Most of the time, the Ha signal in emission nebulae is MUCH stronger then the OIII or SII signal. Take my recent image of SH2-112 as an example:
All these are stretched and look similar in strength. But combining them results in this:
Increasing the red component brings out more SII details but also leads to a very red background:
To avoid that, I extract the luminance of this image:
Using HistrogramTransformation, I Auto zero shadows and highlights and use this as a mask for this image. The same red stretch now results in:
The sky stays dark and more SII details in the nebula. I don't worry about the magenta stars as I will fix them at the end anyway.