I've been wanting to try a mosaic for some time but the timing never quite worked out. But when I framed up M33 on this telescope and saw that it couldn't quite fit into a single frame I decided the time was right.  Two panels framed the galaxy nicely but doing HaLRGB made for a lot of moving parts.

The data was acquired from October 31, 2021 through November 13, 2021 and was a total of 95.5 hours across the two panes. Here's how the acquisition broke down:

Pane 1:           46h 20m
Ha: 6h 20m
L:    10h
R:    10h
G:   10h
B:   10h

Pane 2:       49h 10m
Ha:  6h
L:    12h 20m
R:    11h 30m
G:    9h 20m
B:    10h

That's by far the largest project by telescope time I've done. The L and Ha were binned 1x1 and the R, G and B were 2x2.

I had two goals for this image. The first was to successfully complete a mosaic and the second was to keep my tendency to go overboard with contrast on galaxies under control. I definitely succeeded on the former and I think I succeeded on the latter though that's a bit subjective. I don't know why I have this tendency, especially with galaxies. I don't do it to terrestrial photography but for some reason I get sucked into trying to make every last detail be obviously visible to the detriment of the image. I tried to restrain myself on this one. 

This was some of the most complex processing I've done on an image. The data was so good I wanted to do it justice. I can't say it's perfect but I'm pretty happy with the result. It's the best I can currently do.

Here's how the mosaic was assembled:

1. Used dynamic crop to trim away the edges of all 10 masters (2 each of Ha, L, R, G and B).
2. Run the Image Solver script on all 10 masters
3. Run Mosaic By Coordinates script on each par of images per filter.
4. Run Trim Mosaic Tile on on all 10 masters
5. Run dnaLinearFit on the second image of each filter set using the first as reference.
6. Run Photometric Mosaic on each par of images making up one filter.

That produced five master images, one per filter with the panes joined up. However, for reasons that I don't understand, two of the images had a size one pixel off from the others in one dimension. This was caused in the Mosaic By Coordinates step and there didn't seem to be anything I could do to prevent it so I used StarAlignment to scale up the trouble makers to the same size as the others. This was imperceptible to the eye but it made the geometries match so I could combine them later.

Now that I had the mosaic panels joined I could start on normal image processing:

L:
dynamic crop
deconvolution
DBE
histogram transformation

H:
dynamic crop
starXterminator
masked stretch
tgv denoise
Curves (to clip the background)

R, G and B:
dynamic crop
DBE

RGB:
channel combination
PCC
histogram transformation (not a full stretch)
masked stretch (to finish the stretch)
LRGBCombination (to add in L)
Pixel Math (to add in H using RBA's method)
starXterminator (extracting stars)
I pulled a lum mask from this that I then used MLT to remove the first 8 layers until I was left with a diffuse galaxy shape). This mask was used for most of the next several steps either directly or inverted depending not he step until the stars were restored.
LHE @ 2 scales (normal mask)
Dark Structure Enhance (no mask)
Curves (inverted mask) to bring up the levels on the fringes
Curves (normal mask) slight color boost
Curves (normal mask) additional slight color boost
Curves (normal mask) slight contrast enhancement
Color Saturation (no mask) moderate boost to blues and yellows
Pixel Math in stars (using inverted mask)
EZ Star Reduction (using star mask generated by AdvStarmask script)
Curves (no mask) final contrast adjustment
dynamic crop (final dimensions)

In addition to keeping the edges of the galaxy soft while letting the detail toward the center come out I wanted to have enough color to provide some visual interest while not getting garish with the color. 

I also had a happy accident when putting the stars back. I intended to do that without a mask but accidentally left the inverted lum mask on. It had the result of minimizing the stars in front of the galaxy while leaving the stars toward the edges less affected. Overall it had the effect of giving M33 some context in the sky but letting the center of the galaxy shine through less obstructed. This might be a good technique for adding stars back in on galaxies and smaller nebulae where they don't fill the frame.