[AstroFest] A flirt with Andromeda

[Camelopardalis]

Folks,

I caught the eye of Princess Andromeda for a few nights up at AstroFest and also managed to catch all the nasty atmospheric conditions along with it. This is about half the data I captured, the rest is worse :sadeyes:

Anyhow, I'm not wild about this, so wanted to put it out there for suggestions. What's the best way to handle stars? This image was processed without a star mask, and as there's also a bout of unsavoury gradient from one of the nights contributing to this data set. I've tried playing around in DBE and can't really get it levelled out :help: Consequently, I've not stretched it as much as I would like too. Maybe it just needs more and/or better data?

Any/all tips and tricks appreciated :D

Cheers,
Dunk

[Rod771]

Not much wrong with that, Dunk. What don't you like about the stars?

Sometimes I'll take shorter exposures for the star cores only, then combine that data with longer exposed data using the HDR Composition tool. Helps retain core data/colour when stretching.

[multiweb]

Hey Dunk, that looks awesome! The contrast is great. I can't see anything wrong with the stars. I hear you about the gradients and seeing. I had to dump a lot of subs too.

[AstroJason]

Yeah Dunk, like the guys have said, this looks great mate. Have you tried one of Carboni's gradient removal tools in PS? Might be worth a shot.

[Rex]

Nothing wrong that I can see either Dunk. Great image. Dbe should be able to remove the gradient. In case u don't know, you can apply DBE more than once if it doesn't come clean after the first go just run it again while the sample points are in placee. I asume u ensured the sample points weren't ontop of any stars? Takes a bit of patience but is definately worth the effort to check each sample point before running it. Hope that helps.

[Camelopardalis]

Thanks Rod! The stars just seem a bit bloated at 100% because of the stretching, I just haven't found a star mask method yet that doesn't leave pandas eyes :sadeyes: Good tip on the HDR, thanks for that :thumbsup:



Thanks Marc! Yeah I knew it would be a challenge being low and all, I guess we were lucky to have so many clear nights, even if they weren't perfect.



Thanks Jason! I hadn't tried it but I'll give it a shot :thumbsup:



Thanks Rex! I guess I'm just pixel peeping, watching them grow with the drag of the slider :eyepop: I didn't think to run DBE multiple times, but I will try it again, thanks for the tip. Yeah I usually zoom in and check the sample points aren't on stars, nebulosity or galactic fuzz...especially tricky on this one as the fuzz was corner to corner.

[Retrograde]

Looks very smooth and detailed to my (admittedly less than expert) eyes Dunk. Very nice.

Compares well to my memories of the old Mt Palomar shots which first captured my imagination back when I was a teenager.

[Rod771]

I follow Juan's steps detailed in his post on the Pixinsight forum to produce a star mask Click here (http://pixinsight.com/forum/index.php?topic=6001.msg40719#msg40 719)

If you find that some of the larger stars are not include in the generated mask you can use Range selection to select the larger stars, add some softness to the selected regions with its "softness" slider. Then use Pixel Math to add the Range mask and Star mask together to create your final star mask.

Works for me. :thumbsup:

[multiweb]

Good stuff. Thanks for the link. Bookmarked. :thumbsup:

[Camelopardalis]

Here's a quick repro I did with a little more DBE...

[Camelopardalis]

Thanks Pete! Glad to hear I'm heading in the right direction :) It's always hard to know what objects should really look like.



Great stuff Rod - thanks for that, I'll give it a go :thumbsup: the stars are noticeably tighter in the raw files than the tickled result.

[multiweb]

I'm jealous mate. You've got heaps of contrast in there. :thumbsup: I'm gonna repro mine and try that DBE (hasta la vista gradients) thingy too.

[strongmanmike]

Hmmm that DBE seems quite effective here ...I always wonder if it is always accurate though, say when gradients are being removed from areas full of complex and faint galactic cirrus...? (and no I am not bagging PI :rolleyes:). I guess once you have its nuances and settings right and know what it can and can't do (like Decon/Wavelettes :scared3:) it's probably very powerful?

Mike

[Brisbaneskys]

Nice Image. Here something you can try when using DBE. Make a clone of the image, then over stretch the clone image to see the gradient better. This will allow you to plot your point, more accurately. create a process icon then activate your main image. For the correction select "Division", this will flatten the image, also remove the color cast. Now select "subtraction" to remove/reduce the gradient. Subtraction can be used several time without degrading the image.
DBE is not a replacement for flats and won't get rid of imperfections like dust mots etc. But it does a pretty good job with gradients.

Robin

[Camelopardalis]

Thanks Marc - between the two of us we must have some decent data :D you're welcome to add it to your subs, although it's only DSLR data ;)



Thanks Robin - I'll give it a try! It's quite a complex gradient, as Andromeda was scraping along the northern horizon, so it's the right hand side of the image that is the worst.

Makes me wonder if there is a way to remove the gradient in preprocessing :question: I might be able to get away with rescuing a few more subs from the junk pile then.

[MortonH]

I like the first version best. Admittedly I'm on my mobile just now so probably not seeing all the nuances. Very nice!

[Rod771]

No worries Guys :thumbsup:

[AG Hybrid]

They look fine. Stop whining Englishman.

[AstroJason]

haha I actually agree with Morton, I prefer the original Dunk, it just looks more natural. When you compare the two. The repro seems to have bloated the stars a little, seemed to reveal more noise and revealed a few dust bunnies or something towards the top of the image. That said, it has improved contrast and enhanced the fainter regions of the galaxy. If you never posted the original and just shared the repro I would still think it was a great image! That scope is killer!

[Camelopardalis]

Thanks Morton!



Thanks Adrian! You miss the whining Englishman really :lol:



Thanks Jason! This is still a work in progress as I'm trying to rescue some additional subs, but it'll give me the opportunity to revisit the "look and feel" :D

[RobF]

Flirting with pretty princesses can be damned frustrating at processing time unfortunately Dunk. We were lucky to get so many fog free'ish nights but M31 is still below 25 degrees at AF most of the time which is a lot of atmosphere to deal with.

For galaxy processing this one is a bit like the neb problems with M42. Large dynamic range, but also lots of stars to add to the headaches. You've got so many great suggestions there. All I could add would be to consider playing with:

- consider collecting a bit more blue data in the field to compensate for the low altitude (think of how orange the moon gets when down low). You've managed to get the blue star clouds anyway with whatever you did...
- initial Masked Stretch to help keep the stars down
- Many of the intensity transforms in Pixinsight on faint DSOs are best used with star masks as you've noted - things like median transforms, LAHE, unsharp mask etc.
- Later you can invert the mask and work on the stars a little - e.g. colour balance, deconvolution if so inclined, or morphological transform to erode them down a little. This can certainly make the galaxy pop, but also risks looking "fake" - there has been a representative spread of personal opinions already in your thread here whether star minimisation is good or not :)
- DBE: results can vary a lot depending on how samples are selected - for broad gradients across a field with a diffuse DSO in the middle less points can be better. Rather than auto-generating sample points, try putting points around the outskirts of the field but avoid going too near the galaxy. If you can see any outline of the galaxy in the correction frame you need to try something different. As others have said, don't be afraid to do a few passes to get it right. First effort might only have 6-10 sample points around the outskirts.
- Colour balance - you need decent flat frames if shooting RGB otherwise its so easy to skew the colour balance between the bright centre and faint outer regions using DBE etc. Linear Fit seems to help even out separate RGB frames (not sure what camera you were using - ignore if OSC)

Still think you've great a great image there. Lots more to celebrate than worry about :thumbsup:

[RickS]

Looks pretty good, Dunk. I imaged this puppy from Leyburn once and it was a PITA to process. Just too much atmosphere in the way...

Cheers,
Rick.

[Camelopardalis]

Thanks Rob! She sure is a tricksy Princess :sadeyes:



Thanks Rick! Sadly I wasn't into imaging before flipping hemispheres, so I'll take what I can get, but yeah it's certainly been a learning curve.

[Camelopardalis]

Many thanks to everyone for their comments and tips, much appreciated :thumbsup: it's certainly been a good learning experience.

I've taken another look at the data and rescued another hour of data, not at the sharp end of the subs but at least/maybe contributing a bit more colour. Going through the repro, I hope I somewhat neutralised the gradient and I was able to tease out the Ha regions a little more than before. There's a bit of a blue cast to the background that seems to disappear if I clip the black end a bit, but I don't want to do that.

For now, at least, this is my final go :D it's a little more colourful, a little more detailed and the stars are a little sharper, even though it's probably impossible to tell with the file size limit and compression :shrug:

Cheers,
Dunk