First go at M81 and M82

[glowingturnip]

looks pretty good - your small stars are all still there and look sharp, and I think I see all the detail in the galaxy.  Apply it, put a better stretch on it, and check you haven't introduced any funny looking artifacts anywhere

[SkyExplorer]

Looking at your picture of M81/M82, was that taken in light polluted skies? I'm wondering where all this red background has come from on my data.

Is that LP and sensor noise?

In your opinion, what should I be expecting to get this image looking like?

[glowingturnip]

ahem no, I do my imaging in a dark sky site in southern Spain - plus side is the beautiful clear skies, downside is I don't get out there nearly enough, I'm having serious withdrawal, haven't imaged anything since October.

You should be able to get at least as good a pic as mine, you've definitely got the data for it, much more than I had.  You don't have a supernova though   

There's always a colour cast with raw data, but Background-Neutralisation and Colour-Calibration will get rid of those.  Is the latest run of your processing after background neut and color calibn ?  There does look to be an equal amount of red and blue in your stretched background.

There are a couple of ways of killing background colour blotches.  firstly make a lightness mask and mask for the background only, then you can:

- in MLT again, Starlight Transform, Target Chrominance, 7 layers, turn off layers 4 to 7 and apply

- in Curves Transformation, the last tab on the right (in the middle of the dialogue box) is colour saturation - just drag the right-hand point down to about half way

Tell you what, if you want, you can send me a link to your raw tiff from DSS in dropbox and I'll go through my normal routines to see if we get to the same place.  I can let you have my process icons for each step so you can reproduce it.

[glowingturnip]

so here's where I've got to so far, about to the same stage you're at:

(posting it here, the background looks worse than it does in PI, but it's not a final image yet)

I've put a zip file here (https://www.dropbox.com/s/nigc7i063o5z7g0/M81_M82.zip?dl=0) with process icons etc for the different steps so you can play around with it as you like (nothing nasty in that zip, I promise).  Here's what I did so far:

- an initial crop to make it the same as yours (that crop included in the zip)

- DBE twice

- background neutralisation and then colour calibration using the previews shown in the screen grab

- MLT denoise

- TGV denoise

- then I did another DBE since the background was still quite lumpy - sometimes it helps to do a DBE after denoising since the sample points are then more uniform, and it did in this case

- an initial non-linear histogram stretch

- then a couple of background colour removal techniques (MLT and desaturation, mentioned earlier in this thread), masked for background (mask included)

I haven't done a deconvolution, which you could do.  I'd normally do that after denoising but before the stretch (has to be done on a linear image)

A couple of points about the raw file. 

Firstly as someone else noted, there is some quite heavy banding in the image, which you can see most strongly if you do an autostretch after the first DBE, but I can also still see when posting here (but can't see when viewing in PI, annoying, please do look at the bundled fits file).  There is a script in Pixinsight called CanonBandingReduction, which you could experiment with, though I've never used it so can't advise.

Secondly, the initial raw file had a very strong green cast to it.  Were you shooting through a light pollution filter which may have attenuated the red channel ?  Not a problem though, Background Neutralisation and Colour Calibration have completely removed the cast.  Notice that I haven't used SNCR yet, and probably won't - I'm thinking that maybe your use of SNCR damaged the data in your main data channel and gave you the fuzzy look to M81.

The image is only about half complete at the moment, with several steps to do, and more in the area where personal taste starts coming in to it:

- HDR and LHE masked for the galaxies only, maybe fading the effects back if they look too strong.  M81 could benefit from this.  (these have to be done after the first non-linear stretch, they are non-linear processes)

- sharpening, masked for galaxies only.  I like High-Pass sharpening even though it's not considered the 'Pixinsight way', and usually have to hop over to photoshop to get an HP filter to bring back to use in PI (I make my galaxy masks in photoshop too).  MLT can also be used quite well for sharpening.  UnsharpMask is usually too strong, and best avoided.  M81 can probably take some sharpening, it looks a touch soft at the moment.

- various colour separation and intensification (saturation) techniques for the stars and galaxies.  You've actually got some good colour in this image in the stars and M82 in particular, so it should look really good.

- specific curve transformations to help lift the fainter detail in M81 away from the background, and we may have to get a little creative to sort out that background banding better.

Do you want me to carry on with the next steps, or do you want to take it from here ?

Cheers, and thanks for letting me play, did I mention my AP-withdrawal ?

[SkyExplorer]

Wow - Thanks a lot Stuart, that image is looking great.

Yes I do use an LP filter, I guess I could try it without for the more southern sky objects.  M81 / M82 is er north, so for me being south of London, it places it nicely in the London LP zone.

To help with your AP withdrawal  , if you could finish the image I would be most grateful, that way I can use your version as a reference for my learning.  It would be kind of "here is the finished article" this is what can be achieved with my data.

This is also a proving group for me to decide if I should buy PixInsight and so seeing what you can do to my image will help me make that decision.

I have a lot of data from previous imaging session I did a few years ago, one of M31.  Once I get to grips with the process of image processing I may go back and have a look a re-processing the data from that.

Thanks again Stuart, I am very grateful for all your help.

Steve

[glowingturnip]

okidoke matey, i'll pull out all the stops, see what I can do

[SkyExplorer]

Can't wait to see the finished one 

Thanks for the process icons as well.  One question?  Can you only load one at a time?  I started off with the initial crop fit file and then applied the processes one by one to see how they worked compared to my attempts.

I also noticed the first DBE process used the default settings apart from sample radius which was 3 rather than 5.  When linked to the "initial crop".fit it showed 75 samples.  So as an experiment, I reset the DBE tool, changed the sample radius to 3 and clicked on generate, but I only got 33 samples. 

[glowingturnip]

not sure whether you can load more than one at the same time, but you can have more than one open.  From memory (I'm at work), there are two buttons on the process icon toolbar, one for opening the first one and another for opening additional ones.  Seems a bit silly.

Re the first DBE, actually I didn't change the radius size, but I would have changed the tolerance to a higher value until I got a spread of generated sample points across the whole frame.  Then I would have manually deleted any that were too close to the galaxies, and added others in areas of tricky background.  They say you only need a few sample points to give a good background model if they are carefully placed, and I dare say I tend to use too many...

[SkyExplorer]

Oh I see, are the process icons the initial stages of each process, or did you not save the first DBE after you had made adjustments to the settings?

[glowingturnip]

no, it should be the dialogue box for each operation with all the settings you've applied.  I routinely save mine so I can reproduce my steps if I need to go back, but maybe I messed that one up ?

[SkyExplorer]

The second DBE is also set to default options.  If Open the first DBE against the initial crop it shows 75 samples, if I click "generate" I then get 77.  If I then open the second DBE against the "initial crop"  it shows 98 samples, when I click "Generate" it then shows 77 again as both processes have the same settings.  

I am using version 1.8.03 of pixinsight, as this may be the issue?

[glowingturnip]

i'll check mine at home

[glowingturnip]

I see the first DBE with 75 samples, a tolerance of 5 (default 0.5), and mostly in a grid on the fit apart from where I've shuffled them away from the galaxies.  And yes, radius of 3 when the default is 5, not sure how that happened !

and DBE2 has the same settings, but you can see I've added more points manually on the fit in areas of tricky background, so 98 samples

The point being that once you've auto-generated the points, you can then manually move them around or delete them so that they don't land on stars or galaxy/nebulosity, or you can add new ones in areas with tricky background

[SkyExplorer]

:blush:   erm, now you mention it it looks the same on mine. 

When I click Generate, of course it will add what it thinks it needs as samples, however you have altered the amount of samples which is why it is different to when the process icon first opens.

Sorry for being stupid and wasting your time.  

[glowingturnip]

Sorry for being stupid and wasting your time.  

absolutely not, no problem at all !  There are no stupid questions, only stupid answers..  or something like that.

[glowingturnip]

I think I'm finally done, do you like it ?

The .fit fle is here - https://www.dropbox.com/s/6dxfwzlr03qdkvw/LRGB_final.fit?dl=0

Cheers,

[SkyExplorer]

Wow... Looks amazing Stuart. There is a lot of detail in m81 isn't there? Do you have the processes etc for the additional steps you did so I can understand how to process as good as this. How long did it take you? Thanks again

[glowingturnip]

haven't saved all the steps to be honest, and there were a few tricks here and there (the banding in the background was a bit tricky and I had to get a bit creative there), but broadly speaking, the extra steps were as in post 54.

My processing usually takes me a couple of hours a day for a couple of weeks on the sofa in the evenings, and this was no exception.

[SkyExplorer]

Okay, thanks very much for showing me what is possible with a bit of processing knowledge and experience.

My next target will be M33, so will try and get an equal amount of exposure time on that.

Thanks once again Stuart, if you are every around the Epsom area I'll get you a beer

[glowingturnip]

haha, i'll hold you to that, though has been a while, and there are usually horses involved !

[SkyExplorer]

lol, normally around the first Saturday of June in my experience.

[SkyExplorer]

Will now try and have a go at this old data from a few years ago of M33.  I have cropped, DBE, BN, CC, MLT, TGV and DBE'ed.

The stars are a bit "egg" shaped, and I may have to crop to remove the stacking join on the right, but want to try and practice with PixInsight

[glowingturnip]

yep, that looks pretty good, has promise

[SkyExplorer]

There seems to be a lot of black where you would expect to see the whispy outer galaxy.  Is that because I do not have enough data for the fainter outer galaxy?

[glowingturnip]

you seem to have most of it, as far as I can tell, a bit of careful stretching will bring it out. 

You didn't have any DBE sample points on any faint areas of galaxy did you ?  Might explain why it's slightly darker immediately round the edge of the galaxy compared to the rest of the background if so.  Best to err on the side of caution and keep them well clear of anything that could be structure.