Lot’s of talk about NSG, looks like really great stuff. I have checked the code but it is very difficult (or impossible) to include into AI in it’s current form. I am really looking forward for that PCL module, then it should be easy to integrate (maybe except for the knobs).
Jarmo
ok i asked juan about the PJSR interface for SFS and he said it was just a mistake. he offered to change it, but WBPP also depends on this and the WBPP author already made the changes to be compatible with -10 and -11 so reverting is a bad idea as you already pointed out.
rob
Yes, keeping the new indexing is the best choice now. I guess there are no scripts in PixInsight scripts directory other than WBPP that use SFS.
Jarmo
looks like juan has rewritten LN and he believes it can be applied blindly now. apparently this will come as an update to -12 which means that prior versions can still suffer from artifacts if LN is applied blindly… don’t know if you want to think about making it the default again since it sounds like you would have to test the PI version (yet again) to decide whether or not it is on by default.
Sounds interesting, I need to check it out. I saw comments about LN changes in -12 but I understand from your comment that there are more updates coming.
For now, I guess I will keep the current default. With new save options in AI it is easy to make it a default if someone likes it better. If it really is better and problem free in most cases then maybe the default should be changed. But if it changes inside -12 release then it would need a way to detect the changed behavior.
I did some runs with -12 LN and it did work pretty well but I did see some halos around stars. So I did not want to change the default at that point. My testing was limited to a single comet image, though. I really need to find a good set of test images for LN.
Btw, I had a run with an image set of 10 images. My default choice for image rejection then is Averaged Sigma. Maybe it is my ImageIntegration settings (I use defaults), but it did not really do any rejection. I tried all other rejections from Percentile Clip to EDS and they all did a good job on rejecting bad problems. With so many choices available I do not really need to use Averaged Sigma but I found it weird. I have not done more testing like trying it manually.
I repeated the problem with Averaged Sigma rejection just by loading images to the ImageIntegration process. For comparison, I have also a version with Sigma rejection and it worked fine.
I was otherwise using default values for ImageIntegration. Do you know if there some special options for Averaged Sigma that should be used? Other rejections worked fine with defaults.
Here is a link to the image where it easy to see the problem:
do you have the dataset available? juan always wants to see the data when trying to reproduce problems. i’m not sure i’ve ever used that rejection method so i don’t know offhand if special care is needed, but in general with my own data i find that even with windsorized sigma clipping i usually have to adjust the sliders to get proper rejection.
Yes, I have the files. They are registered files. Data is from Telescope Live but I assume it is ok to share as I am sharing only Halpha files. If needed I can share original FITS files.
well, lowering the high sigma to even 2.8 produces reasonable rejection on that dataset. IMO this is normal - all these algorithms are different and there’s probably no way to come up with defaults that always work right. i guess the most flexible thing to do is to expose the high sigma control to the user… i have seen this problem on some number of projects i’ve worked on with AI.js but never really dug into it. i don’t see any user control over rejection algorithm or rejection controls but i might not be up-to-date.
Thanks for checking it out! It makes sense what you said, it is probably impossible that defaults would work well in all cases.
I do have an option for rejection algorithm in section “Processing settings … integration …”. The last collapsed section, if you have an older version. I think I will add some configuration for different options. I already have too many options, so I guess a few more does not hurt 
A bit off topic, I got my first APOD! I am pretty happy!
It must be the first APOD that is processed with AutoIntegrate script
!!! that is great news! it probably is the 1st with AI.js, congratulations! i didn’t realize t.l was using martin pugh’s observatory. it is a beautiful image.