NGC 1097 from the 'burbs

[DiscoDuck]

Tried a bit of an experiment to see if, with enough data, a decent image could be obtained under light pollution. A quick calculation suggested that, for my location, about 7x more data was needed than in a dark sky site - so about 40 hours or so should be the same as one night at a dark site.

The experiment is not conclusive without doing the same object at a dark site! But even after 40+ hours of data, there is still a lot of noise in this image and not great detail - though admittedly some nights the seeing was appalling, so may be a bit of an ambiguous outcome to the experiment.

Though I think it's cool seeing the detail brought out by stacking (see the one sub vs luminance stack comparison image).

Oh, and the 2000+ luminance frames made PixInsight (and my poor CPU) sweat a bit! :-)
Full res available on Astrobin here (https://www.astrobin.com/wvjbti), though please don't go pixel peeping as, as I said, it's a bit rubbish I think.

[Bart]

Hi Disco. There seems to be some unusual streaking in the background, very reminiscent of DSLR images. Other than that, nice galaxy! :)

[codemonkey]

Man, that's a huge effort Paul, good on ya for giving it a shot.

Even if you put in enough hours to get the SNR to be theoretically the same as a dark site, I think the data is still going to be very challenging to work with. IMO galaxies really need dark skies.

I second the above comment re the "walking noise" in the image. Did you dither these images? It looks like you didn't dither and if that's the case I think you'll find you could make quite a big improvement by doing that. Maybe something to think about for next time.

[DiscoDuck]

Thanks guys. Yes, they were dithered. Perhaps I need to dither more! Don't know if there's anything unique about this situation that makes the fixed pattern noise more obvious.

[strongmanmike]

A dedicated effort for sure and a decent result Paul :thumbsup:. I agree with Lee (not Bruce), dark skies also help because you have to do less "work" on the image too, less gradient removal (or none) etc. Dithering is very important I recon, I don't even use darks with my H694, just a 3 pixel dither between light frames, then median combine them in pre processing, it's great :)

Mike

[DiscoDuck]

Agreed that dark skies help - but in their absence was wondering what could be done! I think your comment re less "work" is on the money - took a lot of processing to get something that'd be trivial with decent data!

[Atmos]

It’s a very good test Paul, I did something similar not long ago with NGC 253, testing galaxy imaging under Bortle 7-8 skies.

My theory was that brighter objects aren’t quite as important for dark skies, whether it be bright nebula or bright galaxies. After that it largely comes down to seeing for the amount of detail that you can pick up. Your stars look great but there may be a bit too much noise reduction showing in the galaxy itself; it might be blurring some of the details but then again, not so great seeing destroys that too!

All in all, a great shot Paul :thumbsup:

[DiscoDuck]

Thanks Colin. There wasn't too much noise reduction applied to the galaxy core - I think the limited detail was just the bad seeing. Some nights were up to about 4.5" FWHM and yet I still used that data (I know!). (One night the average FWHM of some frames was well under 2", with the stars in the centre about 1.5". I've never seen seeing that good before. Shame it was only for a bit of a night!).

[Atmos]

Something you could try is having your standard luminance stack with its current FWHM and only stacking luminance subs with FWHM less than 2”.

Let’s call the full stack FS and the sharper stack SS.
After you have stretched your two luminance images to a similar level. Check the average background level (in ADU) and that’ll be your ABL.

iif(FS>(ABL/65535), FS*(1-FS)+SS*FS, FS)

What this this do is add the sharper data in on brighter areas (which intrinsically have better SNR) without affecting the dimmer areas and won’t affect the background at all.

Typing this on my phone at work but I think that code will work in PixelMath.

[DiscoDuck]

That's a great idea Colin.Will give it a go. Thanks

[Atmos]

I don’t often use it for this application but I have for increasing dynamic range in objects like M42.
120s exposures completely blow out the core so I’ll do 10s exposures just for the core. You pre-stretch the 10s stack so that it’s quite bright and then do something similar and it’ll add the non saturated data in and not touch the dimmer areas.

[alpal]

Hi Paul,
that's actually quite a good image.
The center of the galaxy is overexposed but could be improved by
only using RGB data & no Luminance.
The blotchy background can be fixed easily as per here:
http://bf-astro.com/backgndrepair.htm
I get the prize for the worst NGC1097 when
I tried 8 years ago with a DSLR -
13 frames at 5 minutes each at ISO 800
from a light polluted location.

I won't even link the pic it was so bad but it's in my Flickr photos.
:)


cheers
Allan

[DiscoDuck]

Lol. Thanks Allan.

[multiweb]

Very pretty galaxy. Love that sharp dust lane and the vivid colors. :thumbsup::thumbsup::thumbsup:

[DiscoDuck]

Thank you Marc.

[alpal]

Hi Paul,
if you are willing to link all the FITS files including darks & flats
I would love to have a go at processing your picture.
I don't have the time right now to do imaging but I do have some time for processing.
Your picture is a challenge but I think I'm up to the task. :)

cheers
Allan

[DiscoDuck]

Unfortunately Allan it is way too much data to upload ... of order 150gb I recall.

[alpal]

You mean 1.5 Giga Bytes?

[Atmos]

With more than 2000 Lum frames it would be close to 90gb of files in total

[alpal]

Of course - I forgot that it was nearly 47 hours of data.
OK then - maybe just the stacked FITS files for LRGB
already stacked and aligned &
the LRGB flats stacked results &
the dark frame stacked results for 60s & 180s.

That would be 10 FITS files if Paul was kind enough?

cheers
Allan.

[DiscoDuck]

I can have a look when I get home. But you shouldn't need darks and flats if I give you the stacked lrgb images.

[alpal]

Hi Paul - that's correct - I was just interested

in what they looked like:
the noise level of the dark frames,
the illumination of the camera sensor.



cheers
Allan

[DiscoDuck]

Allan (and anyone else who is interested). I've stuck some images here:
https://drive.google.com/drive/folders/1PqLF_ZnhNwXmEupK5Oa_DeQ3gURtM4la?u sp=sharing
(actually, they're busy uploading as I type - since I'm on cable my upload speed is sloooow! :( )



They are LRGB masters, LRGB flats and 60 and 180s darks (all the final stacked masters). The calibration frames are just FYI as the LRGB data is already corrected and stacked.


Note that the RGB masters are cropped (aligned to a cropped partly-processed version of the L master) and so are not the same geometry as the L master.


Also note that, as I said, this was an experiment to see what a lot of bad data could do. As such, I think it is essentially bad data - partly because some of it was captured at poor FWHM, but also because, as Mike and Lee suggested, there's more "work" needed on such data regardless of the theoretical SNR.



I've still to do another L stack with just the good data to get better resolution in the centre, as Colin suggested. I might put that up too when it's done.


Have fun ... and if you get a good result, please let me see (and let me know what you did to get it!).


Thanks,
Paul

[alpal]

Thanks Paul,
I've downloaded the LRGB stacks to make a start on it.
I've already managed to prevent the burn out at the center of the galaxy
by using only an RGB stack before it received its main stretching in Photoshop and a blurred layer mask.
I already pre-stretched them in NASA KITS liberator with function x^(1/5).
This will be a matter of learning what's needed for this image.
It already needed some gradient removal in Fiswork4.
Do you want us to show you only privately with a link if we think we've done well?



cheers
Allan

[DiscoDuck]

Either way is good with me Allan.

[alpal]

Hi Paul,
I need to work more on this image.
Here is a low res sample.
I certainly managed to get rid of the blotchy background -
I think it was the way you stretched it that caused most of that blotchyness -
but I also added noise to finish the background off & get it smooth -

but I don't have anywhere near the galaxy detail that you have.


cheers
Allan

[DiscoDuck]

I redid the version on Astrobin with the background done as Allan suggested and the centre of the galaxy redone with only the best subs (as Colin suggested) to give more detail.


I still think it's not very good data/bad processing ... but better than it was.


I think in future I might try longer subs as it seems that the fixed pattern noise and/or the fixed (ie was random but now frozen) noise of the master dark (which is only about 300 subs i.e. 5 hours) presents a baseline you can't stack better than, of course. Hence increasing the signal via longer exposures might work. Another experiment to try :)

[alpal]

Hi Paul,
yes as per your latest version:

https://www.astrobin.com/full/wvjbti/J/
you certainly got rid of most of that blotchy noise.
It's looking a lot better.
It might not be possible to get such faint galaxy arms showing well
from a light polluted location.
We all need to have our telescopes on a dark mountain in Chile :)
http://www.chart32.de/index.php/recent


cheers
Allan

[DiscoDuck]

Thanks Allan.

[codemonkey]

The new version on Astrobin looks much better! Well done.

[DiscoDuck]

Thanks