StarXterminator vs StarNet2

Having had a forced layoff from astrophotography for the last 8 months, I recently decided to get back into practice with PI and chose to redo some starless images I had done last year.  The reason being as the updated versions of StarXTerminator and Starnet features in PI stopped working and crashed my computor.  I assumed that both updates increased load on my overclocked system.  I therefore dialed the overclock back a bit and got StarXTerminator to work.  In general, the newest version well outperformed that which I used last year.  It better removed much brighter stars in fully processed images.  It also had reduced noise in the background and less induced artifacts (largish circles that replace stars).  Having said this, I noticed that not all images I applied it too were better, or noticeably better.

I have not yet gotten Starnet to work, but I think I know the reason and need to download the accessory files.  Next inside project.

Finally, even 8 months ago, I had switched from Starnet to StarXTerminator for my intended purpose because it tended to work better on unstretched images, for use in the Adam Bloch star reduction process.  But I noticed that it also worked better on stretched images (some) as well.  I look forward to testing them again when I get Starnet up and running again.  Hopefully someone will get you your answer before I do so.  I believe Starnet in PI is 2 or ++, basically updated periodically.

I have to say that I got a bit annoyed with having to keep searching for and reinstalling the video card drivers and data files each time PI updated, or Starnet updated.

Oh, and StarXTerminator has gotten a lot slower to run.  But still well worth it.

Starnet2 I have found to be superior to StarX.

The new version works both on linear and non-linear images. You may be interested in these videos:
(2) StarXterminator and More with Russ Croman - YouTube
(2) New StarXterminator 2 with AI 11 for PixInsight! - YouTube

Both are a bit promotional but still interesting and impressive. Obviously the software won't work equally well on any image. What I like is the attitude to work on improving it further without extra cost for those who already bought the software.

Clear skies
Wolfgang

Maybe I expressed myself poorly. The new version works on linear and non-linear images without active choice of the user. You just apply the software to your image whether it's stretched or not. Russ suggests to apply it to the unstretched image in PixInsight since any sort of processing may alter the characteristics of the stars and may make detection of stars more difficult for the algorithm.

Clear skies
Wolfgang

Maybe I expressed myself poorly. The new version works on linear and non-linear images without active choice of the user. You just apply the software to your image whether it's stretched or not. Russ suggests to apply it to the unstretched image in PixInsight since any sort of processing may alter the characteristics of the stars and may make detection of stars more difficult for the algorithm.

Clear skies
Wolfgang

Hello Wolfgang,

Just finished watching the video, and your correct. The algorithm has it built in where you don ‘t have to tell it linear or non-linear. It just works. I’ve found this to be an excellent tool in my workflow.

Dale

I installed a trial version of StarXteminator 2.0.X this morning having never used it before.  It is really impressive and produces much nicer star masks than Starnet++.  It is also faster (as long as you have Nvidia GPU acceleration enabled).  I also love the fact that it works seamlessly on linear images.  I tried it on twelve of my old images in various states of processing and it worked flawlessly.  Needless to say I had purchased a copy by lunchtime.

Andy Wray:

Dale Penkala:
Even with my Ryzen 9 it takes a little extra time then the previous version

Dale, do you not have an Nvidia graphics card in your processing PC?  If you do, it makes a huge difference to enable GPU acceleration.

Hello Andy,
Yes I’ve got a 3060 12gb GPU with 94gigs of DDR4 ram. I can’t believe how much fast this machine is from my original i5. I’m really glad I have it now because with the scripts and other Pixel Math expressions in PI it sure makes a difference. I really like the new version of StarXTerminator and I’m glad I now have the computing power to fully utilize this plugin for PI.

Dale

Andy Wray:

Dale Penkala:

Andy Wray:

Dale Penkala:
Even with my Ryzen 9 it takes a little extra time then the previous version

Dale, do you not have an Nvidia graphics card in your processing PC?  If you do, it makes a huge difference to enable GPU acceleration.

Hello Andy,
Yes I’ve got a 3060 12gb GPU with 94gigs of DDR4 ram. I can’t believe how much fast this machine is from my original i5. I’m really glad I have it now because with the scripts and other Pixel Math expressions in PI it sure makes a difference. I really like the new version of StarXTerminator and I’m glad I now have the computing power to fully utilize this plugin for PI.

Dale

Just checking:  what I was asking is have you enabled the GPU acceleration?  It's not that straighforward, so was just checking.  It should take about 5 secs on that GPU.

Hi Andy,
I the only thing I have done on is the bench marking with PI and added 6 swap folders. How would you “enable” the GPU acceleration?

Dale

Dale Penkala:

Andy Wray:

Dale Penkala:

Andy Wray:

Dale Penkala:
Even with my Ryzen 9 it takes a little extra time then the previous version

Dale, do you not have an Nvidia graphics card in your processing PC?  If you do, it makes a huge difference to enable GPU acceleration.

Hello Andy,
Yes I’ve got a 3060 12gb GPU with 94gigs of DDR4 ram. I can’t believe how much fast this machine is from my original i5. I’m really glad I have it now because with the scripts and other Pixel Math expressions in PI it sure makes a difference. I really like the new version of StarXTerminator and I’m glad I now have the computing power to fully utilize this plugin for PI.

Dale

Just checking:  what I was asking is have you enabled the GPU acceleration?  It's not that straighforward, so was just checking.  It should take about 5 secs on that GPU.

Hi Andy,
I the only thing I have done on is the bench marking with PI and added 6 swap folders. How would you “enable” the GPU acceleration?

Dale

I did it ages ago, so didn't make detailed notes, but I followed something like this: https://www.williamliphotos.com/starnet-cuda

Dale Penkala:
Ahh ya I tried to get that to work with my i5 and I had to make an account that stated I was some sort of developer so I didn’t do it. I believe there is a YouTube video on how to do it that I followed and I wasn’t comfortable making that statement. I really would like to have done that however. As you have mentioned its really soposto make a heck of a difference!

Dale

It's an absolute no-brainer .... with your GPU you will basically not worry about StarXterminator or NoiseXterminator ... press the button and before you have really had a chance to think about things both will have completed.  Same with Starnet++ although it is a little bit slower.

Alan Brunelle:

Dale, do you not have an Nvidia graphics card in your processing PC?  If you do, it makes a huge difference to enable GPU acceleration.

Andy,
I may have to look into video card acceleration for StarXTerminator.  I used it for Starnet, and it was amazing.  But StarXTerminator did not use it, or at least use it last year when I was still doing imaging.

But what is clear is that my CPU processors are now being pressed into use with StarXTerminator.  It was very easy to tell when Starnet switched to the GPU, because I could easily hear the different cooling fans switch over from the CPU to the GPU.  If I can get StarXTerminator to do so, I should be able to bump up my overclock on the CPU again.

But either way, it's still very doable.  Just nice to get some value out of the damn GPU!

Alan

@Alan Brunelle if you do please let us know as maybe I’ll have to revisit this idea. I’m not a computer wiz on configuring them so I’d need some guidance and I’ll have to go back to YouTube more than likely.

Dale

Alan Brunelle:
I may have to look into video card acceleration for StarXTerminator.  I used it for Starnet, and it was amazing.  But StarXTerminator did not use it, or at least use it last year when I was still doing imaging.

That's odd ... I just installed StarXterminator 2.0 for the first time this morning and it used the GPU acceleration that I had set up for Starnet++ out of the box.  It does need the latest Nvidia drivers, CUDA files and the right tensorflow libraries to match though.