This image was difficult to process and I tried numerous approaches until I found one that mostly worked. Much of last month we had smoke covering the western US. This resulted in many rejected subs and the ones that were left still had some halos around the stars. I imaged this target over 13 separate nights.
I did standard preprocessing in PI using locaNormilization and DrizzleIntegration. After Integration I extracted a luminance channel. The color channel was handled separately mostly using CurvesTransformations and various color masks. The resulting noise was reduced with Convolution. The color channel was then combined with the Luminance data. Starnet was then used to reduce color and size of the stars. My attempts to use Starnet earlier in the process worsened the halos.
Here is a link to the high resolution version. https://astrob.in/9qrcsw/0/