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

RE: [ccd-newastro] gradients & normalization

Expand Messages
  • Ron Wodaski
    You are dealing with two very different reference systems here. I ve never had success this way, at least not without considerable mental gymnastics to try to
    Message 1 of 7 , Jan 31, 2002
    • 0 Attachment
      You are dealing with two very different reference systems here. I've never
      had success this way, at least not without considerable mental gymnastics to
      try to get the two files to have similar levels.

      The simplest solution is to apply the gradient removal to the image in
      Photoshop. Photoshop just doesn't seem to understand the 16-bit format when
      it comes to saving. I've heard that it actually only saves 15 bits, not 16,
      and if this is true it would be just about impossible to use Photoshop to
      create a gradient removal image in the way you describe.

      Since I had so much trouble with the approach, I only documented the
      all-Photoshop methods in the book. I just can't recommend this approach
      because Photoshop seems intent on making it not work.

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

      -----Original Message-----
      From: jfeldhou [mailto:jfeldhou@...]
      Sent: Thursday, January 31, 2002 5:05 PM
      To: ccd-newastro@yahoogroups.com
      Subject: [ccd-newastro] gradients & normalization


      is it a bad idea to attempt to produce a gradient
      from an image that has been normalized during stacking?

      I attempted to do the following procedure, with very
      poor results. I had 11x600s luminence images,
      preprocessed, stacked(sigmamean). M51 occupied most
      of the corner quadrant, the rest stars of varying bloat.

      I saved the stacked fits image as 16 bit tiff, then
      opened it in photoshop. I used the dust&scratches
      tool at (12,16) cause some stars were big. Then I clone
      tooled out any remaining stars, and the M51 portion.
      after a gausian blur 19, I copied the 16 bit tif back
      to my PC, changed it back to fits, and attempt to
      subtract the blur from my original M51 stack.
      I attempted to subtract the blur, after multiplying
      the blurred pixels by .8 then .7 then .5 .4

      no matter my choice, the target image was erradicated.
      my blur wasn't so good in the area of m51, but I had
      hoped I'd see some worthwhile results on the rest of
      the target frame, and then I could make a better blur.

      is there an obvious place where I went wrong?
      thanks, jim



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



      Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
    • jfeldhou
      this is good to know. more questions... do I need to switch to 8 bit to do the combine in photoshop? I was unable to get layers, nor Apply image to work in
      Message 2 of 7 , Feb 1, 2002
      • 0 Attachment
        this is good to know.
        more questions...

        do I need to switch to 8 bit to do the combine in photoshop?
        I was unable to get layers, nor "Apply image" to work in
        15/16 bit mode? is switching to 8 bits before doing advanced
        processing (ddp,unsharp,L-R) a bad idea?
        is switching to 8 bits a bad idea in general?

        thanks, jim


        --- In ccd-newastro@y..., "Ron Wodaski" <ronw@n...> wrote:
        > You are dealing with two very different reference systems here. I've
        never
        > had success this way, at least not without considerable mental
        gymnastics to
        > try to get the two files to have similar levels.
        >
        > The simplest solution is to apply the gradient removal to the image
        in
        > Photoshop. Photoshop just doesn't seem to understand the 16-bit
        format when
        > it comes to saving. I've heard that it actually only saves 15 bits,
        not 16,
        > and if this is true it would be just about impossible to use
        Photoshop to
        > create a gradient removal image in the way you describe.
        >
        > Since I had so much trouble with the approach, I only documented the
        > all-Photoshop methods in the book. I just can't recommend this
        approach
        > because Photoshop seems intent on making it not work.
        >
        > Ron Wodaski
        > The New CCD Astronomy
        > http://www.newastro.com
        >
        > -----Original Message-----
        > From: jfeldhou [mailto:jfeldhou@y...]
        > Sent: Thursday, January 31, 2002 5:05 PM
        > To: ccd-newastro@y...
        > Subject: [ccd-newastro] gradients & normalization
        >
        >
        > is it a bad idea to attempt to produce a gradient
        > from an image that has been normalized during stacking?
        >
        > I attempted to do the following procedure, with very
        > poor results. I had 11x600s luminence images,
        > preprocessed, stacked(sigmamean). M51 occupied most
        > of the corner quadrant, the rest stars of varying bloat.
        >
        > I saved the stacked fits image as 16 bit tiff, then
        > opened it in photoshop. I used the dust&scratches
        > tool at (12,16) cause some stars were big. Then I clone
        > tooled out any remaining stars, and the M51 portion.
        > after a gausian blur 19, I copied the 16 bit tif back
        > to my PC, changed it back to fits, and attempt to
        > subtract the blur from my original M51 stack.
        > I attempted to subtract the blur, after multiplying
        > the blurred pixels by .8 then .7 then .5 .4
        >
        > no matter my choice, the target image was erradicated.
        > my blur wasn't so good in the area of m51, but I had
        > hoped I'd see some worthwhile results on the rest of
        > the target frame, and then I could make a better blur.
        >
        > is there an obvious place where I went wrong?
        > thanks, jim
        >
        >
        >
        > To unsubscribe from this group, send an email to:
        > ccd-newastro-unsubscribe@e...
        >
        >
        >
        > Your use of Yahoo! Groups is subject to
        http://docs.yahoo.com/info/terms/
      • Mark Jenkins
        Hi Jim, The short answer is yes. You cannot perform many operations in PS without changing to 8 bit mode. In PS 6 certain filters are available in 16bit like
        Message 3 of 7 , Feb 1, 2002
        • 0 Attachment
          Hi Jim,

          The short answer is yes. You cannot perform many operations in PS
          without changing to 8 bit mode.

          In PS 6 certain filters are available in 16bit like Unsharp Mask, and
          Gaussian Blur and a few others.

          As many histogram adjustments as possible should be done in 16 bit but
          some tweaking of the histogram can be done in 8 bit mode if needed.

          Switching to 8 bit is not necessarily a "bad idea" simply because it is
          necessary.

          HTH

          Mark Jenkins
          markj@...

          On Friday, February 1, 2002, at 02:39 AM, jfeldhou wrote:

          > this is good to know.
          > more questions...
          >
          > do I need to switch to 8 bit to do the combine in photoshop?
          > I was unable to get layers, nor "Apply image" to work in
          > 15/16 bit mode? is switching to 8 bits before doing advanced
          > processing (ddp,unsharp,L-R) a bad idea?
          > is switching to 8 bits a bad idea in general?
        • Matt Russell
          Mark, I thought when you trimmed back to 8 bits in Photoshop, you actually lose a lot of data? Can someone clairify this for me? Thanks, Matt
          Message 4 of 7 , Feb 1, 2002
          • 0 Attachment
            Mark,

            I thought when you trimmed back to 8 bits in Photoshop, you actually lose a
            lot of data? Can someone clairify this for me?

            Thanks,

            Matt
          • Ron Wodaski
            When you convert to 8 bits, you are rescaling the data to 256 levels instead of 65,000. So there is a big loss in precision. HOWEVER: for most images, the vast
            Message 5 of 7 , Feb 1, 2002
            • 0 Attachment
              When you convert to 8 bits, you are rescaling the data to 256 levels instead
              of 65,000. So there is a big loss in precision. HOWEVER: for most images,
              the vast bulk of the useful data is actually packed into a very, very small
              range of values (some small percentage of those 65,000 values, as evidenced
              by the thin clump of data at the left end of the raw histogram). Histogram
              adjustments, linear and non-linear, have the effect of discarding specially
              the portions of the data that are less interesting. After you complete this
              process, you have carefully decided yourself what to discard and what to
              keep. Converting to 8 bits after making these decisions means that you lose
              very little meaningful data (if any) when you convert to 8 bits.

              So they key is converting to 8 bits of data at the right moment. See another
              message I am posting about the right order for making changes and converting
              to 8 bits.

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

              -----Original Message-----
              From: Matt Russell [mailto:matthewr@...]
              Sent: Friday, February 01, 2002 9:11 AM
              To: 'ccd-newastro@yahoogroups.com'
              Subject: [ccd-newastro] Re: gradients & normalization


              Mark,

              I thought when you trimmed back to 8 bits in Photoshop, you actually lose a
              lot of data? Can someone clairify this for me?

              Thanks,

              Matt


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



              Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
            • Ron Wodaski
              You want to postpone the change to 8 bits to as late in the process as possible, for the reasons outlined in another reply I posted today. With respect to the
              Message 6 of 7 , Feb 1, 2002
              • 0 Attachment
                You want to postpone the change to 8 bits to as late in the process as
                possible, for the reasons outlined in another reply I posted today. With
                respect to the questions:

                * Yes, you need to eventually switch to 8 bits. Some Photoshop tools only
                work in 8-bit mode, as you've discovered. And you must convert to 8 bits to
                save as JPG anyway.

                * The order in which you do certain processing steps is critical. I rarely,
                if ever, use DDP if I am going to use Photoshop. The histogram tools in
                Photoshop allow you to do a much more precise manual form of DDP. And you
                can use Photoshop's unsharp mask tool in conjunction with a selection to
                sharpen only the portions of the image with good enough S/N for sharpening
                to be effective. Specialized operations like deconvolution should be done on
                the 16-bit file BEFORE you ever bring it into Photoshop. Once you have the
                image in Photoshop, you should finish with it there. Go as far as you can
                and convert to 8-bit mode at the last possible moment, when you have no
                further use for the 16-bit tools. If you do this, the data left when do the
                conversion will suffer minimal loss during the conversion to 8 bits.

                So:

                * Perform specialized processing (deconvolution and other astro-only types
                of tools) to the 16-bit file BEFORE you ever open it in Photoshop. Always
                retain your raw data in case you need to start over!!!

                * Open the image in Photoshop and do linear and non-linear histogram
                changes. This discards less interesting portions of the data, and retains
                the most interesting portions.

                * Convert to 8-bit mode, and now you can use of the tools that were
                unavailable to you while still in 16-bit mode. This includes things like
                gradient removal and combining L and RGB images to create LRGB images. See
                the summary tutorial for creating LRGB images on the book web site's
                tutorial page.

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

                -----Original Message-----
                From: jfeldhou [mailto:jfeldhou@...]
                Sent: Friday, February 01, 2002 12:40 AM
                To: ccd-newastro@yahoogroups.com
                Subject: [ccd-newastro] Re: gradients & normalization


                this is good to know.
                more questions...

                do I need to switch to 8 bit to do the combine in photoshop?
                I was unable to get layers, nor "Apply image" to work in
                15/16 bit mode? is switching to 8 bits before doing advanced
                processing (ddp,unsharp,L-R) a bad idea?
                is switching to 8 bits a bad idea in general?

                thanks, jim


                --- In ccd-newastro@y..., "Ron Wodaski" <ronw@n...> wrote:
                > You are dealing with two very different reference systems here. I've
                never
                > had success this way, at least not without considerable mental
                gymnastics to
                > try to get the two files to have similar levels.
                >
                > The simplest solution is to apply the gradient removal to the image
                in
                > Photoshop. Photoshop just doesn't seem to understand the 16-bit
                format when
                > it comes to saving. I've heard that it actually only saves 15 bits,
                not 16,
                > and if this is true it would be just about impossible to use
                Photoshop to
                > create a gradient removal image in the way you describe.
                >
                > Since I had so much trouble with the approach, I only documented the
                > all-Photoshop methods in the book. I just can't recommend this
                approach
                > because Photoshop seems intent on making it not work.
                >
                > Ron Wodaski
                > The New CCD Astronomy
                > http://www.newastro.com
                >
                > -----Original Message-----
                > From: jfeldhou [mailto:jfeldhou@y...]
                > Sent: Thursday, January 31, 2002 5:05 PM
                > To: ccd-newastro@y...
                > Subject: [ccd-newastro] gradients & normalization
                >
                >
                > is it a bad idea to attempt to produce a gradient
                > from an image that has been normalized during stacking?
                >
                > I attempted to do the following procedure, with very
                > poor results. I had 11x600s luminence images,
                > preprocessed, stacked(sigmamean). M51 occupied most
                > of the corner quadrant, the rest stars of varying bloat.
                >
                > I saved the stacked fits image as 16 bit tiff, then
                > opened it in photoshop. I used the dust&scratches
                > tool at (12,16) cause some stars were big. Then I clone
                > tooled out any remaining stars, and the M51 portion.
                > after a gausian blur 19, I copied the 16 bit tif back
                > to my PC, changed it back to fits, and attempt to
                > subtract the blur from my original M51 stack.
                > I attempted to subtract the blur, after multiplying
                > the blurred pixels by .8 then .7 then .5 .4
                >
                > no matter my choice, the target image was erradicated.
                > my blur wasn't so good in the area of m51, but I had
                > hoped I'd see some worthwhile results on the rest of
                > the target frame, and then I could make a better blur.
                >
                > is there an obvious place where I went wrong?
                > thanks, jim
                >
                >
                >
                > To unsubscribe from this group, send an email to:
                > ccd-newastro-unsubscribe@e...
                >
                >
                >
                > Your use of Yahoo! Groups is subject to
                http://docs.yahoo.com/info/terms/



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



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