Ngc 1097

[rogerg (Roger)]

Hi all,

I've been joying doing some good old deep sky imaging again having moved my telescope from my backyard observatory to my dark sky remote observatory. First target was NGC 1097. I've got a tone of data on it but am not really happy with the end result. Gut feel, I think the problem comes down to my soft LX200 optics. But am interested in the experts opinions here.

I have LRGB data including significant time in the RGB channel (as much in Blue as in Luminance actually!) but to be honest the RGB results look absolutely crap. The Luminance looks good. The individual RGB channels look OK to me, but the combination is absolutely awful. My usual problem of the galaxy lacking colour definition and stars bloated with colour fringing on different sides. So, this is the Luminance.

There's some good detail in the luminance channel but I'm also a bit disappointed by the lack of smooth gradient in the fainter periphery. Posterized almost. All 32 bit processing and haven't managed to attribute this to my image processing. So, perhaps just need 600s instead of 300s exposures to get more data in those softer gradients. Also wondering if I should throw away the 180s exposures and just stack the 300s exposures, perhaps the 180s are bringing down the average in these areas.

https://www.astrobin.com/cfhxrv/B/

Roger.

[gregbradley]

A lovely luminance image.

I usually do 10 minute subs as standard for almost all my images. Too little and you'll get too much colour noise as the read noise hasn't been swamped by the signal.

The faint extension arms would be nice to bring out if its there in the data. It usually takes a lot of exposure to get them though.

Greg.

[codemonkey (Lee)]

Hey Roger,

First off, let me say that the galaxy detail looks pretty good.

There does seem to be an optical issue at play given that you have different star shapes at different places in the image. Could be collimation; I have zero knowledge of that scope though.

I note the background is black clipped; I think this could be the cause of the posterization you refer to. I also note the stars look a bit bloated and also clipped... that could be capture related, but I'm suspecting processing. You may need to work on boosting the signal and increasing contrast without relying so much on black point / white point adjustments, but that's just a guess (i.e. think curves transformations and mid-tone adjustments on histogram transformation rather than the black/white points)

Hope that's of some help.

Cheers,
Lee

[ericwbenson (Eric)]

Hi Roger,

I agree with both what Greg and Lee have to say.

When I was using an C11 + ST8 combo at 0.5"/pix I found 10 minutes subs to be a minimum. If the sky was dark enough (and I suspect they are for you) 20 or 30 min subs would be even better, especially for RGB where the signal is at least a third of lum. I still do 30 min subs with the CDK20 at f/6.8.

There is a bit of collimation issue, the bottom is better than the top, but not much. What you are seeing here is the inherently small corrected field of an SCT. So you are seeing coma. You question your optics, but that is easy to check, under very good seeing take short (~1-10 sec) exposures and measure the FWHM. If it is < 1.5" then the optics are probably good enough for your site, you are seeing limited not diffraction limited. The detail visible in the inner bar and nucleus would lead me to conclude the optics are fine.

The background is slightly black clipped as Lee mentioned, and you are stretching the data pretty hard. The bright stars could be handled a bit better, but again using an SCT I always found it difficult to keep the really bright stars looking natural. I find the diffraction spikes in the RC/CDKs help distract the eye enough from the funny looking disk. The little disk in the core of bright stars is a sharpening artefact, you can use a mask on the bright stars when sharpening other details to prevent this.

Here's a link to a similar brightness/size Northern hemisphere galaxy taken with my C11 and ST8 on a G11 in my suburban Ottawa backyard back in 2006.
https://www.faintgalaxy.com/ngc4651.html
It also had 'optical jets'. Notice the stars are a bit rounder near the edges of the chip, but I had the Celestron reducer-corrector, I don't think the CCDT67 does any correction, it's only a reducer.

N.B. NGC1097 is actually a tough object to process, the range in brightness between the core and faint extensions is huge so you have done really well here.

Cheers,
EB

[rogerg (Roger)]

Thanks all, much appreciated.

Collimation - good point. I forgot to check this after moving the telescope out to the new property. Likely has been knocked out of alignment at least slightly during transit.

Exposure depth - 300s is convenient because I can do that unguided ... 600s I should do but will see if there's a guide star

Background clipped - yes, improvement there.

Stretched lots - yes. Although, almost all in curves. I find it impossible to not bloat the stars, not sure if it's processing I need to work on or just my optics. Things like the screen stretch function in PixInsight result in the same bloating.

So..... where to from here...

I'll check the collimation and see if I can do some 600s exposures. As for the RGB data, I'm at a loss of what to do with that, the results I have look so awful, and that might just be collimation and so throw-away data.

Thanks!

Roger.

[ericwbenson (Eric)]

Hi Roger,


You could software bin all your current data 2x2, it makes for a smaller final image, but the noise reduction would be very strong and the effect of non-round stars less prominent.

I always find the color data helps the strongly stretched bright star shapes since the hue variations mask the otherwise sharp edge dropoff.


Regards,

EB

[PRejto (Peter)]

Hi Roger,

I think it is neatly impossible to stretch an image without some form of star protection, usually a mask, or even a star removal process. Recently I have started to use the Histogram Transformation tool in Pixinsight - stretching only enough to not bloat the stars. I don't stretch until I like how the object of interest looks! I then make a star mask in order to continue the stretch still reserving plenty of room for later tweaks. Bottom line I think one must treat the stars as one photo and the object of interest as another.

Peter

[rogerg (Roger)]

Originally Posted by PRejto

Hi Roger,

I think it is neatly impossible to stretch an image without some form of star protection, usually a mask, or even a star removal process. Recently I have started to use the Histogram Transformation tool in Pixinsight - stretching only enough to not bloat the stars. I don't stretch until I like how the object of interest looks! I then make a star mask in order to continue the stretch still reserving plenty of room for later tweaks. Bottom line I think one must treat the stars as one photo and the object of interest as another.

Peter

Thanks Peter, fair point. I usually presume other people have some magic up their sleeve in acquiring data to not have problems with things like star bloat, but in reality I should probably allow myself more flexibility in how I process the images and do more masking of stars and the like.

[PRejto (Peter)]

Roger,

When you consider the overall dynamic range of an astrophoto you can see that any brightish star is just much brighter than your object of interest. If you brighten the object without masking the brighter stars the brighter stars are guaranteed to white clip. Yes, for sure, if exposure subs are too long clipping of bright stars cannot be reversed. The solution, as best I can tell, is to take subs of different lengths.....short for good stars and long for good object of interest.

In this photo my subs were 5 min, but what I didn't put up on Astrobin was that I also used 1 min subs for star repair:
https://www.astrobin.com/406649/C/?nc=user

Peter

[multiweb (Marc)]

Great shot Roger. Incredible details in the dust lanes and spiral arms.