I still haven't got around to capturing separate luminance subs so I synthesise by combining RGB. My reason for subtracting green was my assumption that it's the only channel I can be reasonably sure has no significant contribution from hydrogen emission. In a perfect world I'd get a narrowband continuum filter just off from the Hα line (they are available) and subtract that data but as I'll eventually be recombining the Hα with the rest of the data, albeit boosted and with a suitable colour, I think Hα - Green/15 will be good enough.
BTW, I've just discovered that spiral structure in the bulge is also visible if one uses a PixInsight Star Mask (Harry's tutorial explain how) and then do some HDR Multiscale Processing on the luminance image.
OK, that's me done with PI for the day. Processing is going quite well but I think my attention level has peaked!