Two nice objects in the same field of view but different techniques needed to bring out the most detail in each--how to combine narrowand data for part of an image with broadband data for the rest? 

In my case I wanted to use RGB data from my OSC camera captured a few years ago for star colur and M108 but replace a broadband Owl with an HOO version. For better or worse, here's what I did. 

I created starless versions of my Ha and Oiii data, optimized each in non-linear form and roughly matched the background levels. I created an HOO image and tweaked the final result. 

Having registered my old broadband image to my narrowband one, I did fine tune it with tools that I hadn't had available when I collected the original data, but using BXT and SXT on the non-linear image didn't cause any obvious problems. So I have  a full broadband image, an RGB stars-only image and a starless HOO planetary with an HOO galaxy image. 

Matching the backgrounds of the broadband image with the HOO image as closely as possible, I created a preview tightly around M108 in the RGB image. Using the SubstitutewithPreview script, I replaced the HOO M108 with the RGB M108. I then gently used Curves to even up the background; mostly it seemed to be a difference in noise rather than a large difference in actual levels. So now I have a starless image with an HOO M97 and a broadband M108. All that was left was to ScreenStars back in and fine tune things as needed. 

Here is what I got. The framing isn't the greatest, as the original broadband data and the newer narrowband data were actually shot at 90 degrees to each other, resulting in a near-squre final image. The stars are not the best even with BXT, as the broadband data was collected before Astro-Physics made the flattener available for my vintage StarFire130EDF. 

The HOO image is composed of 24 300 second Ha subs ( 2 hours ) and 80 300 second OIII subs (6.7 hours ). Why the wide discrepancy, I can't recall! The RGB image was about 5 hours total integration. I suppose one could argue that just going deeper with broadband imaging would accomplish much the same result, but, well, I didn't go that route, at least not yet. 

I suppose there might be different and probably better ways to do this in PI, but I found this worked fairly well. I just wasn't happy with the results from  either ImageBlend or NBRGBC scripts, but it may be that I am not proficient enough with those to get a good result. 

If anyone has any suggestions for how to go about this in other ways, I would love to know. I suspect PhotoShoppers will have more options and likely an easier path, but I only have PixInsight.

Edit:  I revisited this using the ImageBlend script by Mike Cransfield. After realizing that I needed to replace the HOO galaxy rather than the planetary, I got a  better result, preserving the contrast, detail and colour of the HOO planetary as well as the stars embedded in the nebula and around it .