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

Re: [ccd-newastro] Re: Deconvolution quesiton

Expand Messages
  • Mark Acker
    There was once a tool released for Dell machines that was called the Dell Decrappifier to get rid of all the AOL, Symantec, etc., well crap. :) Mark
    Message 1 of 18 , Sep 10, 2013
    • 0 Attachment
      There was once a tool released for Dell machines that was called the "Dell Decrappifier" to get rid of all the AOL, Symantec, etc., well crap. :)

      Mark


      From: Stuart Forman <s24man@...>
      To: ccd-newastro@yahoogroups.com
      Sent: Tuesday, September 10, 2013 1:53 PM
      Subject: [ccd-newastro] Re: Deconvolution quesiton

       
      Thank you everybody.
      I'll amend my original post that my seeing isn't 1.14 arcsec/pixel. It's just my calculated CCD resolution. My seeing isn't nearly that good.

      I like the word "decrappifying". I'm going to steal it. :-)

      Stuart



    • Stuart Forman
      Stan wrote: So a normal FWHM is probably just under 3 for that scope. (right?) In that case, your avg sampling is near FWHM = 2.5 pix, which is
      Message 2 of 18 , Sep 10, 2013
      • 0 Attachment
        Stan wrote:

        "So a normal FWHM is probably just under 3" for that scope. (right?)
        In that case, your avg sampling is near FWHM = 2.5 pix, which is undersampled, though not dramatically so"

        You are correct. I'm lucky if I can get an FWHM under 3"--my seeing usually is not even quite that good. We have a marine layer that comes in many evenings that crappifies viewing. But it's the deck of cards I am dealt, so I play.

        Stuart
      • nismo.godzilla
        I always blend them together using Photoshop for that very reason. Place the deconvolved image on top with a hide all layer mask and paint through the details
        Message 3 of 18 , Sep 11, 2013
        • 0 Attachment
          I always blend them together using Photoshop for that very reason. Place the deconvolved image on top with a hide all layer mask and paint through the details that you want to add to the image etc... I also then apply a Gaussian blur to soften the effect of the layer mask. Finally you can use the new layer as an "Overlay" on the original image.

          Clear Skies


          Neil.
          www.insideastronomy.com

           



          --- In ccd-newastro@yahoogroups.com, <ccd-newastro@yahoogroups.com> wrote:

          I should amend my statement to say I've never liked *my* results. I assume many great images from others that I've admired have included deconvolution. What I've seen -- just my own efforts -- is that, if the settings are mild, then there's no discernable result, but if they're more agressive, then the results are all too obvious, mainly the dark halos. Since I use CCDStack, I'll certainly be interested in seeing methods relevant to that program.

          --- In ccd-newastro@yahoogroups.com, Neil Fleming <neilfleming@...> wrote:
          >
          > Yes, Curtis, I'll most probably have decon as a segment in my sessions at AIC this year...
          >
          > Yes, most decon tools out there use an unmodified or basic approach of Lucy-Richardson.  It's very easy to get decon artifacts, especially the dark halos around the stars with this algorithm.
          >
          > A few products incorporate a modification that allows for the clipping of the PSF raw wings to eliminate star halos, at least at low/moderate levels (CCDStack and Tria).  Interestingly, Tria also allows for "blind deconvolution", usually available in only very expensive products, which uses the image in concert with a reference image to generate the PSF.
          >
          > To the best of my understanding, neither PI nor MaxIM allow for clipping/bias subdivisions, so you'll need a star mask of sorts to mask out the effects on the stars if you end up using the decon at all.
          >
          >    Cheers...Neil
          >  
          > www.flemingastrophotography.com 
          > Direct from Boston - brilliant diamonds in pea soup
          > Also check out the astro_narrowbandYahoo group!
          >
          >
          >
          > >________________________________
          > > From: CurtisC <calypte@...>
          > >To: ccd-newastro@yahoogroups.com
          > >Sent: Tuesday, September 10, 2013 3:24 AM
          > >Subject: [ccd-newastro] Re: Deconvolution quesiton
          > >
          > >
          > >Maybe we'll see something about deconvolution at AIC :)  I've never liked the results.
          > >
          > >--- In ccd-newastro@yahoogroups.com, Neil Fleming <neilfleming@> wrote:
          > >>
          > >> True...
          > >>
          > >> That being said, if you get good, deep, data, decon can still do a wonderful job without going overboard.  My favorite tool by far for decon remains CCDStack.  I generally do two passes - a very gentle one for the stars, and one stronger one for the main object (like a galaxy).  You can then blend 'em together in either PI or Photoshop.  E.g., here is one that I'm still in the middle of collecting data for.  (1.31 asp):
          > >> http://www.flemingastrophotography.com/astropics/interim/M63_LRGB_100pct.jpg
          > >>
          > >>    ...Neil
          > >>  
          > >> www.flemingastrophotography.com 
          > >> Direct from Boston - brilliant diamonds in pea soup
          > >> Also check out the astro_narrowbandYahoo group!
          > >>
          > >>
          > >>
          > >> >________________________________
          > >> > From: Ron Wodaski <yahoo@>
          > >> >To: ccd-newastro@yahoogroups.com
          > >> >Sent: Monday, September 9, 2013 9:10 PM
          > >> >Subject: Re: [ccd-newastro] Deconvolution quesiton
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >In order to do proper convolution, you would need enough resolution to 'resolve' the convolution in the first place. So you'd want to be oversampled for seeing and optical quality.
          > >> >
          > >> >
          > >> >Ron Wodaski
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >On Sep 10, 2013, at 10:49 AM, Stuart Forman <s24man@> wrote:
          > >> >
          > >> > 
          > >> >>I was going to post this in the CCDStack forum, but since Rod talked about it in his book, and Stan reads this listserve, I thought it might be relevant here.
          > >> >>
          > >> >>My understanding of deconvolution techniques is that it works better on a well sampled or oversampled image.  I've read Rod's book, the Pixinsight website, Starizona, Wikipedia, and anything else the Googles will give me on it.  I have a TEC 140 at F7 imaging with an 8300 chip.  That puts me at 1.14 arcsec/pixel maximum resolution (with a seeing less than that since I live in the SF Bay Area where we have an ever present marine layer). So it appears that I'm generally undersampling the image, in which case deconvolution wouldn't be that helpful, and probably would just add artifacts.
          > >> >>
          > >> >>Am I understanding this correctly?
          > >> >>
          > >> >>Stuart
          > >> >>
          > >> >>
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >
          > >> >
          > >>
          > >
          > >
          > >
          > >
          > >------------------------------------
          > >
          > >
          > >
          > >
          > >
          > >
          >
        Your message has been successfully submitted and would be delivered to recipients shortly.