How to get rid of asteroids?
Last night I took in total 2h of L, R, G and B data of NGC 891. It turns out there is a magnitude 16.4 asteroid close to the galaxy. When I use LN MAD winsor clip with kappa 3.0 and iterations 1 then the asteroid shows up as a small line in the final stack. Even when I set kappa to 1.5 and iterations to 3 the asteroid isn't completely rejected. Should I increase iterations or lower kappa even more?
Did you use dithering while capturing? That should help as well. If not, rejection is the only way and indeed you might want to try more aggressive settings as it's so bright. In such a case I would dial it up pretty extreme until it works and then go back to find a sweet spot.
I did use dithering while capturing but that won't help since all objects, including the moving object, are shifted over the same amount of pixels. OK thanks for the tip. Looks like APP treats sigma clipping in a much softer way than PixInsight, which is fine and just something I need to get used to.
A related question: I don't understand the effect of increasing the number of iterations versus lowering kappa. If I understand correctly then kappa is a measure for the level at which pixels are rejected and iterations simply the number of times that the clipping algorithm is used. What I wonder about is if there is a "golden rule" for what works best: only lowering kappa, only increasing the number of iterations or both. I am afraid the answer is "both" though 😉
You don't mention the number of subs for each stack in this situation and i think that is key here. The standard approach is to lower sigma, then increse the number of iterations. I get rid of satellite trails and asteroids in 6 sub average stacks with simple sigma clipping sigma=2 and it=1. You can also try to use median stacks.
In total 60 subs: 10 Red, 10 Green, 10 Blue and 30 Luminance. Thanks, I will try median stacks instead of average.