New in PixInsight 1.8.5: new large-scale pixel rejection

[chris.bailey]

Jeff

I did a bit of playing with this yesterday. With reject large scale structures enabled you can crank up the sigma low and high (say 6 and 5) and still get the same level of rejection and get some SNR improvement as a result, in fact, with clean individual frames (Cosmetic Correction) you can dispense with the normal rejection altogether. As I tend to image the faint SNR and Sharpless objects this is a significant advantage.

Chris

[javajunkie2121]

Hi Chris:  I tried this with 12 blue subs of M51, 2 of which had star trails. I left the layers/growth values for high large scale structure pixel rejection at defaults of 2

..as soon as I start increasing the pixel rejection sigma low/high values from default 4/3(windsorized sigma clip), I get star trails returning?

also I tried adding "low large scale structure" rejection to high large scale structure rejection with sigma low/high back at defaults...compared to high structure rejection alone I think I'm seeing some loss of detail in the galaxy?  Is the low large scale structure intended for other targets?

jeff

[ngc1535]

JavaJunkie....

Could this be a method of rejecting dust donuts (due to flat errors) just like what I was looking for?

-adam

[pfile]

indeed it should adam, and i am remiss because i had trouble rejecting dust spots during the beta period and i flaked out on providing the sample data to juan. i can't tell if my trouble was due to misconfiguration of large scale rejection or not. i need to follow up with juan on that.

rob

[chris.bailey]

..as soon as I start increasing the pixel rejection sigma low/high values from default 4/3(windsorized sigma clip), I get star trails returning?

Jeff

Is you data dithered in acquisition?

Chris

[javajunkie2121]

yes Chris...was doing a medium dither via PHD2/SGPro with subs

jeff

[chris.bailey]

Jeff

I'm stumped then as the half dozen data sets I have thrown at it have shown similar or better rejection with significantly higher Sigma values and shown a worthwhile improvement in end result S/N.

Chris

[mmirot]

I was able to do a better job with satellite trails than in the past with 9 subs. 

This still required a sigma high of around 2.5 -2.9  using a Winsorized sigma set. 


I am using the large scale high settings of 2 for scale and 4 or 5 for growth. Only small traces on trails remain now remain.

These are best setting that I could see working with my images.  I don't know if this is truly optimal or typical.

Large scale low did not seem to have any impact.
I suspect Large scale low won't be usefully for satellite trails, RBI artifacts and cosmetic rays. These are bright structures.

I had been using Windsorized sigma high of 3.5 - 4.5.
However, I was also removing bright trails from the stack to use 3.5 - 4.5 too. 

Max

[ddahlstrom]

I was having difficulty figuring this out.  I have an image with 7 600s DSLR subs, one having a satellite trail.  Turning on Large-Scale Pixel Rejection (reject high) at the defaults of 2 and 2 seemed to do nothing, and raising the values higher yielded no improvement.  I eventually found the secret sauce for this image to be lowering layers (high) to 1 and increasing growth (high) to 7.  This, at last (and with default sigmas), completely eliminated the satellite trail in the integrations.   

[lucchett]

Hi, I read the thread and from my understanding activating Large scale rejection seems always beneficial.
Am I right that I should consider this a default option for any Image?

Thank you,
Andrea

[jkmorse]

Juan & Vicent,

I am just starting to scratch the surface on the new 1.8.5 tools, but this one is magical in its results.  I am going back to redo old processing projects and am amazed at the improvement generated by large-scale pixel rejection when I compare the old results with the new.  It just about eliminates the distracting blotchiness in my backgrounds and otherwise makes for a far cleaner image, with no discernable damage to good signal.  And the high and low rejection maps are things of beauty in showing the problem areas that have been fixed.

Just had to say thanks,

Jim

[Kerrywnm1]

All,

I am not having that great of results using ImageIntegration and large-scale pixel rejection to remove the satellite streak superhighway in my images. I have posted over on "Image Processing Challenges"  (https://pixinsight.com/forum/index.php?topic=13238.msg80235#msg80235) the info and a link to my data. Maybe some of you could please take a look and tell me what I can do?

Thanks,
Kerry

[pfile]

hi, as i said over there the best current technique is to paint over the satellite trails with white or black lines using pixelmath, and then normal pixel rejection will take care of it.

rob