|
I sometimes use a very low pass Chrominance Noise Reduction (2,1 going down) in narrowband images as all those tiny stars are a nightmare to pick up with a star mask. I find it is sometimes easier to get a decent star mask handling on the super bright stars while in linear data because there is such a difference between bright and not bright. I find that sometimes the bright stars are not as well masked as they can be, this leads to artefacts on bright stars when doing any kind of dynamic range compression (HDR/LHE).
I have since started making ranged star masks and using PixelMath to make a master star mask. The top field (cannot remember what it is off hand, has a default of 5) I make a mask at either 2,5&8 or 2,4,6&8. I usually keep the fields below for scale masking at 1,1,1 (large small and compensation). With PixelMath I'll use Max(smallmask,mediummask,...). If any of the brightest stars look a bit soft I'll either create an even bigger scaled mask (10) and see if that helps but at this stage it sometimes starts picking up large scale nebulosity. Failing that, I'll put $T*2 into PixelMath and hit it on the large scale masks to give it a better representation on the brighter stars.
Creating a strong star mask ends up being on of the first things I do as I have found so many processes effect the stars!
|