NGC4631

[JohnP]

OK as you know I spent Friday night around Chris's place doing some imaging. I wanted to compare Bromley with skies at Chris's place.

Below are a couple of screen prints of raws (top Bromley & bottom Chris place) - The imaging setup was identical i.e. ED80, Atik16ic, CLS filter, 500secs. The only difference was that there was a 24% illuminated moon at Chris's. Yes I know not a lot of difference but I think it did contribute a lot to background level as this definitely dropped as moon got lower. As you can see from the statistics analysis the average background is about 2.5 times lower at Chris's. I would say this would be close to 3X lower if moon wasn't around.

I have spent some time messing with the raws. I had a total of 8X500secs to play with from Bromley & 10X500secs from Chris's. I was actually quite suprised as after giving both my best processing effort there wasn't a lot of difference in the actual galaxy quality - the only difference was background was noisier with Bromley image although, again not as bad as I thought it was going to be. I think this all goes to show that the CLS filter I use is actually doing a reasonable job of letting me image from Bromley.

I also had a go at stacking the full set of images (i.e. 8 Bromley & 10 Chris's) & this gave best result. I have attached the processed L below. I also took about 1/2hr each of RGB when at Chris's but for some reason I am having real problems getting any decent colour. I suspect that the mistake I made was to leave the CLS filter on & this has had a big effect on the RGB channels. I may well try getting some unfiltered RGB from Bromley next time out.

Anyhow that's it for now. Comments welcome.

Cheers, John.


PS - Chris think I will be back to your place sometime soon....

[Rocket Pooch]

Nice and smooth....

Can you post the same with the darks removed from each so the camera noise is gone.

[Mac]

Interesting to note,

There is nearly a 10% differnce in the number of pixels counted,
how much would this affect the measurements?

Mac.

[Fay]

Well John, great amount of detail, I must say I think that is good!!!!!! I think we will have to have an RGB session!!!

[MarkS]

Very good image John - I love the detail and those tiny galaxies in the background.  You give an interesting analysis of background light pollution.

I agree with that analysis. 
Using a CLS filter I would get a 3-4x difference between Sidcup and deepest Kent.
With no filter I would get a 20x difference between Sidcup and deepest Kent.

So a CLS filter does do a good job.

Mark

[JohnP]

Thks all for comments:

Chris - I didn't take/ use darks. The average ADU though is around 300 - so I guess you can subtract this from both sets of subs.

Mac - I hadn't noticed that the count was different - I had just assumed it used all the pixels in the box but obviously it doesn't. Anyhow I have played around with moving the square over the image & it doesn't make any difference - regardless of where I place the box the background count difference is always approx 2.5 give or take...

Fay - Thks Fay - yep RGB is a right pain...

Mark - Good to know that your analysis agrees

Cheers - Just need to sort out colour now.

John.

[mickw]

Nice one John with good detail

[MarkS]

I also had a go at stacking the full set of images (i.e. 8 Bromley & 10 Chris's) & this gave best result.

John,

I forgot to mention that combining  two sets of subs taken under different conditions is an interesting problem, if you want to do it optimally.  In your example, a "Chris" sub has a higher S/N ratio than a "Bromley" sub because of the different background ADU counts.  Therefore intuitively we want a "Chris" sub to have a bigger contribution to the final stack.  Or to phrase the problem more precisely, we want to weight each sub in such a manner that we minimise the S/N ratio of the final stacked image.  Fortunately (and without boring you with maths) there is an easy way to do this.  If your "Bromley" subs have 2.5x the background ADU of the "Chris" subs then each "Bromley" sub should be weighted by 1/2.5 (the reciprocal of 2.5) in your final stack.

I have frequently had to do this when combining subs taken from Blacklands, Lydd, Kelling and Riberac.

Regards,

Mark

[JohnP]

I did think about whether adding noisier subs to less noisey would result in a better image - I just compared the stack of 8 from bromley, 10 from chris & the combined 18 & the 18 certainly looked the best. But what you are saying is that I should be able to get even better if I combine correctly (I am not sure how to do this in aip4win but will try). Could I use pixel math to divide each of the bromley subs by 2.5 before combining? Or could I stack the bromley subs as normal & then divide the result by 2.5 before combining with the 10 chris subs?

Thks,  John

[JohnP]

OK Mark - Thks for that. I divided the bromley subs by 2.5 before adding to Chris's subs.

Image is below - I think the main thing that has improved is the feint nebula that surrounds the galaxy - it is now a lot more prominent after doing what you suggested so many thanks for that.

I still need to sort out my colour - Pretty pleased with the L now.

Cheers,  John

[MarkS]

John,

That is looking better now - glad to be of service.

Previously you were unwittingly adding too much noise when you added the unscaled Bromley subs.

Mark

[JohnP]

Hi Mark - yep I understand that now & thks for help John

[Fay]

Don't understand  what you have done John, but really looks good!!

[JohnP]

Thks Fay. I can show you next time I see you. John

[RobertM]

Nice Capture John with lots of detail showing through.  The background looks a bit too black, it that my monitor ?