I've procrastinated a long time with facing the challenge of doing justice to the Horsehead and Flame nebulae -- a challenge for me anyway due to my past mediocre results when I tried combining strong standalone Ha data with RGB data.  There seems to be a vast array of methods for combining these channels that vary with each target.

The RGB data alone and the Ha alone both make interesting images and I've seen many good examples of both here on Astrobin as well as other places.  One of the challenges for me has been to combine these two interesting sets of signals into one composite image while keeping the contrasty and detailed Ha signal from overwhelming the more subtle details and brightness levels of the RGB signals.  I provide a mouseover with my image above to show how much detail and contrast is in the Ha signal alone.  I feel like I hit on a good balance with the processing steps I used here. 

For anyone interested, my processing steps (all with PixInsight) are provided below.

All Subframes
Calibrate flats with flat darks
Calibrate lights with master flats and master darks
Cosmetic Correction to remove hot and cold pixels from lights
Run Subframe Selector on each filter set of lights to remove lights with poor FWHM and eccentricity
Register and integrate 10 best lights from red filter set for use as alignment reference
Register all lights using above alignment reference
NormalizeScaleGradient each set of lights by filter
Integrate each filter set of lights using NSG weights

R G B and Ha masters
Dynamic Crop
Mure Denoise
Dynamic Background Extraction

R G B masters only
TGVDenoise for noise reduction with heavily stretched lum mask and local support using stretched lum

R G B  and Ha masters
StarNetv2

Ha master
Generalized Hyperbolic Stretch to stretch nonlinear and clip empty black pixels
Curves Transform to apply slight s-curve for increased contrast

Starless RGB combo
PixelMath to combine separate starless R, G and B channels
Autocolor script to calibrate color
MMT for noise reduction (8 layers) with strong lum mask
SCNR to reduce green
Generalized Hyperbolic Stretch to stretch nonlinear and clip empty black pixels
Color Saturation of selective areas of nebula with GAME mask

Starless RGB combo and Ha master
PixelMath to combine Ha with Red Channel
HistogramTransform to selectively reduce red cast from flame nebula with GAME mask
Curves Transform to selectively adjust flame nebula color with GAME mask

RGB Stars combo
PixelMath to combine separate R, G and B channels
Autocolor script to color calibrate
Repaired HSV Script to restore color to saturated/ off-colored star cores
MMT for noise reduction (8 layers) with strong lum mask
Color Saturation after using Histogram Transform to expand high range (to boost star color)
SCNR to reduce green
Generalized Hyperbolic Stretch to stretch nonlinear and clip empty black pixels (less stretch than starless RGB to avoid star bloat)
Curves Transform to apply s-curve for increased contrast

Starless HaRGB combo and RGB Stars combo
PixelMath to blend nebula and stars using screen blend expression
Curves Transform to apply s-curve for increased contrast
Unsharp Mask
Dark Structure Enhancement
Signature annotate
Resample to 50%