Pelican Palette Pixinsight Problems

[Oblong]

Hi everyone,

I'm trying to get a decent Hubble Palette effect from some narrowband data on the Pelican Nebula. This is only 6x10m on each channel for the moment (clear, dark nights are rare and short up here). R,G,B is S2,Ha,O3 as a starting point. First, I tried using SCNR in Pixinsight to remove excess green (about the simplest suggestion I could find on the internet), and got something like this:

For comparison, I then had a go with the Photoshop method suggested in Sky at Night Magazine's AP special. This gives surprisingly different results:

I strongly suspect that neither of these attempts is optimal. Is there a published formula for how NASA do it, or is there always an element of subjectivity involved in balancing the relative magnitudes of the various emission lines? Perhaps someone can point me towards a more sophisticated way of doing Hubblesque palettes in Pixinsight?

Thanks a lot!