M78 is difficult but rewarding object.
What was the secret with comet image?
M78... I'd tried imaging it last spring (after I'd first learned of it), but no guiding and limited time to image it before it set left me with just a handful of images to integrate. When I got the chance to image it over the winter I jumped on it.
Pixinsight was used for the processing of the comet, so how I did it is specific to it in some ways.
It required two images, a comet aligned image and a star aligned image, and Pixelmath to combine them.
Getting the rejection to remove (mostly) the star streaks caused by aligning to the comet, I still had to go in and use some cloning to get rid of some residual streaks. I ended up also using the Large-scale Pixel Rejection in the integration. I processed the comet aligned image until I thought it looked how I wanted.
I then registered/integrated a star aligned image (I think I used the comet aligned images for this) using rejection to git rid of the comet's core (leaving the gas cloud around it). I did the "usual" processing. Then I applied a median transform and with layers 1 - 6 selected, and 7+ deselected (this got rid of the comet's gas cloud in the star aligned image).
To combine the comet aligned image to the star aligned image I used the Pixelmath expression iif($T < 0.1, integration_comet, integration_comet + $T) The 0.1 was settled on after using a preview of a section of background to get statistics, and adjusted up/down via trial/error until I was reasonably happy with how the stars were looking. More processing followed to "blend" in the stars into the comet's cloud better.