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

Strange Intermittent Artifact (PI Suggestions?)

  • Please log in to reply
20 replies to this topic

#1 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 01:24 PM

I am getting a Strange Intermittent Artifact on some of my M42 subs, and I am soliciting PI processing work arounds to salvage otherwise goods subs, probably using PixelMath.  This artifact is just in a small portion of the frame.  My problem is that M42 is only visible for at most 3 hours (w/ tall trees to East and my house to the West), and this time of the year it is seldom clear.  I have not had a clear night in a month and the next week is bad.  My main problem is my Red filter as only 36/70 subs are good, yielding only 72mins and most of the dust is in Red.  My Blue and Green both have 70+ good subs, so I can use only the non-aberrated subs.

 

The OTA is an A-P 110GTX w/ 110FF and the camera is a QHY600M.  These artifacts are on some calibrated subs, but not on others.  All subs are using the same flats/darks.  The flats were captured on earlier sessions, but the OTA has remained, undisturbed on the mount on the interim.  The problem does not seem to be the flats, as in subsequent nights the problems is there for ~ an hour and then everything is fine for the balance of the evening.  Then the problem reoccurs the next night in a similar manner.  The artifact seems like some light has been moved to the left, leaving a shadow to the right.  The artifact is on all filters within the aberrant timeframes.  It seems like something on the sensor glass (more than a dust spot and elongated) that appears and then disappears.

 

Here is a starless RGB combine, with the artifact being above the Running Man, showing a ~vertical dark gouge with extra light to the left.   

 

M42_110GTX_RGB_Starless (Artifact).jpg

 

I have masters with and without the aberrant subs included.  I was thinking about masking an ellipse (GAME) around the problem area, using starless inputs, and adjusting the large set's (w/ aberration) master's masked area using the smaller set's $T/med($T) of the same masked area.

 

If I had only the larger master (w/ aberration), I have in the past used the following PixelMath to dilute the problems of a moving spot:

 

iif($T<med($T)-x,max($T,avg($T,med($T)+y)),iif($T>med($T)+x,min($T,avg($T,med($T)-y)),$T))

 

with x and y being PM symbols that have small input values.  You can assume x=0 and y=0 as a 1st approximation.

 

Any ideas on the PM solution or cause?


Edited by SilverLitz, 21 January 2025 - 01:25 PM.


#2 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 6,565
  • Joined: 14 Jun 2009

Posted 21 January 2025 - 01:45 PM

which of the 3 artifacts are you talking about? your posted image is too dim, had to put into PI and boost with STF...

 

Screenshot 2025-01-21 at 10.42.25 AM Large.jpeg

 

if it's the one in the middle, that is really quite strange as it's the same shape repeated light and dark. the uncalibrated subs really look like this?

 

rob

 



#3 Poynting

Poynting

    Viking 1

  • *****
  • Posts: 724
  • Joined: 01 Apr 2012
  • Loc: La Vernia, TX

Posted 21 January 2025 - 01:49 PM

Condensation or frost appearing and disappearing once the dew heater catches up? 



#4 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 01:54 PM

which of the 3 artifacts are you talking about? your posted image is too dim, had to put into PI and boost with STF...

 

attachicon.gif Screenshot 2025-01-21 at 10.42.25 AM Large.jpeg

 

if it's the one in the middle, that is really quite strange as it's the same shape repeated light and dark. the uncalibrated subs really look like this?

 

rob

Rob,

 

It is the strange middle one.  This is also not a stacking artifact, as its basic shape is in all the aberrant subs.

 

The horizontal lines seem to come w/ stacking, as it is not obviously visible in the individual subs.



#5 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 02:09 PM

The middle aberration is in the uncalibrated subs, with varying intensity.



#6 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 6,565
  • Joined: 14 Jun 2009

Posted 21 January 2025 - 02:23 PM

what's very strange about this is the brighter artifact looks like it could belong in the nebula. and then the darker one looks like it's shadow. i don't have a clue how to fix this with pixelmath, i think this is something for clonestamp or a healing brush or something but it's so large that it's not easily erased. but something really weird is going on here.

 

i had a reflection problem with the 110TCC installed but it turned out that i had the filters AR side toward the sensor. the reflection only occurred in OIII and flipping that filter fixed the problem. the reflections were of course stars, and the position changed depending on what side of the meridian i was on. hard to see how you'd get a reflection of a dim extended object though.

 

rob



#7 D_talley

D_talley

    Fly Me to the Moon

  • *****
  • Posts: 5,425
  • Joined: 07 Jul 2005
  • Loc: Albuquerque, New Mexico

Posted 21 January 2025 - 02:23 PM

As for the artifacts on the left and right of the image, I would see them in the older CCD sensors which would be caused by bright stars.  This is the first time I have seen this in a CMOS camera. 



#8 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 02:42 PM

An aside, SetiAstro's Satellite Trail removal in his CosmicClarity Suite is VERY beneficial for M42, as several satellite trails are in almost every sub (most to the right of M42 in ~vertical lines) and PI was not completely rejecting them.  I batched this outside PI, running it on my calibrated subs, and then stacked these satellite removed subs.


Edited by SilverLitz, 21 January 2025 - 04:01 PM.

  • pfile likes this

#9 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 03:23 PM

The light aberration (to the left of the shadow) is not part of the nebula.  It has the same shape as the shadow, but the inverse (light not shadow).

 

Here is the Red channel without the aberrated subs:

 

M42_110GTX_120s_R_starless.jpg



#10 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 03:28 PM

Here is the Red channel with the aberrated subs, notice both the light and shadow "figure 8" aberrations:

 

M42_110GTX_120s_R2_starless (Artifact).jpg

 

Using all 70 subs, result is much lower noise.

 

Also, the dark horizontal aberrations do seem to line up with the bright (and not removed by SXT) star, as suggested.



#11 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 6,565
  • Joined: 14 Jun 2009

Posted 21 January 2025 - 03:54 PM

i wonder if you should take some new R flats this evening and see if there's something there. is there a time-based pattern with your aberrated subs vs. non aberrated? meaning does the artifact only show up on calibrated images taken more recently?

 

blinking those two images i do see another artifact near the bottom of M42, which seems to be diagonally opposed to your obvious artifact. so it does seem to move when there is a meridian flip.

 

rob



#12 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 21 January 2025 - 04:23 PM

Can I get PixelMath to generate ratios greater than 1? (w/ Rescale Result unchecked, the results are still <= 1)  I want $T/med($T) to give an output that can exceed 1.

 

And, can PixelMath give a median of only the masked part of the image?



#13 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 6,565
  • Joined: 14 Jun 2009

Posted 21 January 2025 - 07:09 PM

PI normalizes everything to the range 0.0-1.0, so you can't get values > 1.0. with rescale result turned off, pixels brighter than 1.0 clamp to 1.0. with rescale result on, it does the math, looks for the brightest pixel, and scales the result to be in the range 0.0-1.0. so i guess you'd have to scale your input image and multiply by the factor you want with rescale result off, making sure that you don't clip any output pixels.

 

masks in PI are pretty much write-only. they are applied to the image as the image is being written by whatever process you are running. i think to do what you want you'd have to mask the image with a binary mask, then apply the pixelmath '0' to the image, thus zeroing out the unmasked areas. then run the statistics process and by default it should discard all pixels of value 0 and 1, so effectively you'll get the median value for the masked part.

 

an exception i think is StarXterminator, which looks at the mask when reading the image to create the star image. but that's unusual behavior for PI.

 

rob


  • SilverLitz likes this

#14 drmikevt

drmikevt

    Surveyor 1

  • *****
  • Posts: 1,561
  • Joined: 22 Jun 2015
  • Loc: Burlington, VT

Posted 22 January 2025 - 07:54 AM

masks in PI are pretty much write-only. they are applied to the image as the image is being written by whatever process you are running. i think to do what you want you'd have to mask the image with a binary mask, then apply the pixelmath '0' to the image, thus zeroing out the unmasked areas. then run the statistics process and by default it should discard all pixels of value 0 and 1, so effectively you'll get the median value for the masked part.

Or maybe make a duplicate image and crop down to the area you want to get the median of


  • pfile likes this

#15 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 22 January 2025 - 11:01 AM

Here is a side by side comparison of a small preview around the aberration with the smaller (w/o aberrant subs) combine on the left vs. larger (aberrated) combine on the right.  The area of the aberration looks like a lower leg (below knee w/ foot).  The area of the "foot" is not affected, as the aberration goes from the knee to the ankle.

 

Red Starless Spot Compare.jpg

 

The left (w/o aberrant subs) has a higher median of 3.20844e-03, compared to the right (aberrated) at 2.980339e-03.  This median is of this preview only, but includes the local area surrounding the aberration.

 

I am thinking about substituting the unaberrated left multiplied by this median ratio (2.980339 / 3.20844) into the aberrated right.  This substitution will be only in an even smaller area being an ellipse more tightly bounding the aberration.

 



#16 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 22 January 2025 - 11:23 AM

Further supporting my assertion that the problem is not the flats, is that earlier in the evening of all these nights, I shot another target with this setup (Lobsterclaw+Bubble) without any aberrations.  This would also question the possibility of issue being the dew heater, as it would be fine for ~4 hours (on Lobsterclaw+Bubble), then bad for ~1 hour (M42), then good for ~2 hours (M42), with this behavior also repeating the following night.

 

That would lead me to believe the cause must be some sort of reflection that is very sensitive to where the OTA is pointing (alt-az).  However, the 1st night that I imaged M42 (2 weeks earlier) did not have this issue.

 

The differences between the shadow (0.0029) and the brighter reflection (0.0030) is small, but exaggerated by the STF.


Edited by SilverLitz, 22 January 2025 - 11:24 AM.


#17 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 23 January 2025 - 01:05 PM

The fixed worked very well.  The mask for the replacement needed to be an elliptical GradientEdge mask with extra blurring.

 

Here is a side by side comparison of a small preview around the aberration with the smaller (w/o aberrant subs) combine on the left vs. larger (aberrated, after correction) combine on the right.  The replaced ellipse is ~40% of this preview.  So, I get the benefit of much better SnR of the larger (but aberrated) data set of of the vast majority of the frame, while the replaced ellipse has been corrected, but with more noise in the corrected area.

 

Red Starless Spot Compare (Corrected).jpg

 

Here is the very simple PM correction:

 

M42_110GTX_120s_R_starless * beta  

(with beta=0.9289)

 

 



#18 pfile

pfile

    Fly Me to the Moon

  • -----
  • Posts: 6,565
  • Joined: 14 Jun 2009

Posted 23 January 2025 - 01:08 PM

how exactly did you make the mask? using GAME?

 

thanks

 

rob



#19 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 23 January 2025 - 01:36 PM

how exactly did you make the mask? using GAME?

 

thanks

 

rob

I used GAME (GradientEdge) plus several iterations of MaskBlur script (Bill Blanshan???)


  • pfile likes this

#20 drmikevt

drmikevt

    Surveyor 1

  • *****
  • Posts: 1,561
  • Joined: 22 Jun 2015
  • Loc: Burlington, VT

Posted 23 January 2025 - 04:26 PM

I used GAME (GradientEdge) plus several iterations of MaskBlur script (Bill Blanshan???)

Is there a link for that script?  I've never heard of it or seen it.  Can't find it with a quick google search.



#21 SilverLitz

SilverLitz

    Gemini

  • *****
  • topic starter
  • Posts: 3,422
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 23 January 2025 - 04:48 PM

This site has links to getting Blanshan's masking scripts (scroll down):

 

https://cosgrovescos...uperpower-in-pi

 

If not, you can use Convolution




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