The open cluster M37 contains almost 2000 stars and is spread across 20 light years. Most of the stars are young blue/white but it also contains red giant orange suns which provide considerable colour contrast. I first imaged M37 a couple of years ago but I decided to add to the data since I was never entirely happy about the quality of the some of the sub-frames. I also decided to try out some of my newly learnt stellar processing techniques to improve the data which I describe below for anyone that might be interested.
Since I often image in non-ideal conditions, I sometimes find that the red stacked channel has a larger FWHM values than the blue or green channels, if I combine this data to obtain an RGB image. the result is a red fringe around the stars. For blue stars, this is particularly apparent and creates magenta halos. This time, I decided to shrink the red channel via a ring mask using PI's erosion filter. The ring mask protects the star core, minimizing damage and core dimming. The main issue is generating a good ring mask which captures the majority of the red fringing. After watching one of the Adam Block's Pixinsight's tutorials ( https://adamblockstudios.com/) on how to de-emphase stars, I decided to apply a technique he recommends. Basically, you first create a star mask which includes all the stars and all the halos. This mask should be white across the star and halo and slightly feathered around the edges. Having done this, you then subtract from this the Red channel lum information to create the ring mask. The reason this creates an accurate ring mask is that the lum data contains accurate information on how the light profile various from the stars center. So, for example, at the stars core, the star mask will be close to 1 (in PI everything is normalised to 1), whilst the lum information will also be close to 1, so if you subtract one from the other you end up with something that is close to 0 (eg black) at the core. Thus, at the core you will be protecting the star's core almost 100%. Outside of the star mask (eg outside of the halos) you will have 0, whilst in the lum channel you might have 0.2 (say representing a non-stellar structure), if you subtract 0.2 from 0 you get 0, since you cannot have negative values in PI, so outside the star mask the ring mask will be 0 eg black. Within the halo, the ring mask will mimic the stars halo hence generating an accurate ring mask.
The LRGB image below was taken with my Esprit 150 and represents just over 15 hours integration time.