Andromeda

[rickwayne]

I suppose I'm not really a beginner anymore, since I've been whacking my head against this stuff for years now. Still, this is where I mostly hang out, so here you go. I've spent more like seven nights on this, but these two nights gave me the best data, it would appear. Just shy of five hours total of LRGB. Tech deets at astrobin.

[callisto]

WOW 😮

[Adam1234]

Incredible 

[maw lod qan]

Amazing! 

I was out this evening and turned in that direction for a long exposure widefield. I was happy just getting it in that image!

You should be tickled pink with yours!

[orion25]

Stunning!

[rickwayne]

Oh thank you, thank you, it does my flabby little black heart good to hear those lovely things.

Of course like all of us I adored my image for maybe thirty whole seconds before its flaws started to pulsate and flash, but hey, I have stars that aren't little arrowheads in this one!

This is my very first commissioned astrophotograph -- my wife  wanted an Andromeda for the wall but knew how much  I shuddered and twitched every time I looked at my old one. She agrees that it needs more margin, but is not as bothered as I am by the puzzling lack of color.

[wimvb]

7 hours ago, rickwayne said:

but is not as bothered as I am by the puzzling lack of color.

That’s what bothered me too. The detail is fine, but at 4+ hours, I would expect more colour. What does the rgb image look like? You also have dark areas in the lower right and upper left around the galaxy. This is usually a result of inadequate gradient removal. Very common if you use ABE in Pixinsight, but I have no idea if APP caused it in your image.

Edited November 18, 2020 by wimvb

[gorann]

I like the thick cloudy feeling you got in the central parts of the galaxy, but like Wim I suspect that some background correction procedure has created a too abrupt end to the galaxy, it actually stretches out quite a bit. To get some more colour you could try to change it temporarily to LabColor in PS and tweak the a and b curves in s-shapes after anchoring them in the center. There are other tricks too - maybe google astrophotography and colour boosting.

[ollypenrice]

I'm right with Wim and Goran on their comments. The bulk of the galaxy is thick and rich and satisfying. Like them, I think the dark ends have been created by a gradient tool reading the ends of the galaxy as background sky and pulling them down.

And then there's the colour, which is almost entirely lacking. This will be something to do with your workflow and can almost certainly be addressed but what software do you use? I know some tricks in Photoshop if they'd be any help.

Olly

[wimvb]

If you upload the masters (or pm me) I can have a look at them and maybe come up with a remedy. Mind you, I don’t use APP or PS.

[wimvb]

Since the galaxy is starless, I suspect you blended two images? This can also be the reason why you have those two dark patches.

Edited November 18, 2020 by wimvb

[tomato]

The dust lane detail is there in abundance, but as Wim has noted the foreground stars in front of the galaxy are missing as is distinct star formations within the galaxy, NGC 206 for example. Is this a result of processing techniques used to pull out the dust detail I wonder?

On the colour aspect, I find this target  can behave strangely. I pulled stronger  colour from a single OSC camera than I could with RGB filters on a mono set up, although that is certainly not a like for like comparison. But there should certainly be some there from 5 hours of LRGB integration.

[Jamgood]

I find Andromeda a bit of a nightmare to be honest. Easiest target to find/capture but the hardest to process. My last and best attempt was with my old set up. (130PDS, EQ3Pro, Canon EOS200D) I got about 5hrs of 1min subs.

Never looked quite right but as I'm learning, I'm happy with it. It was nice to catch a meteor in one of the subs.

I'll try again with my new set up sometime. With the HEQ5 I can get much longer and more details subs.

Although yours is devoid of stars and colour, I quite like it. It's almost black and white. Looks arty. 

[rickwayne]

Yes, the galaxy image is out of starnet++; I used Screen blending mode in Photoshop on the original image to put the stars back, then messed around till I liked the look. One thing Screen mode did was brighten the galaxy intolerably -- it's a technique that seems to work better on dimmer nebulae -- so I just masked it out. And probably went a little too crazy on the mask in doing so, leaving the objectionable black borders.

The color...I dunno. If I apply much stretch at all to the original image the stars all wind up near 250 for red, green, and blue. I tried exporting the combined image from APP with no stretch, and starnet didn't have enough to work with. Tried again with just a teensy smidgen of stretch so that the nebulosity was just detectable, and by cranking the Saturation slider on the result in PS I was able to see color in stars, but it didn't look natural. And the core of the galaxy gained some color, but it was more yellow-rose than straw colored. That also induced some REALLY ugly gradients between the panels of the mosaic, since they were shot at different moon stages. APP's light-pollution/gradient reduction tool knocked back a lot of that, but by no means all.

I'll see about posting some of the data. Thanks for the help folks!

[rickwayne]

4 hours ago, Jamgood said:

I find Andromeda a bit of a nightmare to be honest. Easiest target to find/capture but the hardest to process.

That's one of the reasons it's so appealing to me -- a target that literally anyone with a camera can get a picture of, but a near-infinite well of challenge. M42, another "beginner" target, is exactly the same way. Wax on, wax off.

[vlaiv]

For some reason - galaxy is clipped. There are a lot of processing artifacts in the image - uneven background color and similar.

Galaxy nebulosity in reality extends at least to stars that I marked - but there is black ring around galaxy before that. Couple of hours of exposure should have easily shown that part.

This is probably due to aggressive use of background wipe techniques.

Starless version of galaxy is indeed interesting, but I don't really like stars that have been kept in the image, they are strangely processed - as if out of focus, lacking definitive core.

Smaller stars are nice and tight / pinpoint - those are fine, but look at larger ones, they don't look as good.

 

[Viktiste]

Any chance you could post the individual L,R,G,B stacks?

[callisto]

Colors/no colors...large stars/small stars....For a noob of A/P this is a fantastic image and I would proudly have it on my wall....well done 

 

Mark

[rickwayne]

Well, I'm not exactly crying in my beer over it, but the whole idea is to get better, right?

Vlaiv, the weird gray stars are partly an artifact of how I blended the "stars" layer back in, via Screen blend mode in Photoshop and the layer opacity. I was going for a more subdued look so that the stars didn't distract so much from the galaxy, but absolutely gave up too soon! Having the stars gray makes them "pop" less, to be sure, but I agree that's not acceptable. I've got a new version coming that should be a lot better with that.

As far as I can tell I don't actually have any good nebulosity in that awful black region, but maybe I'm wrong. I don't know why the bigger stars are so fuzzy -- I was reasonably strict in culling my subs. Maybe it's an artifact of the Screen mode again -- if I'm going to be culling the stars in front of the galaxy for artistic reasons anyway, perhaps I should just mask in the surrounding region unchanged instead of relying on a blend mode. I should probably figure out how to shrink the stars, a technique I've never mastered.

If anyone would like to play with the integrated channel stacks, please feel free, they're available from my Google Drive. I'll warn you that there are some pretty awful gradients due to (a) spreading the imaging over nights with wildly varying Moon phases and (b) struggling with Ekos's mosaic scheduler so that some nights were entirely missing parts of the image! If anyone would find useful the details of how I set this up in Astro Pixel Processor, let me know.

Thanks again for the help. I'm delighted to have informed criticism.

[vlaiv]

31 minutes ago, rickwayne said:

Vlaiv, the weird gray stars are partly an artifact of how I blended the "stars" layer back in, via Screen blend mode in Photoshop and the layer opacity. I was going for a more subdued look so that the stars didn't distract so much from the galaxy, but absolutely gave up too soon! Having the stars gray makes them "pop" less, to be sure, but I agree that's not acceptable. I've got a new version coming that should be a lot better with that.

As far as I can tell I don't actually have any good nebulosity in that awful black region, but maybe I'm wrong. I don't know why the bigger stars are so fuzzy -- I was reasonably strict in culling my subs. Maybe it's an artifact of the Screen mode again -- if I'm going to be culling the stars in front of the galaxy for artistic reasons anyway, perhaps I should just mask in the surrounding region unchanged instead of relying on a blend mode. I should probably figure out how to shrink the stars, a technique I've never mastered.

I've found that following works rather good for star blend with nebulosity if one uses StarNet++ to do starless image:

1. Work with luminance only - leave color as is and do either LAB or ratio RGB transfer of color

2. With luminance - do linear stretch first - until you almost hit saturation in brightest parts of target. Don't worry about stars clipping at this point. You want dynamics of image as best spread over 16bit range so you don't loose anything over limited precision - this would not be necessary if StarNet++ worked in 32bit mode - but it does not

3. Produce starless image

4. Take image from step 2 and subtract starless image. This will give you only stars. Save that image

In the end, take image from step 4 - pure stars and as top layer - set it to max (or brighten in PS if I'm not mistaken - you want it to show if it is brighter than background layer - and stars always are).

I think you should be careful with setting background removal parameters as it tends to produce dark halos around bright objects.

Here is 1 hour of M31 from very light polluted location (border of red/white zone) done with color camera and Luminance extracted:

I scaled it down so you can easily see extent of the galaxy (and since it was taken with wide photo lens - it does not look nice as at native resolution ). Even with this short exposure it's starting to show its true extent - and it is well beyond even those stars that I marked in your image.

It almost reaches from the top to the bottom of this image. Note that M32 is completely surrounded with faint glow from M31 - it is not "detached", and even M110 looks like it is joined to M31 by faint nebulosity.

I'll see what I can do with your files - at least demonstrate starless approach if anything.

 

[vlaiv]

I'm not sure I can do anything with this data, as it has not stitched properly - not sure what software / workflow was used for stitching, but here is deep stretch of luminance:

Maybe try first to see if you can get decent linear data. My advice would be to use following approach:

1. stack each panel separately

2. wipe background of each panel

3. align panels to form mosaic (but don't stitch them yet)

4. do linear fit of all panels to one baseline panel (maybe central) in overlapping region

5. Then "flatten" the image to get linear stitched mosaic

[rickwayne]

Yeah, doing it on different nights really raised hob with the backgrounds. I don't think it was APP's stitching that was the problem so much as the data itself. I composited it with Astro Pixel Processor with local normalization set to 1st degree and 2 iterations, with multiblend-background set to 5% as I recall. The LNC adds a LOT of time to the processing so I didn't want to turn it up any higher than I had to.

I experimented with almost firewalling both of those and while the gradients didn't go away, they receded far enough that the portion of interest is better salvageable, and probably APP's light-pollution removal tool will be able to deal with it. Oh well, what's another nine or ten hours of processing!

Thanks for the help. I don't know why I'd forgotten about subtracting the starless from the original to leave a star layer, I used to do that. Age, I guess.

[Sunshine]

Gorgeous!

[ollypenrice]

Rather than putting your stars back in blend mode Screen you could try blend mode Lighten. That's working for me.

Olly

[rickwayne]

I have tried that previously with good results. Thanks!

BTW my redo is cooking in Astro Pixel Processor as I type. I set local normalization correction degree and iterations very high and am reaping the result...5 hours on and the progress bar is at 10%...