Comet alignment integration - star trail rejection advice needed

[Jfakatselis]

I am having problems rejecting stair trails in my comet aligned subs integration.
I have been using the Windsorized Sigma Clipping with various values of sigma high ranging from the default of 4.0 all the way down to 0.25. Can't seems to get rid of the star trails completely. Here are a few samples with the sigma values listed in the image file names. Images have been fully calibrated and only had bkgnd neutralization to this point.

Can anyone provide any insight to a method for removing the residual trailing?
I'm using a Canon DSLR, so these are OSC images. Images are of Comet Linear/252p taken 4-6-2016. 38 x 90s subs at ISO 1600.

Many thanks in advance,
Jim Fakatselis

[Stu]

Lower the sigma high clip value in your image integration--I've gone as low as zero.  Then clone stamp the rest out. 

[Jfakatselis]

Thanks for the message, however, I do not believe cloning, a la Photoshop is the best option. Besides Sigmas of 0 also reduce the comet coma significantly. All those photons thrown away!
I was hoping that some other type of algorithm or combination of processes would be preferred for such a powerful tool as PI.
Would certainly appreciate alternative options.

Jim

[gvanhau]

Hello

This old post may help you.
https://pixinsight.com/forum/index.php?topic=2008.15

Regards
Geert

[Jfakatselis]

Thank you Geert. I did come across that post from Juan. It was written before the latest Comet Stack process was as developed as it is today but the essence is similar I will attempt to try it to see how well it will work on my image set.

I think this is a particularly difficult problem to resolve with any software, including PI. When the star trailing of a comet integrated image is as faint as sections of the comet, it seems that a sigma clipping technique is not the only process that is required.

However, as so many posts have appeared concerning this issue I was hoping to spur interest in some newer, innovative technique/process or combination of techniques/processes to handle this type of image with the present tool sets.

Thanks,
Jim

[Stu]

One thing you can possibly try is pixelmath.  Reject the stars as much as you feel comfortable, and then darken the background as much as you can.  At the same time, with your stars image, make the background a little lighter.  Then use the MAX function to take the maximum values of both images. 

Just a thought.