StarXTerminator v Starnet

[AstronomyUkraine]

Today I downloaded a trial of StarXterminator to see for myself how it stacked up against Starnet. The same image was used for both tests, just stretched using STF and HistogramTransformation, no other processing was done. The xisf was used in the Starnet version, which was then converted to tiff for StarXterminator in Photoshop.

 

There are 3 sets of images. The full size starless version of each image, then a close up of both the stars and resulting starless images performed by both applications. The results are conclusive on this particular image, which was probably one of the more difficult to remove the stars, because of the dense star field. StarXterminator performed the better of the two on this particular image, and I suspect it would perform better on most images. I see no discernible artifacts on the Xterminator images, except for a little smudge where the two bright stars where located. The Starnet version unfortunately leaves behind artifacts on both the starless and stars images.

 

Starnet is still a great tool, but I think StarXterminator has now overtaken it.

 

Update! Since downloading StarXterminator, they have released a version for Pixinsight, which also works on linear images. I'm interested to see how it works on linear images, this opens a lot of possibilities in processing.

 

Starnet Starless Full Size

 

 

StarXterminator Starless Full Size

 

 

Starnet Starless CloseUp

 

 

StarXterminator Starless CloseUp

 

 

Starnet Stars CloseUp

 

 

StarXterminator Stars CloseUp

 

 

 

Edited October 10, 2021 by AstronomyUkraine

[TerryMcK]

Great that it has also been released for PI too. I would give it a go in PI. Although converting to TIFF for PS and then reloading the resultant starless image in PI is not really an issue. You can still apply the xisf header from the original image to the starlessTIFF once reloaded in PI so you don’t lose anything.

Brian does StarXterminator also produce a separate image containing the stars like the starmask in StarNet?

[AstronomyUkraine]

2 minutes ago, TerryMcK said:

Great that it has also been released for PI too. I would give it a go in PI. Although converting to TIFF for PS and then reloading the resultant starless image in PI is not really an issue. You can still apply the xisf header from the original image to the starlessTIFF once reloaded in PI so you don’t lose anything.

Brian does StarXterminator also produce a separate image containing the stars like the starmask in StarNet?

What I found so far Terry. The module does produce a separate star mask, but the results are not great in Linear. The starless image is great, but the star mask is a disaster. Stretching the image then generating the star mask results in perfect stars.

[TerryMcK]

I’ll download a trial for the PI version in that case. Hopefully if after purchasing a license it gives free upgrades as bugs are found.

[AstronomyUkraine]

4 minutes ago, TerryMcK said:

I’ll download a trial for the PI version in that case. Hopefully if after purchasing a license it gives free upgrades as bugs are found.

It does warn you to keep the file in a safe place, as PI will remove the process during an upgrade. I didn't find any info about updates.

[TerryMcK]

Sold on the PI version of StarXterminator. Not only is it much quicker than StarNet it also did not produce a nasty halo around where one of the stars was in my California Nebula.

Both of these images started off the same with just some basic stretching and ABE applied.

First the Starnet version:

 

Note that big halo around the star

 

The same image run through StarXTerminator:

 

 

Completely devoid of artefacts.

 

I'll run it for the trial period to evaluate it fully then will probably make a purchase. I didn't look at the Photoshop version but some people on other forums have said it is better than the PI version but didn't say what was better.

[AstronomyUkraine]

4 hours ago, TerryMcK said:

Sold on the PI version of StarXterminator. Not only is it much quicker than StarNet it also did not produce a nasty halo around where one of the stars was in my California Nebula.

Both of these images started off the same with just some basic stretching and ABE applied.

First the Starnet version:

 

Note that big halo around the star

 

The same image run through StarXTerminator:

 

 

Completely devoid of artefacts.

 

I'll run it for the trial period to evaluate it fully then will probably make a purchase. I didn't look at the Photoshop version but some people on other forums have said it is better than the PI version but didn't say what was better.

 

The PI version only come out on the 8th, so is probably not advanced as the PS version yet. I can't say I notice any differences in quality using either software. Maybe it's a speed thing. Nope, the same image was used in PS and PI, both took exactly the same amount of time.

Edited October 11, 2021 by AstronomyUkraine

[AstronomyUkraine]

Just come across a potential bug in StarXterminator in Pixinsight. When lightly stretching an image before using StarXterminator, I get an error message coming up as soon as I apply the process.

 

"The target view appears to be linear (unstretched). StarXTerminator may give poor results without the 'Linear' option selected."

 

I ignored the message, and the process worked fine.

[TerryMcK]

I've just bought the permanent license for StarXTerminator for PixInsight. Slightly different installation from the trial otherwise it barfs.

• First of all open up the expired trial from Process>Object Recognition>StarXTerminator
• Click on the spanner icon (wrench) and enter the email address and license key
• Then close the trial which has been upgraded to permanent - but currently version 1.0.0
• Next go to Process>Modules>Manage modules
• Scroll down until you find StarXTerminator and double click the green tick (checkmark) and click done
• Click on Resources>Updates>Manage repositories
• Click Add and then add in the url you were sent from rc-astro.com (I won't reproduce it here)
• Click OK
• Click Resources>Updates>Check for updates and you will see two files to download
• Carry on with your normal PI update process and next time you reload PI you will see the full installation as before.

 

This time it has changed to version 1.1.0 which has a different AI version 6 installed. Russell says this about the update:

"This package installs AI version 6 for StarXTerminator. This update implements matching of noise statistics for replaced pixels, resulting in a more natural appearance and elimination of the "smooth" areas around stars seen in earlier AI versions. This version of the AI requires StarXTerminator module version 1.1.0 or later."

 

No doubt there are different ways to install the new version but the above process worked for me.

[AstronomyUkraine]

20 minutes ago, TerryMcK said:

I've just bought the permanent license for StarXTerminator for PixInsight. Slightly different installation from the trial otherwise it barfs.

• First of all open up the expired trial from Process>Object Recognition>StarXTerminator
• Click on the spanner icon (wrench) and enter the email address and license key
• Then close the trial which has been upgraded to permanent - but currently version 1.0.0
• Next go to process>modules>Manage modules
• Scroll down until you find StarXTerminator and double click the green tick (checkmark) and click done
• Click on Resources > updates> manage repositories
• Click Add and then add in the url you were sent from rc-astro.com (I won't reproduce it here)
• Click OK
• CLick Resources>Updates>Check for updates and you will see two files to download
• Carry on with your normal PI update process and next time you reload PI you will see the full installation as before.

 

This time it has changed to version 1.1.0 which has a different AI version 6 installed. Russell says this about the update:

"This package installs AI version 6 for StarXTerminator. This update implements matching of noise statistics for replaced pixels, resulting in a more natural appearance and elimination of the "smooth" areas around stars seen in earlier AI versions. This version of the AI requires StarXTerminator module version 1.1.0 or later."

 

No doubt there are different ways to install the new version but the above process worked for me.

Just done the update exactly the way you did Terry, worked like a charm. My trial version expired today, and I had no hesitation in upgrading to the full version. Well worth the $60.

[AstronomyUkraine]

Just had a quick play with the full version, and I'm really impressed. As you see from the image, which is some Ha data of NGC 1491, there are no visible artifacts from the star removal process. I saw two, what I thought were artifacts, but they didn't line up with any stars. I annotated the image, to find they are very small galaxies (relatively speaking), which could have easily been mistaken for stars. Very impressive!

 

[MarkAR]

Another method that looks like it works really well is Defect Map.

 

[AstronomyUkraine]

20 hours ago, MarkAR said:

Another method that looks like it works really well is Defect Map.

I tried this method, it works OK. The end result is better than using Starnet alone, but still falls short of StarXterminator, and is a lot more convoluted to achieve the end result. Plus you are still left with stars that have the grid pattern around them.