StarTools 1.9 preview

[Carles]

Thanks for the update, @admin

indeed, the speed works good, thank you for brining back some controls. Althought the mask leaves out some big stars that are impossible to get rid of the ringing effect. What Still like from 1,8 SVD is the ability to change sample size, that can have a big impact depending on dataset.

[decay]

Yes, as Mike said having all this discussions it's easy to forget what awesome results SV Decon yields. This was more of an experiment rather than a serious attempt to take a shot of M57. This object is too small for my 8" f/5 Newtonian (and for the seeing here) and on top of that I had a strong oscillation on the Ra axis . But this is what SV Decon can do:

2023-05-30 21_00_21-2023-05-30 12_31_51-StarTools.jpg - IrfanView.jpg (14.93 KiB) Viewed 31975 times

Wow! Simply awesome

The only thing I do not understand: The smaller (egg shaped) stars do collapse into small circles or even into four pixels. But what about the larger stars, like the one at the top right? The core seems to be tightened, but the overall shape is not shrinked? Any ideas?

Best regards, Dietmar.

[Dereksc]

I just released a new alpha with more control over the de-ringing parameters (as well as a substantial improvement to processing speed when using intra-iteration resampling). I think you have all helped me realise that there are just too many ways datasets can be compromised that require more hands-on control, e.g. the non-linear saturation-approaching case, as well as imperfect sampling). I have separated-out the single parameter into three parameters.

Let me know if that solves your particular issues - I have had good success getting rid of ringing in most of the cases you have been bringing up and demonstrating...

Hello Ivo,
Thanks for all your hard work! Because this is my first post let me also say congratulations on such a fine product.
I run into a hard crash using StarTools_1_9_552alpha - Linux64. The crash occurs in SVDecon mode of operation as soon as I press "Deap space (stars available)". The crash does not occur when using 551alpha so I hope it turns out to be something small.
The StarTools logfile does not contain any information about the crash, however here is an entry from the Linux system log file that shows the fault:

Screenshot_20230530_162709.png (92.7 KiB) Viewed 31955 times

Here are my system details:

Screenshot_20230530_164156.png (90.19 KiB) Viewed 31955 times

I tried several different projects that had previously run through 551alpha successfully - they all resulted in the same failure mode.
If you need any more information please let me know.
Regards
Derek

[Mike in Rancho]

Hi Derek.

Another in the Threadripper club, huh? I feel so left out. Don't know if that 1030 GPU is up to snuff though.

Althought the mask leaves out some big stars that are impossible to get rid of the ringing effect. What Still like from 1,8 SVD is the ability to change sample size, that can have a big impact depending on dataset.

Well I think I might be the king of ringing here. Might even have to donate my 8-hour M101 to Ivo for the "yes it's worse than you think" dataset archive...

I'm getting closer with 552 though. Have you tried adding a few ticks of increase dynamic range yet? That was the last key for me, after trying the focus and fuzz all over the place, and dumping linearity down to 1%. Must mean I have some really blown stars? Linearity to zero though gets weird. As does too low on the fuzz, which starts making polygonal shapes.

I still don't like being so low on the fuzz just for the problem stars, since that bloats the others that I would like to have tighter, but oh well. Anyway I also noticed that small changes to those sliders can make a difference.

I still have a couple stars that won't behave well but they are better now with the range extension and a couple clicks different fuzz. Odd too that it isn't the small or medium-small stars, or the biggest and brightest, but sort of a medium bright that is causing the most trouble.

Another thing I see is perhaps a little too strong on the shadowing around the bigger diffraction spikes (longitudinal ringing?), and a bit of scattered dark noise. I wonder if any or all of this stuff could be enhanced a little bit by the increased dark structure change.

Just added a bit more in denoise scaling to try to calm that down some. Overall the tracked denoise seems to be working really well though. I haven't even been using SS lately.

[Dereksc]

Hi Derek.

Another in the Threadripper club, huh? I feel so left out. Don't know if that 1030 GPU is up to snuff though.

Hi Mike,

Nice to see you. When I put together my linux box in 2017 I specially selected that graphics card for its performance
I am still waiting to get some new images of M101 (and SN2023ixf). May gray (and I expect June gloom) is in full force here. At least it gives me chance to go through some of my older projects and reprocess them using ST1.9.
Good job with all your alpha testing

Regards

Derek

[Mike in Rancho]

Hi Mike,

Nice to see you. When I put together my linux box in 2017 I specially selected that graphics card for its performance
I am still waiting to get some new images of M101 (and SN2023ixf). May gray (and I expect June gloom) is in full force here. At least it gives me chance to go through some of my older projects and reprocess them using ST1.9.
Good job with all your alpha testing

Regards

Derek

Slightly older era 6th gen Intel here, so I'm slower on CPU and down on cores and threads too.

I don't have anything Linux, so hopefully just a tweak or implementation update and you'll be good to go. Crashing at deep space in SVD seems like a mask module call though, which off the top of my head has been a point of issue before if maybe there's a hardware inadequacy. But that's just a wild guess. 552 did make a (much-needed) speed increase.

Nvidia generation model numbers followed by a 30 tend to be quite entry-level I thought. How much discrete GB onboard? I originally had a GTX 745 4GB in this thing, which worked, but was perhaps lacking some oomph. So I recently stuffed a 12GB RTX 2060 OC into the case for StarTools purposes.

[Dereksc]

Slightly older era 6th gen Intel here, so I'm slower on CPU and down on cores and threads too.

I don't have anything Linux, so hopefully just a tweak or implementation update and you'll be good to go. Crashing at deep space in SVD seems like a mask module call though, which off the top of my head has been a point of issue before if maybe there's a hardware inadequacy. But that's just a wild guess. 552 did make a (much-needed) speed increase.

Nvidia generation model numbers followed by a 30 tend to be quite entry-level I thought. How much discrete GB onboard? I originally had a GTX 745 4GB in this thing, which worked, but was perhaps lacking some oomph. So I recently stuffed a 12GB RTX 2060 OC into the case for StarTools purposes.

Hi Mike,
It doesn't take long for computers to show their age. In general I am still pretty happy with the Linux box though.
Stacking in PI (WBPP) is not too bad, I usually add the latest images to WBPP first thing in the morning and by the time I have had breakfast/showered/dressed the masters are usually ready for post processing. I think the longest stacking time I have had in PI is around 2hrs for HaLRGB (around 900 x ASI1600 subs). If it's just one night of ASI1600 images it usually takes about 20minutes total for LRGB master generation.
Processing with StarTools is also reasonable although I understand that upgrading the graphics card and utilizing the GPU would speed things up. At the time I put together my system the graphics card was just being used to display an image on the monitor Did the upgraded graphics card make a big difference for you? Maybe that would be a worthwhile upgrade for me at some point.
I'm not sure if the alpha552 crash is due to a hardware inadequacy - it may be - but alpha551 and the current stable version of StarTools have never crashed on my machine.
Regards
Derek

[admin]

I run into a hard crash using StarTools_1_9_552alpha - Linux64. The crash occurs in SVDecon mode of operation as soon as I press "Deap space (stars available)". The crash does not occur when using 551alpha so I hope it turns out to be something small.
The StarTools logfile does not contain any information about the crash, however here is an entry from the Linux system log file that shows the fault:
Screenshot_20230530_162709.png
Here are my system details:
Screenshot_20230530_164156.png
I tried several different projects that had previously run through 551alpha successfully - they all resulted in the same failure mode.
If you need any more information please let me know.

Many thanks for this report Derek.

I also run Linux during development, though running a Focal Fossa derivative (20.04LTS).

I am unable to replicate the issue however. Are you running the GPU or CPU version? If running the GPU version, does the CPU version crash for you as well? What is the size (X by Y resolution) of the dataset you are processing?

It does indeed sound like something might be going wrong in the Mask module. Can you confirm you updated the 'resources' file as well when you updated to 552?

Thank you!

[Dereksc]

I run into a hard crash using StarTools_1_9_552alpha - Linux64. The crash occurs in SVDecon mode of operation as soon as I press "Deap space (stars available)". The crash does not occur when using 551alpha so I hope it turns out to be something small.
The StarTools logfile does not contain any information about the crash, however here is an entry from the Linux system log file that shows the fault:
Screenshot_20230530_162709.png
Here are my system details:
Screenshot_20230530_164156.png
I tried several different projects that had previously run through 551alpha successfully - they all resulted in the same failure mode.
If you need any more information please let me know.

Many thanks for this report Derek.

I also run Linux during development, though running a Focal Fossa derivative (20.04LTS).

I am unable to replicate the issue however. Are you running the GPU or CPU version? If running the GPU version, does the CPU version crash for you as well? What is the size (X by Y resolution) of the dataset you are processing?

It does indeed sound like something might be going wrong in the Mask module. Can you confirm you updated the 'resources' file as well when you updated to 552?

Thank you!

Hello Ivo,
Thanks for the reply. I'm not on my Linux machine at the moment but I installed the alpha552 executable in its own directory so presumably the resource file is contained in the distribution zip file - I didn't change any files apart from the config for bigger text. The crash happens on both versions (GPU and CPU). The log file excerpt is for the CPU version. I tried three different projects and loaded just the lum channel stack to test each one. The fit file is from an unbinned ASI1600MM Pro but I did crop, wipe and then SVDecon.
I'll but up the machine in the morning and check X by Y resolution but it will be smaller than native ASI1600MM resolution. Also if there is anything else I can try to help confirm a Mask Module issue i.e. any other operation besides SVDecon in particular that may show the problem better please let me know. And just to re-confirm alpha551 is running well.
Best Regards

Derek

[Dereksc]

I run into a hard crash using StarTools_1_9_552alpha - Linux64. The crash occurs in SVDecon mode of operation as soon as I press "Deap space (stars available)". The crash does not occur when using 551alpha so I hope it turns out to be something small.
The StarTools logfile does not contain any information about the crash, however here is an entry from the Linux system log file that shows the fault:
Screenshot_20230530_162709.png
Here are my system details:
Screenshot_20230530_164156.png
I tried several different projects that had previously run through 551alpha successfully - they all resulted in the same failure mode.
If you need any more information please let me know.

Many thanks for this report Derek.

I also run Linux during development, though running a Focal Fossa derivative (20.04LTS).

I am unable to replicate the issue however. Are you running the GPU or CPU version? If running the GPU version, does the CPU version crash for you as well? What is the size (X by Y resolution) of the dataset you are processing?

It does indeed sound like something might be going wrong in the Mask module. Can you confirm you updated the 'resources' file as well when you updated to 552?

Thank you!

Hello Ivo,
Thanks for the reply. I'm not on my Linux machine at the moment but I installed the alpha552 executable in its own directory so presumably the resource file is contained in the distribution zip file - I didn't change any files apart from the config for bigger text. The crash happens on both versions (GPU and CPU). The log file excerpt is for the CPU version. I tried three different projects and loaded just the lum channel stack to test each one. The fit file is from an unbinned ASI1600MM Pro but I did crop, wipe and then SVDecon.
I'll but up the machine in the morning and check X by Y resolution but it will be smaller than native ASI1600MM resolution. Also if there is anything else I can try to help confirm a Mask Module issue i.e. any other operation besides SVDecon in particular that may show the problem better please let me know. And just to re-confirm alpha551 is running well.
Best Regards

Derek

Hello Ivo,
I booted up my Linux machine.
Here is a screenshot of the alpha552 directory

Screenshot_20230602_092447.png (184.97 KiB) Viewed 31738 times

Here is a screenshot of my image just before running SVDecon:

Screenshot_20230602_093003_res.png (364.2 KiB) Viewed 31738 times

If I can help in any other way please let me know.
Regards
Derek