65 replies to this topic

[Sun Tzu]

Has anyone compared the new GraXpert Star deconvolution to the Siril built-in deconvolution?

 

The GraXpert one is generally pretty good, but it doesn't handle bright red stars too well - especially when you intend to use Starnet later in the process. 

 

I'd love to use BlurXterminator, but I'm not ready to invest in PI at this point.

What are you seeing with bright red stars when you use GraXpert?

[buramu]

What are you seeing with bright red stars when you use GraXpert?

They become red balls with a tiny white core. Let me see if I can find an example.

[Sun Tzu]

Have you tried color calibration again after deconvolution?  I saw earlier this year that some people where having trouble with the star colors changing after GraXpert deconvolution and that seemed to work for some.  Also, does this happen regardless of what settings you use for decon?

[buramu]

Have you tried color calibration again after deconvolution?  I saw earlier this year that some people where having trouble with the star colors changing after GraXpert deconvolution and that seemed to work for some.  Also, does this happen regardless of what settings you use for decon?

I haven't really delved into it too much. The AI star deconvolution in GraXpert doesn't have too many settings, so not much to play around with.

But what I wanted to try was see if star desaturation in Siril before performing deconvolution alleviates the issue.

[Stevan Klaas]

Reprocessed an older image of Monkey Head with drizzle x2 and GHS. Seems a bit more contrasted...but not very sure of my hands...don't if it is background noise or real nebulosity up. But certainly more meet on the bone in terms of detail to carve out.

 

 

Previous version

 

Edited by Stevan Klaas, 07 May 2025 - 02:57 PM.

[Sun Tzu]

 

Reprocessed an older image of Monkey Head with drizzle x2 and GHS. Seems a bit more contrasted...but not very sure of my hands...don't if it is background noise or real nebulosity up. But certainly more meet on the bone in terms of detail to carve out.

 

 

 

Previous version

 

There is more there but there does seem to be something off with the background. Did you stretch rgb together or did you isolate the red or green channel at some point? Also, were any black point adjustments made? Can't quite put my finger on it. 

[Stevan Klaas]

No isolation, just linear black point thing.

 

I think the gradients were very violent from my balcony, I just need to train. Thanks.

Edited by Stevan Klaas, 07 May 2025 - 04:00 PM.

[Sun Tzu]

 

Reprocessed an older image of Monkey Head with drizzle x2 and GHS. Seems a bit more contrasted...but not very sure of my hands...don't if it is background noise or real nebulosity up. But certainly more meet on the bone in terms of detail to carve out.

 

 

 

Previous version

 

Also, did you try SCNR after GHS?

[Stevan Klaas]

I tried, it just changes a bit the tint of the cyanish background.

 

Even in the older image the color noise of the background it was present but I kept it lower. I think the original stack is not very good.

 

 

Also, did you try SCNR after GHS?

 

Edited by Stevan Klaas, 07 May 2025 - 04:07 PM.

[Sun Tzu]

I tried, it just changes a bit the tint of the cyanish background.

 

Even in the older image the color noise of the background it was present but I kept it lower. I think the original stack is not very good.

My first thought was full moon or other light pollution but it doesn't seem as high in the previous image.

[Sun Tzu]

Ok I have a strange issue.

 

I'm running Siril 1.4.0 beta1 on two PCs (both Windows 11).  On one PC the 2x Drizzle scripts run perfectly.  On the other PC they fail. The scripts are from the same file.  What could I be missing? No memory issues or anything and the PC that works with the scripts is actually the older and slower of the two.  Everything else In Siril works the same on both excepts those specific scripts.

[Peter Besenbruch]

They become red balls with a tiny white core. Let me see if I can find an example.

Hey, I can do that with blue stars, too! I have never gone beyond .5 on the "strength" setting. Mostly I'm down around .2 or lower.

[Sun Tzu]

Hey, I can do that with blue stars, too! I have never gone beyond .5 on the "strength" setting. Mostly I'm down around .2 or lower.

I'm usually .7 or lower on objects and .5 or lower on stars.

[Sun Tzu]

The info below may be helpful when processing galaxy images. Also, there are usually tons of reference images online to point in the direction you may want to go. Of course in the end, it's a personal choice.

 

 

Galaxy Color in Astrophotography 

 

In astrophotography, the colors in images can sometimes be enhanced or adjusted to bring out details and differences in the composition of galaxies.

 

However, the underlying reasons for the natural color variations lie in the stellar populations and dust content.

 

The colors we see in galaxies are primarily determined by the types and ages of the stars that dominate their light, as well as the presence of dust. 

 

Blue Galaxies:

 

 * Young, Hot Stars: Blue galaxies are typically dominated by young, massive stars. These stars are very hot and emit a lot of blue and ultraviolet light. Because they are very luminous, they outshine the older, redder stars in the galaxy.

 

 * Active Star Formation: The presence of many young, blue stars indicates active star formation is occurring within the galaxy. Regions of intense star birth often appear blue due to these newly formed, hot stars.

 

 * Spiral Arms: Spiral galaxies with prominent, blue spiral arms are actively forming stars in these regions, where gas and dust are compressed.

 

Red Galaxies:

 

 * Old, Cooler Stars: Red galaxies are usually composed of older, lower-mass stars. These stars are cooler and emit more light in the red and infrared parts of the spectrum.

 

 * Little to No Star Formation: The lack of young, blue stars suggests that star formation has largely ceased in these galaxies. The existing blue stars have already burned out, leaving the redder stars to dominate the light.

 

 * Elliptical Galaxies: Elliptical galaxies, which are typically older and have used up most of their gas and dust, often appear redder.

 

Mixed Color Galaxies:

 

 * Ongoing Star Formation in Some Regions: Many galaxies show a mix of blue and red hues. This indicates that star formation is still occurring in certain areas (appearing blue), while other regions contain older stellar populations (appearing red or yellow).

 

* Spiral Galaxies with Bulges: Spiral galaxies often have a yellowish or reddish central bulge, which contains older stars, while their spiral arms appear bluer due to ongoing star formation.

 

 * Dust Absorption: Dust within a galaxy can also affect its color. Dust preferentially scatters and absorbs blue light more than red light. Therefore, a galaxy with a lot of dust might appear redder than it would otherwise, even if it contains young stars. Regions with dense dust lanes can appear dark or reddish.

 

Summary:

 

 * Blue light generally indicates the presence of hot, young stars and active star formation.

 

 * Red light generally indicates the presence of cooler, older stars and a lack of significant ongoing star formation.

 

* The overall color of a galaxy is the combined light of all its stars and is influenced by the amount and distribution of dust within it.

 

 

Hope some find this information helpful.

 

Clear Skies everyone. 

[Stevan Klaas]

Very helpful ! Thanks for putting this together.

 

For color calibration, I usually search for the specific galaxy in wikipedia and use it as galaxy type in Spectrometric Color Calibration. It changes quite a lot the tint of the galaxy.

 

I also check astrobin to see how people process it, although there is a bit of everything, so I tend to rely on SPCC.

 

I found this tutorial very useful:

 

https://youtu.be/vuR...MEmmfWmNIIqkfjL

 

The info below may be helpful when processing galaxy images. Also, there are usually tons of reference images online to point in the direction you may want to go. Of course in the end, it's a personal choice.

 

 

Galaxy Color in Astrophotography 

 

In astrophotography, the colors in images can sometimes be enhanced or adjusted to bring out details and differences in the composition of galaxies.

 

However, the underlying reasons for the natural color variations lie in the stellar populations and dust content.

 

The colors we see in galaxies are primarily determined by the types and ages of the stars that dominate their light, as well as the presence of dust. 

 

Blue Galaxies:

 

 * Young, Hot Stars: Blue galaxies are typically dominated by young, massive stars. These stars are very hot and emit a lot of blue and ultraviolet light. Because they are very luminous, they outshine the older, redder stars in the galaxy.

 

 * Active Star Formation: The presence of many young, blue stars indicates active star formation is occurring within the galaxy. Regions of intense star birth often appear blue due to these newly formed, hot stars.

 

 * Spiral Arms: Spiral galaxies with prominent, blue spiral arms are actively forming stars in these regions, where gas and dust are compressed.

 

Red Galaxies:

 

 * Old, Cooler Stars: Red galaxies are usually composed of older, lower-mass stars. These stars are cooler and emit more light in the red and infrared parts of the spectrum.

 

 * Little to No Star Formation: The lack of young, blue stars suggests that star formation has largely ceased in these galaxies. The existing blue stars have already burned out, leaving the redder stars to dominate the light.

 

 * Elliptical Galaxies: Elliptical galaxies, which are typically older and have used up most of their gas and dust, often appear redder.

 

Mixed Color Galaxies:

 

 * Ongoing Star Formation in Some Regions: Many galaxies show a mix of blue and red hues. This indicates that star formation is still occurring in certain areas (appearing blue), while other regions contain older stellar populations (appearing red or yellow).

 

* Spiral Galaxies with Bulges: Spiral galaxies often have a yellowish or reddish central bulge, which contains older stars, while their spiral arms appear bluer due to ongoing star formation.

 

 * Dust Absorption: Dust within a galaxy can also affect its color. Dust preferentially scatters and absorbs blue light more than red light. Therefore, a galaxy with a lot of dust might appear redder than it would otherwise, even if it contains young stars. Regions with dense dust lanes can appear dark or reddish.

 

Summary:

 

 * Blue light generally indicates the presence of hot, young stars and active star formation.

 

 * Red light generally indicates the presence of cooler, older stars and a lack of significant ongoing star formation.

 

* The overall color of a galaxy is the combined light of all its stars and is influenced by the amount and distribution of dust within it.

 

 

Hope some find this information helpful.

 

Clear Skies everyone. 

[Sun Tzu]

Very helpful ! Thanks for putting this together.

 

For color calibration, I usually search for the specific galaxy in wikipedia and use it as galaxy type in Spectrometric Color Calibration. It changes quite a lot the tint of the galaxy.

 

I also check astrobin to see how people process it, although there is a bit of everything, so I tend to rely on SPCC.

 

I found this tutorial very useful:

 

https://youtu.be/vuR...MEmmfWmNIIqkfjL

That videos is a perfect addition to the discussion.  Thanks!

[snakehelah]

Before the SeestarS50 I was a complete newbie with 0 astrophotography experience. I didn't really intend to get the unit unless I could be stacking and editing the data myself.

After watching some basic tutorials I took note of a simple workflow. Got the trials for Pixinsight and the blur/noise/starXterminator plugins to test drive everything. Of course, this is still a work in progress, and I realize that you can go even further beyond with even the stacking process, not to mention the post processing. But that's currently beyond my skill level.

Here is my general workflow with Pixinsight:

Stack using method of your choosing. Then, post process the stacked image:

1. Dynamic Crop if needed.
2. Background Extraction (auto dbe or graxpert)
3. Spectro color calibration (create region of interest preview with empty background for best results)
4. blur exterminator
5. downscale if needed based on drizzle
6. noise exterminator
7. Seti astro statistical stretch
8. Star exterminator (generate star image+unscreen stars)
9. HDRmultiscale transform on the object image if applicable
10. Curve transformation for either images as needed
11. Remove green with SCNR if required
12. You can make a greyscale version of your object to apply a mask to your main object image - check if curve transformation helps or is needed with the mask, invert mask if needed and do another curve transformation
13. After everything is done, remove the mask (mask steps are not necessary, there's probably a better way to do this - I am quite bad with masks).
14. Use Pixel math to combine star+ starless, ~((~X)*(~X)) for expression when using starXterminator
15. Final touches at your discretion - use GIMP or any other software if needed

M101, M106, M63 examples

P.S 

How does everyone save/convert their images when they share them? I feel like I lose way too much detail when converting to JPEG to be able to upload to most sites including here.

Attached Thumbnails

• 
• 
• 

Edited by snakehelah, 08 May 2025 - 05:40 PM.

[Sun Tzu]

Before the SeestarS50 I was a complete newbie with 0 astrophotography experience. I didn't really intend to get the unit unless I could be stacking and editing the data myself.

After watching some basic tutorials I took note of a simple workflow. Got the trials for Pixinsight and the blur/noise/starXterminator plugins to test drive everything. Of course, this is still a work in progress, and I realize that you can go even further beyond with even the stacking process, not to mention the post processing. But that's currently beyond my skill level.

Here is my general workflow with Pixinsight:

Stack using method of your choosing. Then, post process the stacked image:

1. Dynamic Crop if needed.
2. Background Extraction (auto dbe or graxpert)
3. Spectro color calibration (create region of interest preview with empty background for best results)
4. blur exterminator
5. downscale if needed based on drizzle
6. noise exterminator
7. Seti astro statistical stretch
8. Star exterminator (generate star image+unscreen stars)
9. HDRmultiscale transform on the object image if applicable
10. Curve transformation for either images as needed
11. Remove green with SCNR if required
12. You can make a greyscale version of your object to apply a mask to your main object image - check if curve transformation helps or is needed with the mask, invert mask if needed and do another curve transformation
13. After everything is done, remove the mask (mask steps are not necessary, there's probably a better way to do this - I am quite bad with masks).
14. Use Pixel math to combine star+ starless, ~((~X)*(~X)) for expression when using starXterminator
15. Final touches at your discretion - use GIMP or any other software if needed

M101, M106, M63 examples

P.S 

How does everyone upload their images when they share them? I feel like I lose way too much detail when converting to JPEG to be able to upload to most sites including here.

Thanks for sharing. Question- Are you doing the statistical stretch before star removal or after?

[Stevan Klaas]

Very nice shots snakehelah.

 

Usually the “orthodox” pixinsight school applies SCNR (#11) just after SPCC (#3) still in linear stage, also extract the stars (#8) before stretching (#7). I can’t explain the reason but there must be one. Anyway experience is what matters.

 

Also I use Gradient Correction, or MARS instead of autodbe or graxpert (#2).

 

For the mask, in non linear, I tend to use multimedian scale transformation with higher number of layers and remove the first two, the residual R, and the higher one before R that keeps the objects but remove completely the background, I increase contrast a bit to make it more protective, sometimes apply morphological transformation to dilate it, and convolution to smoothen it.

 

Long story short: I still struggle to get the background out of starless, the noise treatment makes is super smooth, but it is still there, only nicely spread like peanut butter everywhere. When you stretch it keeps wanting its revenge.

 

Would be nice if someone would do a “get background out of the starless” AI thing in Siril, Pixinsight or Graxpert to avoid that tedious process.

 

Before the SeestarS50 I was a complete newbie with 0 astrophotography experience. I didn't really intend to get the unit unless I could be stacking and editing the data myself.

After watching some basic tutorials I took note of a simple workflow. Got the trials for Pixinsight and the blur/noise/starXterminator plugins to test drive everything. Of course, this is still a work in progress, and I realize that you can go even further beyond with even the stacking process, not to mention the post processing. But that's currently beyond my skill level.

Here is my general workflow with Pixinsight:

Stack using method of your choosing. Then, post process the stacked image:

1. Dynamic Crop if needed.
2. Background Extraction (auto dbe or graxpert)
3. Spectro color calibration (create region of interest preview with empty background for best results)
4. blur exterminator
5. downscale if needed based on drizzle
6. noise exterminator
7. Seti astro statistical stretch
8. Star exterminator (generate star image+unscreen stars)
9. HDRmultiscale transform on the object image if applicable
10. Curve transformation for either images as needed
11. Remove green with SCNR if required
12. You can make a greyscale version of your object to apply a mask to your main object image - check if curve transformation helps or is needed with the mask, invert mask if needed and do another curve transformation
13. After everything is done, remove the mask (mask steps are not necessary, there's probably a better way to do this - I am quite bad with masks).
14. Use Pixel math to combine star+ starless, ~((~X)*(~X)) for expression when using starXterminator
15. Final touches at your discretion - use GIMP or any other software if needed

M101, M106, M63 examples

P.S 

How does everyone save/convert their images when they share them? I feel like I lose way too much detail when converting to JPEG to be able to upload to most sites including here.

 

Edited by Stevan Klaas, 08 May 2025 - 11:23 PM.

[Sun Tzu]

Very nice shots snakehelah.

 

Usually the “orthodox” pixinsight school applies SCNR (#11) just after SPCC (#3) still in linear stage, also extract the stars (#8) before stretching (#7). I can’t explain the reason but there must be one. Anyway experience is what matters.

 

Also I use Gradient Correction, or MARS instead of autodbe or graxpert (#2).

 

For the mask, in non linear, I tend to use multimedian scale transformation with higher number of layers and remove the first two, the residual R, and the higher one before R that keeps the objects but remove completely the background, I increase contrast a bit to make it more protective, sometimes apply morphological transformation to dilate it, and convolution to smoothen it.

 

Long story short: I still struggle to get the background out of starless, the noise treatment makes is super smooth, but it is still there, only nicely spread like peanut butter everywhere. When you stretch it keeps wanting its revenge.

 

Would be nice if someone would do a “get background out of the starless” AI thing in Siril, Pixinsight or Graxpert to avoid that tedious process.

By applying (8) before (7) you maintain more control of both the stars and the object. You can stretch the object without over stretching the stars and vice versa. Separate, then stretch individually.

[snakehelah]

By applying (8) before (7) you maintain more control of both the stars and the object. You can stretch the object without over stretching the stars and vice versa. Separate, then stretch individually.

Thank you. I will try that on my next galaxy. It's galaxy season so I'm mostly capturing galaxies but weather has been horrible lately without much ability to gather any data.

[Highnite]

Thanks for pointing me here Sun Tzu. I will gain more knowledge, Hopefully to retain it as well. I have already picked up some good suggestions.

[Sun Tzu]

Thanks for pointing me here Sun Tzu. I will gain more knowledge, Hopefully to retain it as well. I have already picked up some good suggestions.

I think it was actually Stevan that did the pointing . Glad you found it. I'm always up for learning so please share anything you may come across that may be helpful to some or all of us.

[Sun Tzu]

I've found lots of useful info at the following site:

 

 https://unofficialse...ku.php?id=start

 

Clear Skies to all

[Stevan Klaas]

Reprocessing of M100 with Drizzle x2 and double integration.

 

After

 

Before