removal removes the light pollution contribution, and
                > afterwards we are left with the more-or-less accurate brightness
                value for
                > each pixel.
                >
                > You will add a small amount of noise by doing this processing (we
                are
                > assuming things about how even the gradient is, for example), but
                the net
                > gain exceeds the cost in noise increase, so it is still worth doing.
                >
                > Ron Wodaski
                > The New CCD Astronomy
                > http://www.newastro.com
                >
                > -----Original Message-----
                > From: Tony Kriz [mailto:tkriz1027@a...]
                > Sent: Friday, August 31, 2001 11:33 AM
                > To: ccd-newastro@y...
                > Subject: [ccd-newastro] Re: L.P. gradients & pixel math
                >
                >
                > Ron:
                >
                > This is good news as I was also concerned that making these
                > adjustments might reduce total image signal.
                >
                > I'm finding your suggestion of using the Hutech + pixel math +
                > flattening the background is enough to virtually eliminate the light
                > pollution gradient problems.
                >
                > I'm curious as to what exacting happens using pixel math that allows
                > it to effect the background and not the image signal. This may be
                > too complicated for my layman brain but inquiring minds want to
                know.
                >
                > Thanks.
                >
                > Tony
                >
                >
                >
                > To unsubscribe from this group, send an email to:
                > ccd-newastro-unsubscribe@egroups.com
                >
                >
                >
                > Your use of Yahoo! Groups is subject to
                http://docs.yahoo.com/info/terms/
              • Ron Wodaski
                MaxIm DL 3.0 deals with the for you. It will normalize images prior to a sum, average, or median combine -- this is the correct procedure. Until it is
                Message 7 of 7 , Sep 1, 2001
                  MaxIm DL 3.0 deals with the for you. It will normalize images prior to a
                  sum, average, or median combine -- this is the correct procedure. Until it
                  is released, you would have to manually normalize prior to a combine.

                  You would then normalize the RGB images to a common background level, and
                  then do a color combine.

                  Ron Wodaski
                  The New CCD Astronomy
                  http://www.newastro.com

                  -----Original Message-----
                  From: Randy Nulman [mailto:rjnulman@...]
                  Sent: Saturday, September 01, 2001 8:57 AM
                  To: ccd-newastro@yahoogroups.com
                  Subject: [ccd-newastro] Re: L.P. gradients & pixel math


                  Ron,

                  Thanks for this information..it's very helpful.
                  If I had a number of red, green, blue exposure, I would first combine
                  them and then perform pixel math on each combined result (prior to
                  the RGB combine routine).

                  If I understand your response correctly, this is incorrect. I
                  believe you're saying to do the pixel math on each individ. exposure
                  prior to any combine...then combine...then do the RGB combine. Is
                  this correct? If so, is there any validity of doing a "final" pixel
                  math correction to the combined result of each channel prior to the
                  RGB combine (or is this overkill?)

                  Thanks,
                  Randy

                  --- In ccd-newastro@y..., "Ron Wodaski" <ronw@n...> wrote:
                  > That solution (filter + pixel math + gradient removal) is pretty
                  much the
                  > recipe for dealing with light-polluted skies. To that I would add:
                  take the
                  > longest possible exposures, and lots of them.
                  >
                  > Consider two pixels in your image. Once is in the background -- no
                  stars, no
                  > nebula, just background. The other pixel is in, say, a galaxy.
                  Let's see
                  > what happens to the photons from those two sky locations from start
                  to
                  > finish.
                  >
                  > Let's say you take a 10-minute exposure. Let's assume that the
                  noise level
                  > is +/-10 ADU. The value of the hypothetical background pixel is
                  made up of:
                  >
                  > * 1,000 ADU (Analog to Digital Units, the measure of brightness)
                  because of
                  > sky glow
                  >
                  > * 242 ADU from light pollution
                  >
                  > * 100 ADU because of a pedestal supplied by the camera driver (SBIG
                  only)
                  >
                  > * -8 ADU because of noise
                  >
                  > So the background pixel, after a dark frame and flat field, has a
                  value of
                  > 1,336 ADU.
                  >
                  > The value of the galaxy pixel is made up of:
                  >
                  > * 422 ADU due to photons from the galaxy
                  >
                  > * 1,000 ADU because of sky glow
                  >
                  > * 330 ADU due to a light pollution gradient (the galaxy is closer
                  to the
                  > light pollution source than the background pixel was, just for the
                  sake of
                  > example)
                  >
                  > * 100 ADU pedestal
                  >
                  > * +2 because of noise (noise is random, so it could be anything
                  between -10
                  > and +10)
                  >
                  > So the galaxy pixel, likewise after reduction, has a value of 1854.
                  The
                  > galaxy pixel is brighter than the background pixel by (1854-1336),
                  or 518
                  > ADU. Some of this is due to the light pollution gradient, which has
                  not yet
                  > been corrected.
                  >
                  > If I subtract 1286 from both pixels, then their values are:
                  >
                  > background: 50 ADU
                  > galaxy pixel: 568 ADU
                  >
                  > The galaxy pixel is still 518 ADU brighter, but if I have taken
                  several
                  > images and want to combine them, they will all have the same
                  background
                  > level, and thus any averaging or other calculations on the image
                  pixels will
                  > be more valid because the brightness values are being measured on
                  the same
                  > scale from the same starting point. That is the advantage you gain
                  by
                  > "normalizing" the images (that is, giving them a uniform background
                  level).
                  >
                  > If you now do a proper gradient removal, the pixels will reach have
                  a more
                  > accurate brightness relationship to one another. The background
                  pixel has
                  > been set to 50, so it is the reference point. This is, in
                  effect, "black."
                  > Since one always measures the darkest part of the background when
                  > normalizing, this background pixel was chosen because it is the
                  darkest part
                  > of the gradient. The difference between the galaxy pixel gradient
                  > contribution (330) and the background (242) is 88 ADU. After
                  normalization,
                  > that is how much the galaxy pixel will be darkened by gradient
                  removal. So
                  > the pixel values after gradient removal are:
                  >
                  > background: 50 ADU (no gradient to remove)
                  > galaxy pixel: 480 ADU
                  >
                  > So the "true" difference in brightness is 430 ADU. Normalization
                  removes sky
                  > glow; gradient removal removes the light pollution contribution, and
                  > afterwards we are left with the more-or-less accurate brightness
                  value for
                  > each pixel.
                  >
                  > You will add a small amount of noise by doing this processing (we
                  are
                  > assuming things about how even the gradient is, for example), but
                  the net
                  > gain exceeds the cost in noise increase, so it is still worth doing.
                  >
                  > Ron Wodaski
                  > The New CCD Astronomy
                  > http://www.newastro.com
                  >
                  > -----Original Message-----
                  > From: Tony Kriz [mailto:tkriz1027@a...]
                  > Sent: Friday, August 31, 2001 11:33 AM
                  > To: ccd-newastro@y...
                  > Subject: [ccd-newastro] Re: L.P. gradients & pixel math
                  >
                  >
                  > Ron:
                  >
                  > This is good news as I was also concerned that making these
                  > adjustments might reduce total image signal.
                  >
                  > I'm finding your suggestion of using the Hutech + pixel math +
                  > flattening the background is enough to virtually eliminate the light
                  > pollution gradient problems.
                  >
                  > I'm curious as to what exacting happens using pixel math that allows
                  > it to effect the background and not the image signal. This may be
                  > too complicated for my layman brain but inquiring minds want to
                  know.
                  >
                  > Thanks.
                  >
                  > Tony
                  >
                  >
                  >
                  > To unsubscribe from this group, send an email to:
                  > ccd-newastro-unsubscribe@egroups.com
                  >
                  >
                  >
                  > Your use of Yahoo! Groups is subject to
                  http://docs.yahoo.com/info/terms/



                  To unsubscribe from this group, send an email to:
                  ccd-newastro-unsubscribe@egroups.com



                  Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
                Your message has been successfully submitted and would be delivered to recipients shortly.