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

Re: [ccd-newastro] Question for Ron or who ever can help.

Expand Messages
  • Bill Logan
    Interesting graph Ron. According to the graph, 50 light frames would be the time to quit as Mark was asking. Correct? Any more frames wouldn t significantly
    Message 1 of 21 , Jun 8, 2009
      Interesting graph Ron. According to the graph, 50 light frames would be the
      time to quit as Mark was asking. Correct? Any more frames wouldn't
      significantly improve the S/N? Of course, 50 exposures say at 600 seconds
      each would require 8.3 hours. With that kind of time and summer months
      approaching, one would have to wait for total darkness and then select an
      object on the eastern horizon and hope you're finish before sunrise.
      Considering atmospheric extinction values low on east and west horizons, the
      first hour and the last hour of imaging may have too much distortion and may
      need to be thrown out. LOL. Thanks for the graph. That made a complicated
      subject simple to understand.

      Bill Logan
      Eagar, Arizona


      ----- Original Message -----
      From: "Yahoo - Wodaski" <yahoo@...>
      To: <ccd-newastro@yahoogroups.com>
      Sent: Monday, June 08, 2009 6:30 PM
      Subject: Re: [ccd-newastro] Question for Ron or who ever can help.


      > Here is an Excel graph which shows the diminishing returns from
      > additional images:
      >
      > http://wodaski.com/ex/signal_to_noise.jpg
      >
      > The numbers along the bottom are the number of images. The numbers at
      > left are the approximate % improvement in S/N by adding ONE more image
      > to the existing stack. The more images you take, the more additional
      > images you need to take to get a significant improvement in S/N. In
      > other words, if you take 5 images and want to see another significant
      > improvement, such as improving dim object signal to noise 2x, then
      > you'll need another 15 images!
      >
      > So you can improve your S/N, but the more images you take, the more
      > challenging it is to get visible improvement.
      >
      > This assumes that individual sub exposures are long enough for shot
      > noise to swamp read noise.
      >
      > Ron Wodaski
      >
      > Mark wrote:
      >> During capture and image stacking at what point does adding more subs not
      >> make a difference any more?
      >> I know the more images stacked improves the signal to noise ratio but
      >> when is it mute to quit adding time or exposures to an image?
      >>
      >> Ive always been curious on this.
      >>
      >> Regards
      >>
      >> Mark
      >>
      >>
      >>
      >>
      >>
      >>
      >>
      >> ------------------------------------
      >>
      >>
      >>
      >>
      >
      >
      > ------------------------------------
      >
      >
      >
      >
      >
    • rgr3218893
      Ron, So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for pretty
      Message 2 of 21 , Jun 8, 2009
        Ron,

        So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...

        I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.

        What is a target S/N?

        Rex

        --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
        >
        > BTW: that doubling S/N thing can readily be calculated. Generally, to
        > double your S/N, you need 4X the present number of images. Thus, if you
        > have 5 images, you need 5*4=20 total to double S/N; once you have 20
        > images, you need 20*4=80 to double your S/N again.
        >
        > Ron W
        >
        > Yahoo - Wodaski wrote:
        > > Here is an Excel graph which shows the diminishing returns from
        > > additional images:
        > >
        > > http://wodaski.com/ex/signal_to_noise.jpg
        > >
        > > The numbers along the bottom are the number of images. The numbers at
        > > left are the approximate % improvement in S/N by adding ONE more image
        > > to the existing stack. The more images you take, the more additional
        > > images you need to take to get a significant improvement in S/N. In
        > > other words, if you take 5 images and want to see another significant
        > > improvement, such as improving dim object signal to noise 2x, then
        > > you'll need another 15 images!
        > >
        > > So you can improve your S/N, but the more images you take, the more
        > > challenging it is to get visible improvement.
        > >
        > > This assumes that individual sub exposures are long enough for shot
        > > noise to swamp read noise.
        > >
        > > Ron Wodaski
        > >
        > > Mark wrote:
        > >> During capture and image stacking at what point does adding more subs not make a difference any more?
        > >> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
        > >>
        > >> Ive always been curious on this.
        > >>
        > >> Regards
        > >>
        > >> Mark
        > >>
        > >>
        > >>
        > >>
        > >>
        > >>
        > >>
        > >> ------------------------------------
        > >>
        > >>
        > >>
        > >>
        > >
        > >
        > > ------------------------------------
        > >
        > >
        > >
        > >
        >
      • Yahoo - Wodaski
        Actually, my point is that there is no hard quit point. 8.3 hours, in today s mega-exposure world, is still considered short (well, by a few people, anyway
        Message 3 of 21 , Jun 8, 2009
          Actually, my point is that there is no hard quit point. 8.3 hours, in
          today's mega-exposure world, is still considered short (well, by a few
          people, anyway <g>). There's no need at all to do all the imaging on one
          night, either. You could image every available night for a month, for
          example. Professionals do that all the time - steal some time here, some
          time there, and then put it all together.

          I see 24 hour exposures, taken over multiple nights. How long was the
          Hubble Ultra Deep Field total exposure time, you ask? The total exposure
          time is just under 1 million seconds, from 400 orbits, with a typical
          sub-exposure time of 1200 seconds. 275 hours, in round numbers.

          And the Hubble has a lot of advantages. <g> So there really is no upper
          limit to total exposure time beyond one's own patience and skill. (To
          realize the benefit of all those exposures, you're going to need really
          low-noise bias and dark frames, which means many, many darks!)

          And if you want to double the S/N after taking 50 images, well, you do
          the math: now you'll need 200 images. <g>

          Bill Logan wrote:
          > Interesting graph Ron. According to the graph, 50 light frames would be the
          > time to quit as Mark was asking. Correct? Any more frames wouldn't
          > significantly improve the S/N? Of course, 50 exposures say at 600 seconds
          > each would require 8.3 hours. With that kind of time and summer months
          > approaching, one would have to wait for total darkness and then select an
          > object on the eastern horizon and hope you're finish before sunrise.
          > Considering atmospheric extinction values low on east and west horizons, the
          > first hour and the last hour of imaging may have too much distortion and may
          > need to be thrown out. LOL. Thanks for the graph. That made a complicated
          > subject simple to understand.
          >
          > Bill Logan
          > Eagar, Arizona
          >
          >
          > ----- Original Message -----
          > From: "Yahoo - Wodaski" <yahoo@...>
          > To: <ccd-newastro@yahoogroups.com>
          > Sent: Monday, June 08, 2009 6:30 PM
          > Subject: Re: [ccd-newastro] Question for Ron or who ever can help.
          >
          >
          >> Here is an Excel graph which shows the diminishing returns from
          >> additional images:
          >>
          >> http://wodaski.com/ex/signal_to_noise.jpg
          >>
          >> The numbers along the bottom are the number of images. The numbers at
          >> left are the approximate % improvement in S/N by adding ONE more image
          >> to the existing stack. The more images you take, the more additional
          >> images you need to take to get a significant improvement in S/N. In
          >> other words, if you take 5 images and want to see another significant
          >> improvement, such as improving dim object signal to noise 2x, then
          >> you'll need another 15 images!
          >>
          >> So you can improve your S/N, but the more images you take, the more
          >> challenging it is to get visible improvement.
          >>
          >> This assumes that individual sub exposures are long enough for shot
          >> noise to swamp read noise.
          >>
          >> Ron Wodaski
          >>
          >> Mark wrote:
          >>> During capture and image stacking at what point does adding more subs not
          >>> make a difference any more?
          >>> I know the more images stacked improves the signal to noise ratio but
          >>> when is it mute to quit adding time or exposures to an image?
          >>>
          >>> Ive always been curious on this.
          >>>
          >>> Regards
          >>>
          >>> Mark
          >>>
          >>>
          >>>
          >>>
          >>>
          >>>
          >>>
          >>> ------------------------------------
          >>>
          >>>
          >>>
          >>>
          >>
          >> ------------------------------------
          >>
          >>
          >>
          >>
          >>
          >
          >
          >
          > ------------------------------------
          >
          >
          >
          >
        • JoeMize
          Never. It all depends upon your patience and how much details and faint stuff you want. IMHO...joe :)) May You Go Among The Imperishable Stars Joe Mize
          Message 4 of 21 , Jun 8, 2009
            Never. It all depends upon your patience and how much details and faint
            stuff you want. IMHO...joe :))


            "May You Go Among The Imperishable Stars"
            Joe Mize www.cav-sfo.com
            Chiefland Astronomy Village, Fla.

            ----- Original Message -----
            From: "Mark" <astronomy@...>
            To: <ccd-newastro@yahoogroups.com>
            Sent: Monday, June 08, 2009 7:53 PM
            Subject: [ccd-newastro] Question for Ron or who ever can help.


            > During capture and image stacking at what point does adding more subs not
            > make a difference any more?
            > I know the more images stacked improves the signal to noise ratio but when
            > is it mute to quit adding time or exposures to an image?
            >
            > Ive always been curious on this.
            >
            > Regards
            >
            > Mark
            >
            >
            >
            >
            >
            >
            >
            > ------------------------------------
            >
            >
            >
            >
            >
          • Bill Logan
            You are absolutely correct Ron. I have never tried multiple nights, but it sounds like something I might try this summer. Someone once told me to get as much
            Message 5 of 21 , Jun 8, 2009
              You are absolutely correct Ron. I have never tried multiple nights, but it
              sounds like something I might try this summer. Someone once told me to get
              as much data as possible so I suppose the number of exposures is infinite.

              Thanks again for your extraordinary knowledge of digital imaging. I am
              always learning something from you.

              Bill Logan
              Eagar, Arizona


              ----- Original Message -----
              From: "Yahoo - Wodaski" <yahoo@...>
              To: <ccd-newastro@yahoogroups.com>
              Sent: Monday, June 08, 2009 8:47 PM
              Subject: Re: [ccd-newastro] Question for Ron or who ever can help.


              > Actually, my point is that there is no hard quit point. 8.3 hours, in
              > today's mega-exposure world, is still considered short (well, by a few
              > people, anyway <g>). There's no need at all to do all the imaging on one
              > night, either. You could image every available night for a month, for
              > example. Professionals do that all the time - steal some time here, some
              > time there, and then put it all together.
              >
              > I see 24 hour exposures, taken over multiple nights. How long was the
              > Hubble Ultra Deep Field total exposure time, you ask? The total exposure
              > time is just under 1 million seconds, from 400 orbits, with a typical
              > sub-exposure time of 1200 seconds. 275 hours, in round numbers.
              >
              > And the Hubble has a lot of advantages. <g> So there really is no upper
              > limit to total exposure time beyond one's own patience and skill. (To
              > realize the benefit of all those exposures, you're going to need really
              > low-noise bias and dark frames, which means many, many darks!)
              >
              > And if you want to double the S/N after taking 50 images, well, you do
              > the math: now you'll need 200 images. <g>
              >
              > Bill Logan wrote:
              >> Interesting graph Ron. According to the graph, 50 light frames would be
              >> the
              >> time to quit as Mark was asking. Correct? Any more frames wouldn't
              >> significantly improve the S/N? Of course, 50 exposures say at 600
              >> seconds
              >> each would require 8.3 hours. With that kind of time and summer months
              >> approaching, one would have to wait for total darkness and then select an
              >> object on the eastern horizon and hope you're finish before sunrise.
              >> Considering atmospheric extinction values low on east and west horizons,
              >> the
              >> first hour and the last hour of imaging may have too much distortion and
              >> may
              >> need to be thrown out. LOL. Thanks for the graph. That made a
              >> complicated
              >> subject simple to understand.
              >>
              >> Bill Logan
              >> Eagar, Arizona
              >>
              >>
              >> ----- Original Message -----
              >> From: "Yahoo - Wodaski" <yahoo@...>
              >> To: <ccd-newastro@yahoogroups.com>
              >> Sent: Monday, June 08, 2009 6:30 PM
              >> Subject: Re: [ccd-newastro] Question for Ron or who ever can help.
              >>
              >>
              >>> Here is an Excel graph which shows the diminishing returns from
              >>> additional images:
              >>>
              >>> http://wodaski.com/ex/signal_to_noise.jpg
              >>>
              >>> The numbers along the bottom are the number of images. The numbers at
              >>> left are the approximate % improvement in S/N by adding ONE more image
              >>> to the existing stack. The more images you take, the more additional
              >>> images you need to take to get a significant improvement in S/N. In
              >>> other words, if you take 5 images and want to see another significant
              >>> improvement, such as improving dim object signal to noise 2x, then
              >>> you'll need another 15 images!
              >>>
              >>> So you can improve your S/N, but the more images you take, the more
              >>> challenging it is to get visible improvement.
              >>>
              >>> This assumes that individual sub exposures are long enough for shot
              >>> noise to swamp read noise.
              >>>
              >>> Ron Wodaski
              >>>
              >>> Mark wrote:
              >>>> During capture and image stacking at what point does adding more subs
              >>>> not
              >>>> make a difference any more?
              >>>> I know the more images stacked improves the signal to noise ratio but
              >>>> when is it mute to quit adding time or exposures to an image?
              >>>>
              >>>> Ive always been curious on this.
              >>>>
              >>>> Regards
              >>>>
              >>>> Mark
              >>>>
              >>>>
              >>>>
              >>>>
              >>>>
              >>>>
              >>>>
              >>>> ------------------------------------
              >>>>
              >>>>
              >>>>
              >>>>
              >>>
              >>> ------------------------------------
              >>>
              >>>
              >>>
              >>>
              >>>
              >>
              >>
              >>
              >> ------------------------------------
              >>
              >>
              >>
              >>
              >
              >
              > ------------------------------------
              >
              >
              >
              >
              >
            • Yahoo - Wodaski
              Well, in truth, there is no such thing as the S/N of a single sub! That s because noise is the uncertainty in the data. To measure the total noise in an image,
              Message 6 of 21 , Jun 8, 2009
                Well, in truth, there is no such thing as the S/N of a single sub!
                That's because noise is the uncertainty in the data. To measure the
                total noise in an image, you have to take a lot of images, and then see
                how wrong that one image is.

                Of course, we can estimate the noise as read noise + shot noise + any
                other instrument noise + (noise from bias, dark, flats). We're not even
                considering dynamic range and contrast transfer, which also have a lot
                to do with what you can image. And don't forget good old FWHM - the
                seeing. And the resolving power of your scope. the quantum efficiency of
                your detector. The reflectivity of any optical surfaces (even coated
                surfaces reflect a little bit). And then there is the little matter of
                detector geometry: what happens to photons that hit various
                non-photosensitive portions of the chip? Do they reflect away? Do they
                create contrast-reducing scatter? Are the wavelength-speecific artifacts
                (e.g., etaloning)?

                And that's just a list of things that start to apply as you get deeper
                exposures and start to have to worry about. not just in the
                calculations; but in how you process the image (wavelength-specific
                etalon effects are a pain in the you-know-what).

                Since every site varies from other sites, and further every site varies
                hour by hour (sometimes minute by minute) there just no answer to your
                question about 'how long'. In the end, some combination of your patience
                and your satisfaction in your work.

                Even your 'hard case' is highly variable - how faint of a wisp? How many
                photons per hour? What would you see if you just took another 450
                images? <g>

                It is far easier to do some test combines with your data and if you are
                not happy, what you can now at least do is think in relatively concrete
                terms about what it would take to get a better image. let's say your
                faint nebula is barely there in an 8-hour exposure. It's going to take
                another 24 hours of exposure time to double the S/N.

                Yikes! maybe 1.4X S/N would be good enough... <g>

                Ron W

                rgr3218893 wrote:
                > Ron,
                >
                > So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...
                >
                > I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.
                >
                > What is a target S/N?
                >
                > Rex
                >
                > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                >> BTW: that doubling S/N thing can readily be calculated. Generally, to
                >> double your S/N, you need 4X the present number of images. Thus, if you
                >> have 5 images, you need 5*4=20 total to double S/N; once you have 20
                >> images, you need 20*4=80 to double your S/N again.
                >>
                >> Ron W
                >>
                >> Yahoo - Wodaski wrote:
                >>> Here is an Excel graph which shows the diminishing returns from
                >>> additional images:
                >>>
                >>> http://wodaski.com/ex/signal_to_noise.jpg
                >>>
                >>> The numbers along the bottom are the number of images. The numbers at
                >>> left are the approximate % improvement in S/N by adding ONE more image
                >>> to the existing stack. The more images you take, the more additional
                >>> images you need to take to get a significant improvement in S/N. In
                >>> other words, if you take 5 images and want to see another significant
                >>> improvement, such as improving dim object signal to noise 2x, then
                >>> you'll need another 15 images!
                >>>
                >>> So you can improve your S/N, but the more images you take, the more
                >>> challenging it is to get visible improvement.
                >>>
                >>> This assumes that individual sub exposures are long enough for shot
                >>> noise to swamp read noise.
                >>>
                >>> Ron Wodaski
                >>>
                >>> Mark wrote:
                >>>> During capture and image stacking at what point does adding more subs not make a difference any more?
                >>>> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
                >>>>
                >>>> Ive always been curious on this.
                >>>>
                >>>> Regards
                >>>>
                >>>> Mark
                >>>>
                >>>>
                >>>>
                >>>>
                >>>>
                >>>>
                >>>>
                >>>> ------------------------------------
                >>>>
                >>>>
                >>>>
                >>>>
                >>>
                >>> ------------------------------------
                >>>
                >>>
                >>>
                >>>
                >
                >
                >
                >
                > ------------------------------------
                >
                >
                >
                >
              • Mark
                Ron, Thanks for all of the insight. I remember this from a while back but thought I had it wrong. You confirmed it now. Thanks again for all of the
                Message 7 of 21 , Jun 9, 2009
                  Ron,
                  Thanks for all of the insight.
                  I remember this from a while back but thought I had it wrong.
                  You confirmed it now.
                  Thanks again for all of the information.

                  Regards

                  Mark

                  --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                  >
                  > Well, in truth, there is no such thing as the S/N of a single sub!
                  > That's because noise is the uncertainty in the data. To measure the
                  > total noise in an image, you have to take a lot of images, and then see
                  > how wrong that one image is.
                  >
                  > Of course, we can estimate the noise as read noise + shot noise + any
                  > other instrument noise + (noise from bias, dark, flats). We're not even
                  > considering dynamic range and contrast transfer, which also have a lot
                  > to do with what you can image. And don't forget good old FWHM - the
                  > seeing. And the resolving power of your scope. the quantum efficiency of
                  > your detector. The reflectivity of any optical surfaces (even coated
                  > surfaces reflect a little bit). And then there is the little matter of
                  > detector geometry: what happens to photons that hit various
                  > non-photosensitive portions of the chip? Do they reflect away? Do they
                  > create contrast-reducing scatter? Are the wavelength-speecific artifacts
                  > (e.g., etaloning)?
                  >
                  > And that's just a list of things that start to apply as you get deeper
                  > exposures and start to have to worry about. not just in the
                  > calculations; but in how you process the image (wavelength-specific
                  > etalon effects are a pain in the you-know-what).
                  >
                  > Since every site varies from other sites, and further every site varies
                  > hour by hour (sometimes minute by minute) there just no answer to your
                  > question about 'how long'. In the end, some combination of your patience
                  > and your satisfaction in your work.
                  >
                  > Even your 'hard case' is highly variable - how faint of a wisp? How many
                  > photons per hour? What would you see if you just took another 450
                  > images? <g>
                  >
                  > It is far easier to do some test combines with your data and if you are
                  > not happy, what you can now at least do is think in relatively concrete
                  > terms about what it would take to get a better image. let's say your
                  > faint nebula is barely there in an 8-hour exposure. It's going to take
                  > another 24 hours of exposure time to double the S/N.
                  >
                  > Yikes! maybe 1.4X S/N would be good enough... <g>
                  >
                  > Ron W
                  >
                  > rgr3218893 wrote:
                  > > Ron,
                  > >
                  > > So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...
                  > >
                  > > I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.
                  > >
                  > > What is a target S/N?
                  > >
                  > > Rex
                  > >
                  > > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                  > >> BTW: that doubling S/N thing can readily be calculated. Generally, to
                  > >> double your S/N, you need 4X the present number of images. Thus, if you
                  > >> have 5 images, you need 5*4=20 total to double S/N; once you have 20
                  > >> images, you need 20*4=80 to double your S/N again.
                  > >>
                  > >> Ron W
                  > >>
                  > >> Yahoo - Wodaski wrote:
                  > >>> Here is an Excel graph which shows the diminishing returns from
                  > >>> additional images:
                  > >>>
                  > >>> http://wodaski.com/ex/signal_to_noise.jpg
                  > >>>
                  > >>> The numbers along the bottom are the number of images. The numbers at
                  > >>> left are the approximate % improvement in S/N by adding ONE more image
                  > >>> to the existing stack. The more images you take, the more additional
                  > >>> images you need to take to get a significant improvement in S/N. In
                  > >>> other words, if you take 5 images and want to see another significant
                  > >>> improvement, such as improving dim object signal to noise 2x, then
                  > >>> you'll need another 15 images!
                  > >>>
                  > >>> So you can improve your S/N, but the more images you take, the more
                  > >>> challenging it is to get visible improvement.
                  > >>>
                  > >>> This assumes that individual sub exposures are long enough for shot
                  > >>> noise to swamp read noise.
                  > >>>
                  > >>> Ron Wodaski
                  > >>>
                  > >>> Mark wrote:
                  > >>>> During capture and image stacking at what point does adding more subs not make a difference any more?
                  > >>>> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
                  > >>>>
                  > >>>> Ive always been curious on this.
                  > >>>>
                  > >>>> Regards
                  > >>>>
                  > >>>> Mark
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>> ------------------------------------
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>>
                  > >>>
                  > >>> ------------------------------------
                  > >>>
                  > >>>
                  > >>>
                  > >>>
                  > >
                  > >
                  > >
                  > >
                  > > ------------------------------------
                  > >
                  > >
                  > >
                  > >
                  >
                • dmwmpd
                  Hi Ron, What is your recommendation for the number of darks to take? I have settled on using twice as many darks as lights. My thinking is that reducing the
                  Message 8 of 21 , Jun 9, 2009
                    Hi Ron,

                    What is your recommendation for the number of darks to take? I have settled on using twice as many darks as lights. My thinking is that reducing the random noise in the darks is important so the master dark doesn't contribute as much noise to the lights during calibration (where the same dark is subtracted from every light).

                    Don

                    --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                    >
                    > Well, in truth, there is no such thing as the S/N of a single sub!
                    > That's because noise is the uncertainty in the data. To measure the
                    > total noise in an image, you have to take a lot of images, and then see
                    > how wrong that one image is.
                    >
                    > Of course, we can estimate the noise as read noise + shot noise + any
                    > other instrument noise + (noise from bias, dark, flats). We're not even
                    > considering dynamic range and contrast transfer, which also have a lot
                    > to do with what you can image. And don't forget good old FWHM - the
                    > seeing. And the resolving power of your scope. the quantum efficiency of
                    > your detector. The reflectivity of any optical surfaces (even coated
                    > surfaces reflect a little bit). And then there is the little matter of
                    > detector geometry: what happens to photons that hit various
                    > non-photosensitive portions of the chip? Do they reflect away? Do they
                    > create contrast-reducing scatter? Are the wavelength-speecific artifacts
                    > (e.g., etaloning)?
                    >
                    > And that's just a list of things that start to apply as you get deeper
                    > exposures and start to have to worry about. not just in the
                    > calculations; but in how you process the image (wavelength-specific
                    > etalon effects are a pain in the you-know-what).
                    >
                    > Since every site varies from other sites, and further every site varies
                    > hour by hour (sometimes minute by minute) there just no answer to your
                    > question about 'how long'. In the end, some combination of your patience
                    > and your satisfaction in your work.
                    >
                    > Even your 'hard case' is highly variable - how faint of a wisp? How many
                    > photons per hour? What would you see if you just took another 450
                    > images? <g>
                    >
                    > It is far easier to do some test combines with your data and if you are
                    > not happy, what you can now at least do is think in relatively concrete
                    > terms about what it would take to get a better image. let's say your
                    > faint nebula is barely there in an 8-hour exposure. It's going to take
                    > another 24 hours of exposure time to double the S/N.
                    >
                    > Yikes! maybe 1.4X S/N would be good enough... <g>
                    >
                    > Ron W
                    >
                  • Yahoo - Wodaski
                    Sounds like you already understand the issue: any number of darks will contribute some noise to the light frames. The more darks you cake, the lower the noise
                    Message 9 of 21 , Jun 9, 2009
                      Sounds like you already understand the issue: any number of darks will
                      contribute some noise to the light frames. The more darks you cake, the
                      lower the noise contribution will be. The rate of return is the same as
                      for light frames: to double the S/N in a dark frame, you need 4x as many
                      frames.

                      Ron W

                      dmwmpd wrote:
                      > Hi Ron,
                      >
                      > What is your recommendation for the number of darks to take? I have settled on using twice as many darks as lights. My thinking is that reducing the random noise in the darks is important so the master dark doesn't contribute as much noise to the lights during calibration (where the same dark is subtracted from every light).
                      >
                      > Don
                      >
                      > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                      >> Well, in truth, there is no such thing as the S/N of a single sub!
                      >> That's because noise is the uncertainty in the data. To measure the
                      >> total noise in an image, you have to take a lot of images, and then see
                      >> how wrong that one image is.
                      >>
                      >> Of course, we can estimate the noise as read noise + shot noise + any
                      >> other instrument noise + (noise from bias, dark, flats). We're not even
                      >> considering dynamic range and contrast transfer, which also have a lot
                      >> to do with what you can image. And don't forget good old FWHM - the
                      >> seeing. And the resolving power of your scope. the quantum efficiency of
                      >> your detector. The reflectivity of any optical surfaces (even coated
                      >> surfaces reflect a little bit). And then there is the little matter of
                      >> detector geometry: what happens to photons that hit various
                      >> non-photosensitive portions of the chip? Do they reflect away? Do they
                      >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                      >> (e.g., etaloning)?
                      >>
                      >> And that's just a list of things that start to apply as you get deeper
                      >> exposures and start to have to worry about. not just in the
                      >> calculations; but in how you process the image (wavelength-specific
                      >> etalon effects are a pain in the you-know-what).
                      >>
                      >> Since every site varies from other sites, and further every site varies
                      >> hour by hour (sometimes minute by minute) there just no answer to your
                      >> question about 'how long'. In the end, some combination of your patience
                      >> and your satisfaction in your work.
                      >>
                      >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                      >> photons per hour? What would you see if you just took another 450
                      >> images? <g>
                      >>
                      >> It is far easier to do some test combines with your data and if you are
                      >> not happy, what you can now at least do is think in relatively concrete
                      >> terms about what it would take to get a better image. let's say your
                      >> faint nebula is barely there in an 8-hour exposure. It's going to take
                      >> another 24 hours of exposure time to double the S/N.
                      >>
                      >> Yikes! maybe 1.4X S/N would be good enough... <g>
                      >>
                      >> Ron W
                      >>
                      >
                      >
                      >
                      > ------------------------------------
                      >
                      >
                      >
                      >
                    • Bill Logan
                      Hi Don, I am not an expert as Ron, but in his book, The New CCD Astronomy, page 256, For best results, I recommend that you take more than one dark frame and
                      Message 10 of 21 , Jun 9, 2009
                        Hi Don, I am not an expert as Ron, but in his book, The New CCD Astronomy,
                        page 256, "For best results, I recommend that you take more than one dark
                        frame and median combine them before applying them to your images." As a
                        rule, I have found that five dark frames taken immediately before and five
                        dark frames after my imaging session will produce a nice combined master
                        dark, but then I live on top of a mountain in northern Arizona and blessed
                        with really dark skies and sometimes dark frames aren't needed. I like to
                        inspect each dark frame checking for cosmic ray hits. If one frame has a
                        hit, it's discarded before combining. The reason that I take five before
                        and five after is the change in temperature. I'll get a good master dark
                        after combining. I also like to make sure that I use the same ISO setting
                        and exposure length as my light frames.

                        Some astrophotographers will have a file of master darks at different
                        exposure lengths and temperatures to be used later. I've done that myself,
                        but I have found that taking fresh dark frames before, during and after
                        every session produces betters results.

                        Bill Logan
                        Eagar, Arizona


                        ----- Original Message -----
                        From: "dmwmpd" <westergren@...>
                        To: <ccd-newastro@yahoogroups.com>
                        Sent: Tuesday, June 09, 2009 10:40 AM
                        Subject: [ccd-newastro] Re: Question for Ron or who ever can help.


                        > Hi Ron,
                        >
                        > What is your recommendation for the number of darks to take? I have
                        > settled on using twice as many darks as lights. My thinking is that
                        > reducing the random noise in the darks is important so the master dark
                        > doesn't contribute as much noise to the lights during calibration (where
                        > the same dark is subtracted from every light).
                        >
                        > Don
                        >
                        > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                        >>
                        >> Well, in truth, there is no such thing as the S/N of a single sub!
                        >> That's because noise is the uncertainty in the data. To measure the
                        >> total noise in an image, you have to take a lot of images, and then see
                        >> how wrong that one image is.
                        >>
                        >> Of course, we can estimate the noise as read noise + shot noise + any
                        >> other instrument noise + (noise from bias, dark, flats). We're not even
                        >> considering dynamic range and contrast transfer, which also have a lot
                        >> to do with what you can image. And don't forget good old FWHM - the
                        >> seeing. And the resolving power of your scope. the quantum efficiency of
                        >> your detector. The reflectivity of any optical surfaces (even coated
                        >> surfaces reflect a little bit). And then there is the little matter of
                        >> detector geometry: what happens to photons that hit various
                        >> non-photosensitive portions of the chip? Do they reflect away? Do they
                        >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                        >> (e.g., etaloning)?
                        >>
                        >> And that's just a list of things that start to apply as you get deeper
                        >> exposures and start to have to worry about. not just in the
                        >> calculations; but in how you process the image (wavelength-specific
                        >> etalon effects are a pain in the you-know-what).
                        >>
                        >> Since every site varies from other sites, and further every site varies
                        >> hour by hour (sometimes minute by minute) there just no answer to your
                        >> question about 'how long'. In the end, some combination of your patience
                        >> and your satisfaction in your work.
                        >>
                        >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                        >> photons per hour? What would you see if you just took another 450
                        >> images? <g>
                        >>
                        >> It is far easier to do some test combines with your data and if you are
                        >> not happy, what you can now at least do is think in relatively concrete
                        >> terms about what it would take to get a better image. let's say your
                        >> faint nebula is barely there in an 8-hour exposure. It's going to take
                        >> another 24 hours of exposure time to double the S/N.
                        >>
                        >> Yikes! maybe 1.4X S/N would be good enough... <g>
                        >>
                        >> Ron W
                        >>
                        >
                        >
                        >
                        > ------------------------------------
                        >
                        >
                        >
                        >
                        >
                      • dmwmpd
                        Thanks Ron, I generally take 10 to 12 sets of LRGB images and use a master dark made from 25 darks to calibrate them. With the recent questions about how many
                        Message 11 of 21 , Jun 9, 2009
                          Thanks Ron,

                          I generally take 10 to 12 sets of LRGB images and use a master dark made from 25 darks to calibrate them. With the recent questions about how many lights to take, I thought it would be good to consider how many darks too.

                          Don

                          --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                          >
                          > Sounds like you already understand the issue: any number of darks will
                          > contribute some noise to the light frames. The more darks you cake, the
                          > lower the noise contribution will be. The rate of return is the same as
                          > for light frames: to double the S/N in a dark frame, you need 4x as many
                          > frames.
                          >
                          > Ron W
                          >
                          > dmwmpd wrote:
                          > > Hi Ron,
                          > >
                          > > What is your recommendation for the number of darks to take? I have settled on using twice as many darks as lights. My thinking is that reducing the random noise in the darks is important so the master dark doesn't contribute as much noise to the lights during calibration (where the same dark is subtracted from every light).
                          > >
                          > > Don
                          > >
                          > > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                          > >> Well, in truth, there is no such thing as the S/N of a single sub!
                          > >> That's because noise is the uncertainty in the data. To measure the
                          > >> total noise in an image, you have to take a lot of images, and then see
                          > >> how wrong that one image is.
                          > >>
                          > >> Of course, we can estimate the noise as read noise + shot noise + any
                          > >> other instrument noise + (noise from bias, dark, flats). We're not even
                          > >> considering dynamic range and contrast transfer, which also have a lot
                          > >> to do with what you can image. And don't forget good old FWHM - the
                          > >> seeing. And the resolving power of your scope. the quantum efficiency of
                          > >> your detector. The reflectivity of any optical surfaces (even coated
                          > >> surfaces reflect a little bit). And then there is the little matter of
                          > >> detector geometry: what happens to photons that hit various
                          > >> non-photosensitive portions of the chip? Do they reflect away? Do they
                          > >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                          > >> (e.g., etaloning)?
                          > >>
                          > >> And that's just a list of things that start to apply as you get deeper
                          > >> exposures and start to have to worry about. not just in the
                          > >> calculations; but in how you process the image (wavelength-specific
                          > >> etalon effects are a pain in the you-know-what).
                          > >>
                          > >> Since every site varies from other sites, and further every site varies
                          > >> hour by hour (sometimes minute by minute) there just no answer to your
                          > >> question about 'how long'. In the end, some combination of your patience
                          > >> and your satisfaction in your work.
                          > >>
                          > >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                          > >> photons per hour? What would you see if you just took another 450
                          > >> images? <g>
                          > >>
                          > >> It is far easier to do some test combines with your data and if you are
                          > >> not happy, what you can now at least do is think in relatively concrete
                          > >> terms about what it would take to get a better image. let's say your
                          > >> faint nebula is barely there in an 8-hour exposure. It's going to take
                          > >> another 24 hours of exposure time to double the S/N.
                          > >>
                          > >> Yikes! maybe 1.4X S/N would be good enough... <g>
                          > >>
                          > >> Ron W
                          > >>
                          > >
                          > >
                          > >
                          > > ------------------------------------
                          > >
                          > >
                          > >
                          > >
                          >
                        • Neil Fleming
                          I d suggest a minimum of 25...fewer than that and the master dark will/may contribute too much noise during calibration.  CCDWare has a calculator on their
                          Message 12 of 21 , Jun 9, 2009
                            I'd suggest a minimum of 25...fewer than that and the master dark will/may contribute too much noise during calibration.  CCDWare has a calculator on their web site to help determine the number for your camera, part of the subexposure calculator.
                             
                               Regards...Neil


                            www.flemingastrophotography.com 
                            Direct from Boston - brilliant diamonds in pea soup
                            Also check out the astro_narrowband Yahoo group!

                            --- On Tue, 6/9/09, Bill Logan <wb9sat@...> wrote:


                            From: Bill Logan <wb9sat@...>
                            Subject: Re: [ccd-newastro] Re: Question for Ron or who ever can help.
                            To: ccd-newastro@yahoogroups.com
                            Date: Tuesday, June 9, 2009, 3:07 PM


                            Hi Don, I am not an expert as Ron, but in his book, The New CCD Astronomy,
                            page 256, "For best results, I recommend that you take more than one dark
                            frame and median combine them before applying them to your images."  As a
                            rule, I have found that five dark frames taken immediately before and five
                            dark frames after my imaging session will produce a nice combined master
                            dark, but then I live on top of a mountain in northern Arizona and blessed
                            with really dark skies and sometimes dark frames aren't needed.  I like to
                            inspect each dark frame checking for cosmic ray hits.  If one frame has a
                            hit, it's discarded before combining.  The reason that I take five before
                            and five after is the change in temperature.  I'll get a good master dark
                            after combining.  I also like to make sure that I use the same ISO setting
                            and exposure length as my light frames.

                            Some astrophotographers will have a file of master darks at different
                            exposure lengths and temperatures to be used later.  I've done that myself,
                            but I have found that taking fresh dark frames before, during and after
                            every session produces betters results.

                            Bill Logan
                            Eagar, Arizona


                            ----- Original Message -----
                            From: "dmwmpd" <westergren@...>
                            To: <ccd-newastro@yahoogroups.com>
                            Sent: Tuesday, June 09, 2009 10:40 AM
                            Subject: [ccd-newastro] Re: Question for Ron or who ever can help.


                            > Hi Ron,
                            >
                            > What is your recommendation for the number of darks to take?  I have
                            > settled on using twice as many darks as lights.  My thinking is that
                            > reducing the random noise in the darks is important so the master dark
                            > doesn't contribute as much noise to the lights during calibration (where
                            > the same dark is subtracted from every light).
                            >
                            > Don
                            >
                            > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                            >>
                            >> Well, in truth, there is no such thing as the S/N of a single sub!
                            >> That's because noise is the uncertainty in the data. To measure the
                            >> total noise in an image, you have to take a lot of images, and then see
                            >> how wrong that one image is.
                            >>
                            >> Of course, we can estimate the noise as read noise + shot noise + any
                            >> other instrument noise + (noise from bias, dark, flats). We're not even
                            >> considering dynamic range and contrast transfer, which also have a lot
                            >> to do with what you can image. And don't forget good old FWHM - the
                            >> seeing. And the resolving power of your scope. the quantum efficiency of
                            >> your detector. The reflectivity of any optical surfaces (even coated
                            >> surfaces reflect a little bit). And then there is the little matter of
                            >> detector geometry: what happens to photons that hit various
                            >> non-photosensitive portions of the chip? Do they reflect away? Do they
                            >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                            >> (e.g., etaloning)?
                            >>
                            >> And that's just a list of things that start to apply as you get deeper
                            >> exposures and start to have to worry about. not just in the
                            >> calculations; but in how you process the image (wavelength-specific
                            >> etalon effects are a pain in the you-know-what).
                            >>
                            >> Since every site varies from other sites, and further every site varies
                            >> hour by hour (sometimes minute by minute) there just no answer to your
                            >> question about 'how long'. In the end, some combination of your patience
                            >> and your satisfaction in your work.
                            >>
                            >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                            >> photons per hour? What would you see if you just took another 450
                            >> images? <g>
                            >>
                            >> It is far easier to do some test combines with your data and if you are
                            >> not happy, what you can now at least do is think in relatively concrete
                            >> terms about what it would take to get a better image. let's say your
                            >> faint nebula is barely there in an 8-hour exposure. It's going to take
                            >> another 24 hours of exposure time to double the S/N.
                            >>
                            >> Yikes! maybe 1.4X S/N would be good enough... <g>
                            >>
                            >> Ron W
                            >>
                            >
                            >
                            >
                            > ------------------------------------
                            >
                            >
                            >
                            >
                            >



                            ------------------------------------






                            [Non-text portions of this message have been removed]
                          • Bill Logan
                            Thanks Neil. I looked up the CCDWare site and could not find the calculator. Could you provide a link? Bill Logan Eagar, Arizona ... From: Neil Fleming
                            Message 13 of 21 , Jun 9, 2009
                              Thanks Neil. I looked up the CCDWare site and could not find the
                              calculator. Could you provide a link?

                              Bill Logan
                              Eagar, Arizona


                              ----- Original Message -----
                              From: "Neil Fleming" <neilfleming@...>
                              To: <ccd-newastro@yahoogroups.com>
                              Sent: Tuesday, June 09, 2009 4:20 PM
                              Subject: Re: [ccd-newastro] Re: Question for Ron or who ever can help.


                              I'd suggest a minimum of 25...fewer than that and the master dark will/may
                              contribute too much noise during calibration. CCDWare has a calculator on
                              their web site to help determine the number for your camera, part of the
                              subexposure calculator.

                              Regards...Neil


                              www.flemingastrophotography.com
                              Direct from Boston - brilliant diamonds in pea soup
                              Also check out the astro_narrowband Yahoo group!

                              --- On Tue, 6/9/09, Bill Logan <wb9sat@...> wrote:


                              From: Bill Logan <wb9sat@...>
                              Subject: Re: [ccd-newastro] Re: Question for Ron or who ever can help.
                              To: ccd-newastro@yahoogroups.com
                              Date: Tuesday, June 9, 2009, 3:07 PM


                              Hi Don, I am not an expert as Ron, but in his book, The New CCD Astronomy,
                              page 256, "For best results, I recommend that you take more than one dark
                              frame and median combine them before applying them to your images." As a
                              rule, I have found that five dark frames taken immediately before and five
                              dark frames after my imaging session will produce a nice combined master
                              dark, but then I live on top of a mountain in northern Arizona and blessed
                              with really dark skies and sometimes dark frames aren't needed. I like to
                              inspect each dark frame checking for cosmic ray hits. If one frame has a
                              hit, it's discarded before combining. The reason that I take five before
                              and five after is the change in temperature. I'll get a good master dark
                              after combining. I also like to make sure that I use the same ISO setting
                              and exposure length as my light frames.

                              Some astrophotographers will have a file of master darks at different
                              exposure lengths and temperatures to be used later. I've done that myself,
                              but I have found that taking fresh dark frames before, during and after
                              every session produces betters results.

                              Bill Logan
                              Eagar, Arizona


                              ----- Original Message -----
                              From: "dmwmpd" <westergren@...>
                              To: <ccd-newastro@yahoogroups.com>
                              Sent: Tuesday, June 09, 2009 10:40 AM
                              Subject: [ccd-newastro] Re: Question for Ron or who ever can help.


                              > Hi Ron,
                              >
                              > What is your recommendation for the number of darks to take? I have
                              > settled on using twice as many darks as lights. My thinking is that
                              > reducing the random noise in the darks is important so the master dark
                              > doesn't contribute as much noise to the lights during calibration (where
                              > the same dark is subtracted from every light).
                              >
                              > Don
                              >
                              > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                              >>
                              >> Well, in truth, there is no such thing as the S/N of a single sub!
                              >> That's because noise is the uncertainty in the data. To measure the
                              >> total noise in an image, you have to take a lot of images, and then see
                              >> how wrong that one image is.
                              >>
                              >> Of course, we can estimate the noise as read noise + shot noise + any
                              >> other instrument noise + (noise from bias, dark, flats). We're not even
                              >> considering dynamic range and contrast transfer, which also have a lot
                              >> to do with what you can image. And don't forget good old FWHM - the
                              >> seeing. And the resolving power of your scope. the quantum efficiency of
                              >> your detector. The reflectivity of any optical surfaces (even coated
                              >> surfaces reflect a little bit). And then there is the little matter of
                              >> detector geometry: what happens to photons that hit various
                              >> non-photosensitive portions of the chip? Do they reflect away? Do they
                              >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                              >> (e.g., etaloning)?
                              >>
                              >> And that's just a list of things that start to apply as you get deeper
                              >> exposures and start to have to worry about. not just in the
                              >> calculations; but in how you process the image (wavelength-specific
                              >> etalon effects are a pain in the you-know-what).
                              >>
                              >> Since every site varies from other sites, and further every site varies
                              >> hour by hour (sometimes minute by minute) there just no answer to your
                              >> question about 'how long'. In the end, some combination of your patience
                              >> and your satisfaction in your work.
                              >>
                              >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                              >> photons per hour? What would you see if you just took another 450
                              >> images? <g>
                              >>
                              >> It is far easier to do some test combines with your data and if you are
                              >> not happy, what you can now at least do is think in relatively concrete
                              >> terms about what it would take to get a better image. let's say your
                              >> faint nebula is barely there in an 8-hour exposure. It's going to take
                              >> another 24 hours of exposure time to double the S/N.
                              >>
                              >> Yikes! maybe 1.4X S/N would be good enough... <g>
                              >>
                              >> Ron W
                              >>
                              >
                              >
                              >
                              > ------------------------------------
                              >
                              >
                              >
                              >
                              >



                              ------------------------------------






                              [Non-text portions of this message have been removed]



                              ------------------------------------
                            • rgr3218893
                              Huh?!! Ron...you could run for public office avoiding committing to a final answer Let s assume that in our nebula, the wisp was the faintest of wisps if
                              Message 14 of 21 , Jun 9, 2009
                                Huh?!! Ron...you could run for public office avoiding committing to a final answer <g>

                                Let's assume that in our nebula, the wisp was the faintest of wisps if ever a wisp there was...

                                what typical S/N do you get at your NM site?

                                Rex

                                --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                                >
                                > Well, in truth, there is no such thing as the S/N of a single sub!
                                > That's because noise is the uncertainty in the data. To measure the
                                > total noise in an image, you have to take a lot of images, and then see
                                > how wrong that one image is.
                                >
                                > Of course, we can estimate the noise as read noise + shot noise + any
                                > other instrument noise + (noise from bias, dark, flats). We're not even
                                > considering dynamic range and contrast transfer, which also have a lot
                                > to do with what you can image. And don't forget good old FWHM - the
                                > seeing. And the resolving power of your scope. the quantum efficiency of
                                > your detector. The reflectivity of any optical surfaces (even coated
                                > surfaces reflect a little bit). And then there is the little matter of
                                > detector geometry: what happens to photons that hit various
                                > non-photosensitive portions of the chip? Do they reflect away? Do they
                                > create contrast-reducing scatter? Are the wavelength-speecific artifacts
                                > (e.g., etaloning)?
                                >
                                > And that's just a list of things that start to apply as you get deeper
                                > exposures and start to have to worry about. not just in the
                                > calculations; but in how you process the image (wavelength-specific
                                > etalon effects are a pain in the you-know-what).
                                >
                                > Since every site varies from other sites, and further every site varies
                                > hour by hour (sometimes minute by minute) there just no answer to your
                                > question about 'how long'. In the end, some combination of your patience
                                > and your satisfaction in your work.
                                >
                                > Even your 'hard case' is highly variable - how faint of a wisp? How many
                                > photons per hour? What would you see if you just took another 450
                                > images? <g>
                                >
                                > It is far easier to do some test combines with your data and if you are
                                > not happy, what you can now at least do is think in relatively concrete
                                > terms about what it would take to get a better image. let's say your
                                > faint nebula is barely there in an 8-hour exposure. It's going to take
                                > another 24 hours of exposure time to double the S/N.
                                >
                                > Yikes! maybe 1.4X S/N would be good enough... <g>
                                >
                                > Ron W
                                >
                                > rgr3218893 wrote:
                                > > Ron,
                                > >
                                > > So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...
                                > >
                                > > I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.
                                > >
                                > > What is a target S/N?
                                > >
                                > > Rex
                                > >
                                > > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                                > >> BTW: that doubling S/N thing can readily be calculated. Generally, to
                                > >> double your S/N, you need 4X the present number of images. Thus, if you
                                > >> have 5 images, you need 5*4=20 total to double S/N; once you have 20
                                > >> images, you need 20*4=80 to double your S/N again.
                                > >>
                                > >> Ron W
                                > >>
                                > >> Yahoo - Wodaski wrote:
                                > >>> Here is an Excel graph which shows the diminishing returns from
                                > >>> additional images:
                                > >>>
                                > >>> http://wodaski.com/ex/signal_to_noise.jpg
                                > >>>
                                > >>> The numbers along the bottom are the number of images. The numbers at
                                > >>> left are the approximate % improvement in S/N by adding ONE more image
                                > >>> to the existing stack. The more images you take, the more additional
                                > >>> images you need to take to get a significant improvement in S/N. In
                                > >>> other words, if you take 5 images and want to see another significant
                                > >>> improvement, such as improving dim object signal to noise 2x, then
                                > >>> you'll need another 15 images!
                                > >>>
                                > >>> So you can improve your S/N, but the more images you take, the more
                                > >>> challenging it is to get visible improvement.
                                > >>>
                                > >>> This assumes that individual sub exposures are long enough for shot
                                > >>> noise to swamp read noise.
                                > >>>
                                > >>> Ron Wodaski
                                > >>>
                                > >>> Mark wrote:
                                > >>>> During capture and image stacking at what point does adding more subs not make a difference any more?
                                > >>>> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
                                > >>>>
                                > >>>> Ive always been curious on this.
                                > >>>>
                                > >>>> Regards
                                > >>>>
                                > >>>> Mark
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>> ------------------------------------
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>>
                                > >>>
                                > >>> ------------------------------------
                                > >>>
                                > >>>
                                > >>>
                                > >>>
                                > >
                                > >
                                > >
                                > >
                                > > ------------------------------------
                                > >
                                > >
                                > >
                                > >
                                >
                              • rgr3218893
                                For your final combined image in one of your best nebular wisps
                                Message 15 of 21 , Jun 9, 2009
                                  For your final combined image in one of your best nebular wisps <g>

                                  --- In ccd-newastro@yahoogroups.com, "rgr3218893" <rgr3218893@...> wrote:
                                  >
                                  > Huh?!! Ron...you could run for public office avoiding committing to a final answer <g>
                                  >
                                  > Let's assume that in our nebula, the wisp was the faintest of wisps if ever a wisp there was...
                                  >
                                  > what typical S/N do you get at your NM site?
                                  >
                                  > Rex
                                  >
                                  > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                                  > >
                                  > > Well, in truth, there is no such thing as the S/N of a single sub!
                                  > > That's because noise is the uncertainty in the data. To measure the
                                  > > total noise in an image, you have to take a lot of images, and then see
                                  > > how wrong that one image is.
                                  > >
                                  > > Of course, we can estimate the noise as read noise + shot noise + any
                                  > > other instrument noise + (noise from bias, dark, flats). We're not even
                                  > > considering dynamic range and contrast transfer, which also have a lot
                                  > > to do with what you can image. And don't forget good old FWHM - the
                                  > > seeing. And the resolving power of your scope. the quantum efficiency of
                                  > > your detector. The reflectivity of any optical surfaces (even coated
                                  > > surfaces reflect a little bit). And then there is the little matter of
                                  > > detector geometry: what happens to photons that hit various
                                  > > non-photosensitive portions of the chip? Do they reflect away? Do they
                                  > > create contrast-reducing scatter? Are the wavelength-speecific artifacts
                                  > > (e.g., etaloning)?
                                  > >
                                  > > And that's just a list of things that start to apply as you get deeper
                                  > > exposures and start to have to worry about. not just in the
                                  > > calculations; but in how you process the image (wavelength-specific
                                  > > etalon effects are a pain in the you-know-what).
                                  > >
                                  > > Since every site varies from other sites, and further every site varies
                                  > > hour by hour (sometimes minute by minute) there just no answer to your
                                  > > question about 'how long'. In the end, some combination of your patience
                                  > > and your satisfaction in your work.
                                  > >
                                  > > Even your 'hard case' is highly variable - how faint of a wisp? How many
                                  > > photons per hour? What would you see if you just took another 450
                                  > > images? <g>
                                  > >
                                  > > It is far easier to do some test combines with your data and if you are
                                  > > not happy, what you can now at least do is think in relatively concrete
                                  > > terms about what it would take to get a better image. let's say your
                                  > > faint nebula is barely there in an 8-hour exposure. It's going to take
                                  > > another 24 hours of exposure time to double the S/N.
                                  > >
                                  > > Yikes! maybe 1.4X S/N would be good enough... <g>
                                  > >
                                  > > Ron W
                                  > >
                                  > > rgr3218893 wrote:
                                  > > > Ron,
                                  > > >
                                  > > > So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...
                                  > > >
                                  > > > I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.
                                  > > >
                                  > > > What is a target S/N?
                                  > > >
                                  > > > Rex
                                  > > >
                                  > > > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                                  > > >> BTW: that doubling S/N thing can readily be calculated. Generally, to
                                  > > >> double your S/N, you need 4X the present number of images. Thus, if you
                                  > > >> have 5 images, you need 5*4=20 total to double S/N; once you have 20
                                  > > >> images, you need 20*4=80 to double your S/N again.
                                  > > >>
                                  > > >> Ron W
                                  > > >>
                                  > > >> Yahoo - Wodaski wrote:
                                  > > >>> Here is an Excel graph which shows the diminishing returns from
                                  > > >>> additional images:
                                  > > >>>
                                  > > >>> http://wodaski.com/ex/signal_to_noise.jpg
                                  > > >>>
                                  > > >>> The numbers along the bottom are the number of images. The numbers at
                                  > > >>> left are the approximate % improvement in S/N by adding ONE more image
                                  > > >>> to the existing stack. The more images you take, the more additional
                                  > > >>> images you need to take to get a significant improvement in S/N. In
                                  > > >>> other words, if you take 5 images and want to see another significant
                                  > > >>> improvement, such as improving dim object signal to noise 2x, then
                                  > > >>> you'll need another 15 images!
                                  > > >>>
                                  > > >>> So you can improve your S/N, but the more images you take, the more
                                  > > >>> challenging it is to get visible improvement.
                                  > > >>>
                                  > > >>> This assumes that individual sub exposures are long enough for shot
                                  > > >>> noise to swamp read noise.
                                  > > >>>
                                  > > >>> Ron Wodaski
                                  > > >>>
                                  > > >>> Mark wrote:
                                  > > >>>> During capture and image stacking at what point does adding more subs not make a difference any more?
                                  > > >>>> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
                                  > > >>>>
                                  > > >>>> Ive always been curious on this.
                                  > > >>>>
                                  > > >>>> Regards
                                  > > >>>>
                                  > > >>>> Mark
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>> ------------------------------------
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>>
                                  > > >>>
                                  > > >>> ------------------------------------
                                  > > >>>
                                  > > >>>
                                  > > >>>
                                  > > >>>
                                  > > >
                                  > > >
                                  > > >
                                  > > >
                                  > > > ------------------------------------
                                  > > >
                                  > > >
                                  > > >
                                  > > >
                                  > >
                                  >
                                • Eagle Station
                                  http://www.ccdware.com/resources/                                             ... From: Bill Logan
                                  Message 16 of 21 , Jun 9, 2009
                                    http://www.ccdware.com/resources/


                                     
                                                                           
                                     

                                    --- On Tue, 6/9/09, Bill Logan <wb9sat@...> wrote:


                                    From: Bill Logan <wb9sat@...>
                                    Subject: Re: [ccd-newastro] Re: Question for Ron or who ever can help.
                                    To: ccd-newastro@yahoogroups.com
                                    Date: Tuesday, June 9, 2009, 8:12 PM


                                    Thanks Neil.  I looked up the CCDWare site and could not find the
                                    calculator.  Could you provide a link?

                                    Bill Logan
                                    Eagar, Arizona
                                  • Yahoo - Wodaski
                                    There is no answer. It really isn t an issue of what S/N you get at this site or that site (although obviously dark skies are going to help). It really
                                    Message 17 of 21 , Jun 9, 2009
                                      There is no answer. <g> It really isn't an issue of what S/N you get at
                                      this site or that site (although obviously dark skies are going to
                                      help). It really does come down to all those variables. And they get
                                      harder and harder to control. So it's not just patience; at some point,
                                      you have to start getting very clever at controlling the variable.

                                      The only limit is how hard you are willing to work at it. <g>

                                      Ron W

                                      rgr3218893 wrote:
                                      > Huh?!! Ron...you could run for public office avoiding committing to a final answer <g>
                                      >
                                      > Let's assume that in our nebula, the wisp was the faintest of wisps if ever a wisp there was...
                                      >
                                      > what typical S/N do you get at your NM site?
                                      >
                                      > Rex
                                      >
                                      > --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@...> wrote:
                                      >> Well, in truth, there is no such thing as the S/N of a single sub!
                                      >> That's because noise is the uncertainty in the data. To measure the
                                      >> total noise in an image, you have to take a lot of images, and then see
                                      >> how wrong that one image is.
                                      >>
                                      >> Of course, we can estimate the noise as read noise + shot noise + any
                                      >> other instrument noise + (noise from bias, dark, flats). We're not even
                                      >> considering dynamic range and contrast transfer, which also have a lot
                                      >> to do with what you can image. And don't forget good old FWHM - the
                                      >> seeing. And the resolving power of your scope. the quantum efficiency of
                                      >> your detector. The reflectivity of any optical surfaces (even coated
                                      >> surfaces reflect a little bit). And then there is the little matter of
                                      >> detector geometry: what happens to photons that hit various
                                      >> non-photosensitive portions of the chip? Do they reflect away? Do they
                                      >> create contrast-reducing scatter? Are the wavelength-speecific artifacts
                                      >> (e.g., etaloning)?
                                      >>
                                      >> And that's just a list of things that start to apply as you get deeper
                                      >> exposures and start to have to worry about. not just in the
                                      >> calculations; but in how you process the image (wavelength-specific
                                      >> etalon effects are a pain in the you-know-what).
                                      >>
                                      >> Since every site varies from other sites, and further every site varies
                                      >> hour by hour (sometimes minute by minute) there just no answer to your
                                      >> question about 'how long'. In the end, some combination of your patience
                                      >> and your satisfaction in your work.
                                      >>
                                      >> Even your 'hard case' is highly variable - how faint of a wisp? How many
                                      >> photons per hour? What would you see if you just took another 450
                                      >> images? <g>
                                      >>
                                      >> It is far easier to do some test combines with your data and if you are
                                      >> not happy, what you can now at least do is think in relatively concrete
                                      >> terms about what it would take to get a better image. let's say your
                                      >> faint nebula is barely there in an 8-hour exposure. It's going to take
                                      >> another 24 hours of exposure time to double the S/N.
                                      >>
                                      >> Yikes! maybe 1.4X S/N would be good enough... <g>
                                      >>
                                      >> Ron W
                                      >>
                                      >> rgr3218893 wrote:
                                      >>> Ron,
                                      >>>
                                      >>> So in the end....when we measure our S/N of a single sub...what is a rule of thumb for a final? Is a S/N of 20, 30 or higher good enough for "pretty pictures"?...assuming that time is not a factor. At what S/N point does the eye or processing mask out any sub improvements...
                                      >>>
                                      >>> I know that depends on the object, but let's pick a hard case....faint wisps of nebula or outer regions of a galaxy.
                                      >>>
                                      >>> What is a target S/N?
                                      >>>
                                      >>> Rex
                                      >>>
                                      >>> --- In ccd-newastro@yahoogroups.com, Yahoo - Wodaski <yahoo@> wrote:
                                      >>>> BTW: that doubling S/N thing can readily be calculated. Generally, to
                                      >>>> double your S/N, you need 4X the present number of images. Thus, if you
                                      >>>> have 5 images, you need 5*4=20 total to double S/N; once you have 20
                                      >>>> images, you need 20*4=80 to double your S/N again.
                                      >>>>
                                      >>>> Ron W
                                      >>>>
                                      >>>> Yahoo - Wodaski wrote:
                                      >>>>> Here is an Excel graph which shows the diminishing returns from
                                      >>>>> additional images:
                                      >>>>>
                                      >>>>> http://wodaski.com/ex/signal_to_noise.jpg
                                      >>>>>
                                      >>>>> The numbers along the bottom are the number of images. The numbers at
                                      >>>>> left are the approximate % improvement in S/N by adding ONE more image
                                      >>>>> to the existing stack. The more images you take, the more additional
                                      >>>>> images you need to take to get a significant improvement in S/N. In
                                      >>>>> other words, if you take 5 images and want to see another significant
                                      >>>>> improvement, such as improving dim object signal to noise 2x, then
                                      >>>>> you'll need another 15 images!
                                      >>>>>
                                      >>>>> So you can improve your S/N, but the more images you take, the more
                                      >>>>> challenging it is to get visible improvement.
                                      >>>>>
                                      >>>>> This assumes that individual sub exposures are long enough for shot
                                      >>>>> noise to swamp read noise.
                                      >>>>>
                                      >>>>> Ron Wodaski
                                      >>>>>
                                      >>>>> Mark wrote:
                                      >>>>>> During capture and image stacking at what point does adding more subs not make a difference any more?
                                      >>>>>> I know the more images stacked improves the signal to noise ratio but when is it mute to quit adding time or exposures to an image?
                                      >>>>>>
                                      >>>>>> Ive always been curious on this.
                                      >>>>>>
                                      >>>>>> Regards
                                      >>>>>>
                                      >>>>>> Mark
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>> ------------------------------------
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>>>
                                      >>>>> ------------------------------------
                                      >>>>>
                                      >>>>>
                                      >>>>>
                                      >>>>>
                                      >>>
                                      >>>
                                      >>>
                                      >>> ------------------------------------
                                      >>>
                                      >>>
                                      >>>
                                      >>>
                                      >
                                      >
                                      >
                                      >
                                      > ------------------------------------
                                      >
                                      >
                                      >
                                      >
                                    Your message has been successfully submitted and would be delivered to recipients shortly.