I began shooting more M106 data this year thinking I would add a bit to some prior data.  I was keen to try an isolate the anomolous arms in Ha and OIII.  I did succeed in capturing those arms, and a nice view of the central core in Ha, but I found it difficult to meaningfully show that data while keeping the core bright enought to look like an actual galaxy core.   Also, I found that the old CCD data was not pairing well with the newer CMOS data so I decided to reshoot entirely from scratch.  Unfortunately, since I was matching the framing to the old data, I repeated the sub-optimal framing.   That meant I had to crop a beautiful little pair of spirals in the corner.  To make myself feel better about that I included a full frame view so we can still see them

I've added before and after images of the Ha, Red and Cleaned Ha data.  These may be more interesting than what the result added to the image!  

Below is a description my primary processing steps for those might be interested.  This was done in PI and PS.

Luminance Processing:

My cutoff for FWHM was a relatively high 2.5" but I had a decent amount of subs below 2.25" - mostly below 2".  I did one Luminance integration with all <2.5" subs and one with all <2.25".  I wanted to use the sharpest for the stars and core where the photon sample size was high enough that I didn't need as much data.  The lower resolution data is fine for the background and dimmer structures.  After creating the two masters, I used a clone of the sharp one, converted it to non-linear, applied an S-curve, and used that as a mask on the sharp master.  I then blended the two linear masters with this Pixelmath expression:

SharpLum*Mask + LessSharpLum*(1-Mask)

I then stretched the result lightly, but still enough that the core was quite bright.  HDRMT was applied to a preview through a star mask.  I then added the preview to the original data using the SubstituteWithPreview script.

Next in PS I applied Russell Croman's NoiseXterminator, masking out most of the highlights as usual.  His process is fast and accurate but I would not apply it to sharp highlights (same with most noise reduction). I then created luminance masks and star masks, subtracting the stars from the luminance mask.  Using levels I ensured that the mask was fully white on the highlights, fully black on the background and shades of gray between.  This mask was used to apply unsharp masking at about the scale of my FWHM (~7 pixels) and 3X my FWHM (~21 pixels) for lane darkening in the core.

Color Processing:

I chose to use LinearFit on the RGB data, fitting to the red channel.  This did a good job of color balancing but I also applied Background Neutralization and Color Calibration processes.     After stretching with Masked Stretch I bumped up the stretch a bit with Histogram Transformation.   I next appled Russell Croman's NoiseXterminator in PI - no masking needed with color data.  I like to apply noise reduction prior to saturating color data.  I then added color saturation via the color saturation process.

In PS I added a bit more vibrance and a touch of saturation.  At this point I copied the luminance to a layer over the color data and set the opacity to 50%.  After merging the luminance layer into the color I have a color layer with lower noise.  A bit more saturation is now possible and we are getting to about the final level of color saturation.  I do this in measured steps to keep control of color noise and prevent clipping the most colorful parts.  

I add the luminance layer again now with full 100% opacity.  Merge that down for an LRGB image to work with.  My next step is more color work but this time in the LAB space.  Small curves are added to the A and B channels to increase the contrast of the color.  At this point the Ha emission areas are looking a lot like Ha data was added even though it wasn't.  

Ha Processing:

M106 has a pair of "anomalous arms" that are primarily Ha and are well outside the plane of rotation.  The upper arm is forked and easy to see because it is above, and thus in front of, the core.  The lower arm is more difficult because we have to look through the galaxy core.  but the magic of an Ha filter works on M106 white light the same way it works on man-made white light so we can see through that core!  

Because the true Ha emission line is less than 1 nanometer, most of the light we capture with a 5nm Ha filter is red continuum and not Ha.  To see the true Ha structure we need to subtract the red data from the Ha data.  Without attempting a true tutorial, here are the steps descriptively:

1) Remove the background levels from the red master [Red-med(Red)]
2) Subtract the cleaned red from the Ha using a scaling factor [Ha-CleanRed*.22]
3) Adjust the scaling factor up or down till you find just the point before parts of the image clip to 0.  [in my case I had a larger pedestal on the Ha so my factor ended up being well above 1]
4) If there are remnants of stars, either clipped black or still white, eliminate those with clone stamp or remove stars from the masters before you start.  You will never want to add Ha data to stars.
5) Finally, eliminate the background of the resulting pure Ha image [PureHa-med(PureHa)]

In most cases you will want to add this CleanPureHa data to both the red channel and the luminance channel using some factor to tune how much power you want the Ha to have.  

In my case here I did not follow the traditional route of adding the Ha.  I tried many methods but ended up adding the Ha data as a soft light layer with a mask limiting the effect to only where the Ha data existed.  This was not a dramatic impact but I liked the overall effect even if the anomalous arms are not super-prominent.

I did some testing with the OIII but I have not yet tried to eliminate the continuum from the OIII data.  Given how I ended up using the Ha data I felt it unlikely I would use it so I decided to go ahead and finish this image.