Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Please Help: Fixing Star Artifacts on Liverpool Telescope Images

  • Please log in to reply
21 replies to this topic

#1 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 12:24 PM

I have known about the Liverpool Telescope Archives for several years. There is an incredible amount of data to access and process in an perfect image scale. While people enjoy doing projects renting telescope time in Chile for the 1 Meter Telescope at a handsome price there is a lot of data up there that is free in the larger 2 meter Liverpool System.

 

You do have to comb through the data to see what is available and sort out the better images.

 

The problem I am asking for help here is dealing with the artifacts in the stars of the raw data images. How can someone fix these stars? Obviously I have not figured this out but there is a way to do it. I was hoping somebody here could show me the way.

 

Any help would be greatly appreciated.

 

Liverpool.jpg

 


  • dswtan likes this

#2 Dynan

Dynan

    Fly Me to the Moon

  • *****
  • Posts: 6,014
  • Joined: 11 Mar 2018
  • Loc: NOLA

Posted 15 May 2021 - 12:38 PM

PI Clone Stamp might work.


  • Hobby Astronomer likes this

#3 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 12:46 PM

Here is where someone can pick up the link to a FITS file to try a solution on this:

 

See: LINK TO THE FITS FILE on the page below.

 

https://telescope.li..._106_1_1_1.html



#4 dswtan

dswtan

    Viking 1

  • *****
  • Posts: 695
  • Joined: 29 Oct 2006
  • Loc: Morgan Hill, CA

Posted 15 May 2021 - 01:22 PM

I'm really interested in the Liverpool archive too, and appreciate the OP again recently drawing attention to it -- thank you.

 

My inclination is that some PixelMath wizardry could address or mitigate this, but I don't personally have that level of skill.

 

There's an interesting ancient thread on the PI forum that may provide some hints and techniques with masked painting (and PI's, um, perspective on that).

https://pixinsight.c...-star-fix.3557/


  • Hobby Astronomer and Dynan like this

#5 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 01:39 PM

Here is a 4 panel mosaic that I stitched together from the Liverpool Telescope Archives. The stars circled have the artifact problem. See next post for example of the stars being fixed. I just do not know how to do it.

 

M16_4Panel_HA_Mosaic_60Sec_V.1SM.jpg


Edited by Hobby Astronomer, 15 May 2021 - 01:44 PM.


#6 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 01:42 PM

This is a lum channel extraction from Goran Nilsson's Liverpool Telescope Wilkipedia Image here: https://en.wikipedia...l_Telescope.jpg

 

Somehow he fixed the stars in his image.

 

_1024px_The_Eagle_Nebula_M16_Goran_Nilsson__The_Liverpool_Telescope_L.jpg


Edited by Hobby Astronomer, 15 May 2021 - 01:44 PM.


#7 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 01:56 PM

h_e_20200220_53_1_1_1.jpg

 

Above is a single sub from the Liverpool Telescope Archives: 120 Sec  Bessell-B  2020-02-20 23:03:29  NGC2261

 

Below is the the image by Daniel Nobre where where the star artifacts are fixed.

 

get.jpg?insecure



#8 rekokich

rekokich

    Apollo

  • *****
  • Posts: 1,130
  • Joined: 14 Sep 2014
  • Loc: Virginia

Posted 15 May 2021 - 09:18 PM

StarTools
open Spots 1
mask module, brush mode Lasso, lasso a mask around the spots but within the star image as in Spots 2, keep,
repair module, algorithm Redistribute, Core is Avg Location, do,
final result is Spots 3

 

SPOTS.jpg


  • dswtan, Hobby Astronomer and Dynan like this

#9 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 09:30 PM

StarTools
open Spots 1
mask module, brush mode Lasso, lasso a mask around the spots but within the star image as in Spots 2, keep,
repair module, algorithm Redistribute, Core is Avg Location, do,
final result is Spots 3

 

attachicon.gifSPOTS.jpg

Wooooooowwww. waytogo.gif

 

bow.gifbow.gifbow.gifbow.gifbow.gifbow.gifbow.gif


Edited by Hobby Astronomer, 15 May 2021 - 09:32 PM.


#10 MikeK314

MikeK314

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 09 Oct 2013

Posted 15 May 2021 - 09:57 PM

If you have Photoshop, lasso around the spot but stay within the star, then run the Maximum filter (under Other in the Filters menu), setting the size to be large enough to capture the whole spot.  That'll do it.  I would guess that GIMP has this functionality too, but I don't use GIMP so can't say for sure.


  • dswtan and Hobby Astronomer like this

#11 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 10:12 PM

If you have Photoshop, lasso around the spot but stay within the star, then run the Maximum filter (under Other in the Filters menu), setting the size to be large enough to capture the whole spot.  That'll do it.  I would guess that GIMP has this functionality too, but I don't use GIMP so can't say for sure.

Thanks Mike! waytogo.gif

 

bow.gifbow.gifbow.gifbow.gifbow.gifbow.gifbow.gif



#12 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 10:19 PM

Thanks Rekokich and Mike. Thanks Dynan.

 

I tried clonestamp in PI. Sort of worked but was not very elegant.

 

I am going to try out star tools to see what it is like and try the process there.

 

Also going to try Mikes approach with Affinity Photo which is a Photoshop clone.

 

Wonder if there is a way to do this in Pixinsight besides the clone stamp method?



#13 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 15 May 2021 - 10:22 PM

Here is the clonestamp method used on the HA channel. Sort of works but would really be trouble in a multi-channel image such as an LRGB image.

 

M16_4Panel_HA_Mosaic_60Sec_V.1Clonestamp.jpg



#14 pejorde

pejorde

    Explorer 1

  • *****
  • Posts: 70
  • Joined: 27 Nov 2010
  • Loc: Norway

Posted 17 May 2021 - 03:18 PM

These spots have negative pixel values in the fits file and the black color is simply the software's way of presenting such values. The simplest approach, especially if you will be processing many such images,  would be to write a small program or script that replaces negative pixel values with the max value, which will show up as white:

 

h_e_20200629_106_1_1_1.jpg

 

Per Erik


  • lambermo, pfile, Hobby Astronomer and 2 others like this

#15 astroian

astroian

    Ranger 4

  • -----
  • Posts: 380
  • Joined: 21 Sep 2016
  • Loc: Swindon, UK

Posted 18 May 2021 - 01:34 PM

Simplest solution in PI I can think of is to create a mask that protects everything except the centres of those stars with the problem (plus a small border). Apply the mask then use pixel math to write 1 to the image. Only the unprotected parts will be set to 1. If you want to blend it a bit then use the convolution process to blur the mask a bit.

Cheers,
Ian
  • Hobby Astronomer likes this

#16 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 5,734
  • Joined: 14 Jun 2009

Posted 19 May 2021 - 10:57 AM

similarly you can just go ahead and register/integrate and then use the readout cursor to see what the values are like in those black parts in the stars. in the sample fits above, the following expression filled in the stars properly:

 

iif($T<0.072,1,$T)

 

then you can check to make sure you didn't blow away any real data by applying the pixelmath expression:

$T==1

 

and you'll get a "map" of what pixels were replaced.

 

you'll have to play with the value 0.072 to find the right one but once you find it, you're done. you can also choose a value < 1 to replace so that you're matching the saturated value in the rest of the star.

 

rob


  • Hobby Astronomer likes this

#17 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 19 May 2021 - 09:23 PM

similarly you can just go ahead and register/integrate and then use the readout cursor to see what the values are like in those black parts in the stars. in the sample fits above, the following expression filled in the stars properly:

 

iif($T<0.072,1,$T)

 

then you can check to make sure you didn't blow away any real data by applying the pixelmath expression:

$T==1

 

and you'll get a "map" of what pixels were replaced.

 

you'll have to play with the value 0.072 to find the right one but once you find it, you're done. you can also choose a value < 1 to replace so that you're matching the saturated value in the rest of the star.

 

rob

 

Well here is 12 subs of 30 seconds in the red channel stacked. Method works.

integration.jpg


  • dswtan and pfile like this

#18 ChrisWhite

ChrisWhite

    Fly Me to the Moon

  • *****
  • Posts: 6,863
  • Joined: 28 Feb 2015
  • Loc: Northeast USA

Posted 20 May 2021 - 07:59 AM

I think the easiest way for you to eliminate these is by raising your high sigma clipping value in image integration.  Keep raising it just to the point where they disappear and you still get the rejection benefits where you need them.  A little trial and error but I'm almost certain this will fix your problem. 

 

EDIT-  When you say RAW data, are you talking about individual subs?  Or integrations?


Edited by ChrisWhite, 20 May 2021 - 08:01 AM.


#19 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 5,734
  • Joined: 14 Jun 2009

Posted 20 May 2021 - 02:22 PM

this is in the subs... they are FITS files with extent -32627 to +32627 which is a little weird. i played around with the FITS file hander on PI to try to get it to interpret the negative values differently but no matter what i did there were 0 (actually very small values) in the star cores. PI might be translating -1000 to +1000 for instance rather than scaling such that -1000 == 0.

 

so bottom line is rejection may not work since all the values in the star cores are similar and so are not outliers.

 

rob


  • dswtan likes this

#20 ChrisWhite

ChrisWhite

    Fly Me to the Moon

  • *****
  • Posts: 6,863
  • Joined: 28 Feb 2015
  • Loc: Northeast USA

Posted 20 May 2021 - 05:16 PM

this is in the subs... they are FITS files with extent -32627 to +32627 which is a little weird. i played around with the FITS file hander on PI to try to get it to interpret the negative values differently but no matter what i did there were 0 (actually very small values) in the star cores. PI might be translating -1000 to +1000 for instance rather than scaling such that -1000 == 0.

so bottom line is rejection may not work since all the values in the star cores are similar and so are not outliers.

rob


Thanks Rob. I initially thought this was an integration artifact. Strange indeed.

#21 Hobby Astronomer

Hobby Astronomer

    Surveyor 1

  • *****
  • topic starter
  • Posts: 1,956
  • Joined: 18 Dec 2013
  • Loc: Texas

Posted 20 May 2021 - 09:13 PM

post-225131-0-71700800-1621477351_thumb.

 

I applied the Pixel Math only to the final image. If this was a serious image I would have applied the Pixel Math to all the subs.

 

Happy so far.

 

HA


  • dswtan likes this

#22 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 5,734
  • Joined: 14 Jun 2009

Posted 21 May 2021 - 05:40 PM

Thanks Rob. I initially thought this was an integration artifact. Strange indeed.

yes - there is a weird artifact which i think mainly shows up with drizzle where you get black star cores. i think this happens because all the data in every sub (where there is a saturated star) is rejected and so PI doesn't know any other value to put there but 0.

 

this had been happening for a long time and no one could really explain it but then someone uploaded a dataset that causes the problem and Juan figured it out:

 

https://pixinsight.c...4725/post-98095

 

this problem looks a lot like that one but amazingly it is different. something new every day in this hobby...

 

rob


Edited by pfile, 21 May 2021 - 05:44 PM.

  • dswtan and ChrisWhite like this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics