[RCC] Andromeda Galaxy - help with processing

Hi @Pedro A. Sampaio ,

i also think that your image is good. Some things might help you getting even better: ABE should only be used for a first look. Better use DBE. I am mostly following the workflow defined by @Kayron Mercieca here: LVAs Tutorials
Even if you follow another workflow, reading Kayrons tutorials may give you some hints what to improve?

CS
Michael

Hi Pedro,
This is a pretty decent image, especially for first light from astro-photography.  I would play around with curves to adjust the colour (remove some red) and contrast (S-curve) a little more, at this stage.
If I might suggest a way forward - one that focuses on learning rather than producing the "best " image right away.  Keep it simple, and add complexity with time.   
I would put away Starnet for now and learn star-masking, stretch with a star mask if you need to - this is sort of a manual masked stretch - after Archsinh  (be careful with black-point on archsinh).  This will give you a sense of how much to stretch stars versus other stuff.  You shouldn't need Starnet or even star masks, really, for Andromeda.  I would use Arcsinh stretch at first, then  use curves along with histogram display to stretch the rest of the way.   Using curves will give you a feel for the data and also give some control over the centre of the galaxy versus the stars vs the galaxy arms.  I know this seems loosy goosy, but it will give you a better sense of the data.
I would also put away Photometric CC (at least, for now) and do some manual background neutralization and colour calibration so you get a sense of what each does, and what limits to place on background versus stars & galaxy.
Try and get your colour balance correct prior to stretching, even if it isn't saturated enough.  Make sure you normalize your individual channels (likely means giving blue a boost with linear fit?).  I dont use SCNR to reduce green from OSC cameras - rather, extract your channels, normalize or adjust contributions with LRGB sliders, and recombine.
I agree with above - are you using any noise reduction?  If you are using filters - use Mure NR, otherwise MLT on linear + MMT, Atrous, or ACDNR in stretched.  I also agree with above on DBE vs ABE in this case - with manual point placement (no stars and off galaxy). -  prior to colour calibration.
You set yourself a high bar for your first image - low in the sky + using a reducer.  (Not a fan of reducers, too much fussing with backfocus)   So kudos to you for producing a great image.
If you save your process steps in Pixinsight and project file, you should be able to return to any point, and change what you do next.  If your colour went wrong, or you created artifacts, you can determine when this occured and go back to that step, fix and carry on.  Just start out simple first and remember there is no single workflow or parameters that always work and there is no single workflow that is "the best".
Cheers and Clear Skies,
Dave

Thanks for all the replies! Lots to unpack here!

@Michael Timm - Having no background in photography or imaging processing, I learned PI basically through LVA. My workflow is based on it, although I still skip some steps, like convolution - I've tried a few times, but the results were never good. Still, it's been a while since I've read all the tutorials, and I totally forgot he had a step-by-step tutorial for Andromeda! 

@Gernot Schreider  - Following LVA's tutorials, I started using only DBE. But I wasn't getting good results with my images, with very wild background models. Now that I think of it, it may have to do that my first targets were either large nebula complexes that took almost all frame (like Rho Ophiuchi) or packed with dense star fields (like  M8 and M20), where it was very difficult to place the background points. I then stopped using DBE and migrated to ABE, using functions of 1 or 2. That gave good results, TBH (see my previous images in my profiles). But I never tried using ABE + DBE, will try it out! Also, I stopped using BN for the same reasons (it was difficult to find a good background model in those images), and Photometric CC did some background neut. Also, never tried AutoHistogram, will check it out! 

@David Payne - before migrating to my new APO, I've been using zoom lenses. They don't work well with StarNet - elongated stars and CA meant StarNet would generate terrible artifacts. I used a lot of star masks, before I started using StarNet for my most recent images. I do need to learn a lot, specially with curves! Channel extraction is another thing I never tried. 

As for stretching methods, I initially used regular histogram, since that was what I was used to do (I started AP with GIMP). Then tried Masked_Stretch but never got good results (stars were horrendous). I stretched some images with Arcsinh_Stretch and for the most part I liked the results - specially since it seems to preserve color and protect highlights. But I guess the lesson is that each image is different and experimentation is key. 

As for noise reduction, I do MLT for noise reduction in the linear phase, and ACDNR right after stretching. I've been meddling with EZ Denoise, with some success.

Thanks again for the awesome feedback! 

I revised my original image, using curves - adjusted to reduce the reds, increased blue a little bit, increased luminance just a tad, and changed the temperature to a colder.

I'll probably reprocess the image in the near future, but probably after I gather more data on this subject.



My first Andromeda