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

Rainbow stars

  • Please log in to reply
24 replies to this topic

#1 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 14 September 2019 - 04:09 PM

I'm seeing odd multi-colored effects in my stars and can't quite figure out what's going on.  Here's a picture of a particularly bad region in M29, both an individual sub and a lightly-processed stack of about 1000 taken over about a week, same stars, zoomed quite a ways in.  For both images I did background neutralization and an ArcsinhStretch.

 

RainbowStars.jpg

 

Here are some (I think) relevant facts:

  • You can see that the color distortions are visible even in one sub.
  • I'm using a new camera, a ZWO ASI294MC one-shot-color, with offset = 5 and gain = 120.  This is the third image from this camera I've examined.  The problem was also visible in the other two.  I don't see it in the DSLR images I made with the same system before this camera, but that was a year ago too.
  • The color distortions vary at different places in the image.  Sometimes they're not even evident.  In this sample it looks like blue is coming in at about 4:00 and magenta about 10:00 on all three big stars, but that isn't generally the case across the image.
  • The humidity during the image acquisition was pretty high, upwards of 90%.
  • The noise background in the individual sub is fairly even, and doesn't indicate blotches of color corresponding to the color distortions in the stars.  You can see this better if you zoom in even further.
  • My scope is an 8" Ritchey–Chrétien, and I'm using an Astro Physics CCDT67 compressor at about 0.75 compression.  Dew heaters are keeping the primary & secondary mirrors clear (although I must admit I'm guessing about the secondary).  That's about it for the optical chain.
  • My mirrors are not spotless, but they don't look filmy to the naked eye.
  • All subs were shot at altitude >=50 degrees, so I don't think atmospheric diffraction is happening.

Any ideas?


Edited by TinySpeck, 14 September 2019 - 05:54 PM.


#2 Brett Waller

Brett Waller

    Mariner 2

  • *****
  • Posts: 263
  • Joined: 18 Aug 2006
  • Loc: Staunton, VA

Posted 14 September 2019 - 04:28 PM

Is the pattern of coloration radially symmetric with respect to the center of the image?  In other words, are blue and red always on the same sides toward and away from the center?  If so, use Channel Extraction to separate the RGB channels, then Star Alignment to register them, and Channel Combination to recombine them into the RGB image. 

 

I am assuming you are using a OSC camera and the red and blue points of focus are at slightly different points, thus the plate scale varies with wavelength. Star Alignment will re-register the RGB panes and should improve the result.

 

Brett



#3 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 14 September 2019 - 05:09 PM

Interesting idea, Brett, thanks.  I checked again, carefully, and unfortunately there isn't a pattern like what you describe.  I don't see a pattern at all.  Some central stars are bad and some good, and same with the outer regions.

 

Here is a particularly wacky one I found while looking around:

 

Rainbow2.jpg

 

What th'...??



#4 Gipht

Gipht

    Surveyor 1

  • *****
  • Posts: 1888
  • Joined: 12 Nov 2016
  • Loc: Prescott Valley, AZ.

Posted 14 September 2019 - 05:37 PM

Just as a check, you might try a picture without the focal reducer.  There may be a spacing issue between the camera and the reducer or a compatibility issue between the optics and the reducer.


  • TinySpeck likes this

#5 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 14 September 2019 - 06:01 PM

Just as a check, you might try a picture without the focal reducer.  There may be a spacing issue between the camera and the reducer or a compatibility issue between the optics and the reducer.

I just edited my original post to mention that I didn't see the problem with the same scope & reducer using a DSLR last year.  Also, if it was a lens effect I wouldn't expect the effect to be so random across the image.

 

The effect doesn't change over a week or two of imaging and I have dew heaters, so it's probably not dew.  I'm leaning toward some kind of film on my mirrors now.  The RC is an open design so the neighbor's BBQ smoke, exhaust, etc, might build up.  That could be an uneven effect, too.  It may be that the film has built up since last year when I got undistorted color images from my DSLR.

 

Does this make sense?



#6 niccoc1603

niccoc1603

    Ranger 4

  • -----
  • Posts: 318
  • Joined: 16 Aug 2018
  • Loc: Firenze, Italy

Posted 14 September 2019 - 07:41 PM

Can’t really help with the diagnose of the problem, but in my experience ArcsinStretch has the tendency to create various weird color artefacts in stars.
I like using AS, but I always replace the stars core with a HT stretched version. The operation is simple with Pixelmath and a proper star mask

Edited by niccoc1603, 14 September 2019 - 07:41 PM.


#7 hoxca

hoxca

    Explorer 1

  • -----
  • Posts: 93
  • Joined: 02 Mar 2016

Posted 16 September 2019 - 06:55 PM

Hum...

I also notice some of this artifact with my ASI071 and FSQ106.

It should be also correlated for sure with the tiny pixel size of the sensor / and my overall sampling of 1.854 arcsercs/pixel.

I think that my 6d was more forgiving in this regard.

 

A debayer drizzle doest help to fix this behavior.

I think that i did/do a mistake when i register my ligth frames.

 

So i did a quick test with a manual registration just to see the diff.

For the record here a result on a small sample of my SH2-120 frames.

 

initial debayer drizzle :

 

before.jpg

 

manual register of the V,B channels to R after integration :

 

after.jpg

 

Correcting the problem afterward is not the way to go as the process add a bunch of chromatic noise to the image frown.gif

I'm searching the root cause of this in a pre-processing workflow.

 

Maybe a registration problem ?



#8 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 16 September 2019 - 07:01 PM

I had another suggestion that it might be due to stacking registration error, too, but I'm seeing what looks like the same color distortion even in one sub (see original post).  I might try drizzle debayer though, just to see if it helps.



#9 t-ara-fan

t-ara-fan

    Viking 1

  • -----
  • Posts: 733
  • Joined: 20 Sep 2017
  • Loc: 50° 13' N

Posted 17 September 2019 - 01:05 AM

I had stars with uneven colors. They went away after I collimated my EdgeHD-8.  Collimated it a few times ;)

 

suZ0OsI.jpg



#10 Pete_xl

Pete_xl

    Viking 1

  • -----
  • Posts: 740
  • Joined: 29 Jan 2015
  • Loc: Germany

Posted 17 September 2019 - 04:00 AM

I also have had this effect with my ASI294mcp connected to the TS Photoline 130/910 mm triplet apo. The target was quite low above the horizon. The rainbow stars were  spread all over the fov and separate registration of channels did not help. Some weeks ago I had imaged M32 higher in the skies and all was ok.

 

I thought maybe this was a combo-effect of tiny pixels and atmospheric dispersion. I do not have enough imaging sessions with the 294 to find out how reproducable the effect is because I normally use my ASI1600mmp.



#11 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 17 September 2019 - 10:19 AM

Thanks for the helpful suggestions, folks!

 

I had stars with uneven colors. They went away after I collimated my EdgeHD-8.  Collimated it a few times

Very interesting.  I haven't collimated since getting the scope almost exactly a year ago.  Star shape is good across the field and I was hoping to avoid collimating for a while.  I didn't know it could affect color!

 

... The target was quite low above the horizon. The rainbow stars were  spread all over the fov and separate registration of channels did not help. Some weeks ago I had imaged M32 higher in the skies and all was ok....

I too was seeing atmospheric refraction effects with low targets some time ago, so now I'm keeping the altitude over 50 degrees.  This includes the targets which have shown these rainbow stars, so I know that's not the problem this time.

 

Last night I zoomed in on one particularly chaotic star in both the single sub (above) and the stack (below):

 

rainbow 3.jpg

 

I zoomed until the pixel sizes were about the same between the two.  I think I can see the beginnings of the same color distortions in the sub as in the stack.  Just further support that the problem seems to be at the individual sub level.



#12 Tayson82

Tayson82

    Explorer 1

  • -----
  • Posts: 61
  • Joined: 17 May 2018
  • Loc: Poland / Wolomin

Posted 18 September 2019 - 05:32 AM

manual register of the V,B channels to R after integration :

How did You do that?



#13 ManuelJ

ManuelJ

    Apollo

  • -----
  • Posts: 1450
  • Joined: 19 Dec 2005
  • Loc: Madrid, Spain

Posted 19 September 2019 - 11:13 AM

I think that your reducer is acting like a prism, producing wedge color. Try rotating it (and only the reducer) and see if it follows the pattern.

#14 ManuelJ

ManuelJ

    Apollo

  • -----
  • Posts: 1450
  • Joined: 19 Dec 2005
  • Loc: Madrid, Spain

Posted 19 September 2019 - 11:16 AM

Hum...
I also notice some of this artifact with my ASI071 and FSQ106.
It should be also correlated for sure with the tiny pixel size of the sensor / and my overall sampling of 1.854 arcsercs/pixel.
I think that my 6d was more forgiving in this regard.

A debayer drizzle doest help to fix this behavior.
I think that i did/do a mistake when i register my ligth frames.

So i did a quick test with a manual registration just to see the diff.
For the record here a result on a small sample of my SH2-120 frames.

initial debayer drizzle :

before.jpg

manual register of the V,B channels to R after integration :

after.jpg

Correcting the problem afterward is not the way to go as the process add a bunch of chromatic noise to the image frown.gif
I'm searching the root cause of this in a pre-processing workflow.

Maybe a registration problem ?


Do your stars look like that with a high power eyepiece? If yes, your telescope is out of collimation.

#15 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 19 September 2019 - 12:36 PM

I think that your reducer is acting like a prism, producing wedge color. Try rotating it (and only the reducer) and see if it follows the pattern.

 

Do your stars look like that with a high power eyepiece? If yes, your telescope is out of collimation.

Thanks for your thoughts, Manuel.  My star color distortion doesn't seem to have any regularity to it across the image, and the color distortion on some stars looks completely random (not prism-like, see above).  I've been looking closely at some older photos, though, and I think I can see the same color distortions so maybe the reducer has been distorting the color all along.  I will see if I can isolate that.

 

I've seen a couple suggestions here that collimation could be the problem, too.  That's a good idea about checking with an eyepiece, although I'm not sure I could see the same effect as a 60-s exposure (which is already pretty subtle).  I didn't know Ritchey–Chrétien collimation error could result in color distortion, but it's a good lead to follow.


Edited by TinySpeck, 19 September 2019 - 12:43 PM.


#16 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 19 September 2019 - 08:08 PM

UPDATE: I've just gone through a dozen or so images over the past year, and there is color distortion like this in most of them.  A few of them show almost no distortion, but most of the rest of them do to some extent.

 

So I think I have two problems: I'm getting more critical, and something in my optical chain isn't quite perfect.  I really appreciate the suggestions here.  I will work on collimation and close examination of my reducer.



#17 hoxca

hoxca

    Explorer 1

  • -----
  • Posts: 93
  • Joined: 02 Mar 2016

Posted 20 September 2019 - 06:01 PM

How did You do that?

Split RGB channel then there is a process in pixinsight call dynamic alignment that just do that wink.gif

 

dynamic_alignment.jpg


Edited by hoxca, 20 September 2019 - 06:03 PM.


#18 hoxca

hoxca

    Explorer 1

  • -----
  • Posts: 93
  • Joined: 02 Mar 2016

Posted 20 September 2019 - 06:08 PM

In my case using the distortion correction with 50 iterations in the register process solve the behavior.

This is not burning a lot of CPU time. For my lights, the processus converge before 3 iterations.

 

definitly another pebkac !



#19 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 21 September 2019 - 11:57 AM

Split RGB channel then there is a process in pixinsight call dynamic alignment that just do that wink.gif

 

attachicon.gif dynamic_alignment.jpg

I checked this out.  DynamicAlignment is a tedious manual process of selecting stars, and only works with a pair of images (two of the three RGB channels).  It must be repeated, including star selection as far as I can tell, for the other RGB channel.  According to Juan Conejero at PixInsight the StarAlignment process uses the same 2D surface-fit algorithm, but it finds stars automatically (per your parameters) and aligns all three channels at once.  Have a look at that instead, it should do the same thing but easier and better.


Edited by TinySpeck, 21 September 2019 - 12:23 PM.


#20 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 21 September 2019 - 12:20 PM

... Star Alignment will re-register the RGB panes and should improve the result.

 

Brett

When I tried this earlier I didn't have the StarAlignment parameters set correctly for my stretched image, so it wasn't working as well as it should.  I'm experimenting further and getting better results, so this is definitely a great suggestion, Brett, thank you.  By careful blinking between the split R, G, and B channels set to the same zoom and center, I can definitely see that they are not all aligned.

 

I've been examining my R, G, B split files and seeing distortion in only the R and B channels too.  G stars look very nice and uniform throughout the image.  Here is a comparison of the particularly wacky star I've been picking on above.  G is on top, B on the bottom:

 

G and B.jpg

 

See the bite chomped out of the upper left quadrant of B?  I'm seeing asymmetries and defects like this all over the image, in both R and B but not G.  So this makes me think it has something to do with debayering.  Since you can see the individual pixels in these images (each pixel corresponds to one of the four Bayer pixels), it seems odd that the distortion would extend over so many complete RGGB quads, but I'm not sure how else to explain how nice G looks and how lousy B and R look.  To be continued...



#21 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 21 September 2019 - 01:13 PM

I think that your reducer is acting like a prism, producing wedge color. Try rotating it (and only the reducer) and see if it follows the pattern.

It IS my reducer!  Almost all of my images have used it, but I dug up some old test shots I did with no reducer and there is no color distortion whatsoever in the individual RGB subs.  Star color is very pure over the entire image.  The split R / G / B channels line up right on top of one another and there is no difference in distortion or asymmetry in the R, G, or B channels either.

 

So leave it to a lens to screw up a pristine achromatic mirrored telescope!  Now, what to do to repair all my distorted reduced images?  There may be some un-lenser I can use on each RGB channel -- this will take some research.


  • ManuelJ likes this

#22 ManuelJ

ManuelJ

    Apollo

  • -----
  • Posts: 1450
  • Joined: 19 Dec 2005
  • Loc: Madrid, Spain

Posted 22 September 2019 - 07:54 AM

It IS my reducer! Almost all of my images have used it, but I dug up some old test shots I did with no reducer and there is no color distortion whatsoever in the individual RGB subs. Star color is very pure over the entire image. The split R / G / B channels line up right on top of one another and there is no difference in distortion or asymmetry in the R, G, or B channels either.

So leave it to a lens to screw up a pristine achromatic mirrored telescope! Now, what to do to repair all my distorted reduced images? There may be some un-lenser I can use on each RGB channel -- this will take some research.


Having suffered something similar, I tried to fix it up via software, but ended up deleting data and having to capture it again.

Results are mediocre. At least you found the cause, good for you!
  • TinySpeck likes this

#23 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 22 September 2019 - 12:22 PM

Just as a check, you might try a picture without the focal reducer.  There may be a spacing issue between the camera and the reducer or a compatibility issue between the optics and the reducer.

 

Having suffered something similar, I tried to fix it up via software, but ended up deleting data and having to capture it again.

Results are mediocre. At least you found the cause, good for you!

Thanks for the suggestions about the reducer, which turned out to be the culprit in this mystery!  Ugh, I hope I don't have to re-do all the images I took with it...


Edited by TinySpeck, 22 September 2019 - 12:23 PM.


#24 niccoc1603

niccoc1603

    Ranger 4

  • -----
  • Posts: 318
  • Joined: 16 Aug 2018
  • Loc: Firenze, Italy

Posted 22 September 2019 - 02:03 PM

What reducer?

#25 TinySpeck

TinySpeck

    Mariner 2

  • *****
  • topic starter
  • Posts: 272
  • Joined: 08 Oct 2017
  • Loc: Seattle area

Posted 22 September 2019 - 05:58 PM

What reducer?

It's the AstroPhysics CCDT67.  The general consensus I found was that this was a good reducer to use with my scope.  At first I was using it near the nominal reduction factor (0.64) and seeing coma around the periphery.  I moved the camera closer to the reducer for a reduction factor of 0.73 and the coma improved significantly.  Most of my imaging has been done at that reduction factor.




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