M51 in LRGB, a rework

[Tzetze]

This data was captured back in the Spring of this year before upgrading from an eq6 to an eq8 mount. I wasn't satisfied with my original attempts at processing so decided to try a rework of it.

I tried using the PhotometricColorCalibration process with 3 different options for white reference but in the end settled for using a more hands on approach with the ColorCalibration process.

Now that I've upgraded my mount and improved the stability of my rig, I hope to capture a fresh set of data of this target soon. Criticism and pointers towards improvement are most welcome.

Lum: 22x, 900s, 1x1

RGB: 7x ea, 225s, 2x2

10" Newt at f4.7, EQ6, Atik 460ex, Astrodon LRGB E series.

 

Edited December 15, 2019 by Tzetze
Added HDR version

[JamesF]

Love the outer luminosity.  Every time I look I see another galaxy

James

[alan potts]

Superb shot, so much to look at and for in this shot, Thanks for posting.

Alan

[carastro]

Excellent.

[lux eterna]

Not much to improve here. The background is perfect. If anything, I would desaturate the blue a bit. And maybe you could work out a little more detail in the core if it was not as bright as it is here, but overall a very nice image.

Ragnar

[GiorgioF]

An image to keep as example for both acquisition and post processing.

[Tzetze]

Thank you all for your kind comments. As mentioned, the rig used wasn't the most stable and so star shapes are not great and there is a general bluriness/softness to the image (moreso than the oversampling should produce).

4 hours ago, lux eterna said:

Not much to improve here. The background is perfect. If anything, I would desaturate the blue a bit. And maybe you could work out a little more detail in the core if it was not as bright as it is here, but overall a very nice image.

Ragnar

Thank you Ragnar. I agree on both points. In fact, the saturation has been dialled back some from where it had been during processing. Perhaps a little further desaturation would be for the best.

Galaxy core is a little overblown and star cores are not great either. I had finished a working of the processing but wasn't pleased with the result - I had bad pixels showing up in the RGB combination and blue tints throughout the background. I had linear fit all channels to the blue channel (which I think was too noisy) and ended up reworking the RGB. I registered the colour channels against the fully processed luminance and had to further crop all of the LRGB. This then meant that I wasn't able to use the linear luminance channel to produce a starmask and, after too many attempts to produce a decent mask, decided to proceed without one. Lesson learned: Corners can't be cut without losing something in this game.

Thanks for the pointers and they certainly have been taken on board but I likely won't process this target again until I gather some fresh crispy data. 

[alan4908]

Very good. I do like the colours !

In terms of points for improvements that you've asked for: on the core you might want to try using PI's HDR multiscale transform to see if you can obtain more detail.  Select an area somewhat larger than the core as a preview, then apply HDR to only the preview using the settings of your choice. When you are happy with the result, run the script substitute with preview (under utilities). Your HDR core image with then be automatically incorporated into the main image (hopefully) without a noticeable join. I used this Adam Block technique on my own effort and managed to get good core results - see my album Deep Sky III.

The other suggestion is that your red channel is not quite aligned to the other channels, you can see this quite clearly on the right hand side of the image. I presume this is due to a slight collocation error.  To correct for this in post processing - in PI's (latest release) in star alignment try selecting thin plate splines as the registration model and then check the distortion checkbox.  

Alan

 

[Tzetze]

Thank you Alan. Those are some great pointers to work on. I skipped the HDR process completely as applying to a range mask, while improving the core, was not improving the whole image. Now that I think of it in hindsight, this was a wide range mask, if I were to use a narrower range mask (with subtracted starmask) I should get better results. However, I do like your tip on applying to preview selection only. I wasn't familiar with that script, sounds very useful indeed.

I hadn't noticed the red misalignment and to be honest I still can't see it but this is either due to a slight red/green blindness on my part or I don't know what I'm looking for.   I'll use your recommended settings in future though. Thanks again, that's very helpful.

[Tzetze]

I worked a HDR process into the image, albeit at a very late stage and added to the original post. I would rather add the HDR to just the luminance layer and soften the effect in blending during LRGB combination. Again, a decent starmask would have smoothed out the core better than this result.

[alan4908]

16 hours ago, Tzetze said:

I hadn't noticed the red misalignment and to be honest I still can't see it but this is either due to a slight red/green blindness on my part or I don't know what I'm looking for.   I'll use your recommended settings in future though. Thanks again, that's very helpful.

I find the best way to inspect collimation errors is to pick an individual star and then zoom in on it to inspect it in detail. What you should see is all the colours aligning, in your case the red channel is slightly misaligned.  To explain a little further, if I select a star at the right hand edge of the frame I get the image below. The star should be blue but you have a red misalignment effect as shown below in this highly magnified image. Hope this helps !

Alan

 

[Tzetze]

On 16/12/2019 at 08:35, alan4908 said:

I find the best way to inspect collimation errors is to pick an individual star and then zoom in on it to inspect it in detail. What you should see is all the colours aligning, in your case the red channel is slightly misaligned.  To explain a little further, if I select a star at the right hand edge of the frame I get the image below. The star should be blue but you have a red misalignment effect as shown below in this highly magnified image. Hope this helps !

Alan

Apologies for late reply, very busy time at work for a while.

This does indeed help a lot. I lately did some work on my collimation and thought I had improved things quite a bit but it looks like the misalignment is still present.

Maybe not quite as pronounced as in the M51 image but this is something to work on. I wonder if this is something that can be further improved with collimation. I guess this comes down to optics and OTA quality. I'm curious to see how much improvement can be made through pre-processing though. Roll on Christmas and some free time!

[ollypenrice]

I think the image is very good indeed. Colour and depth are there already. I can't help thinking you could get slightly sharper small-scale detail, though. But this is great.

Olly

[Stub Mandrel]

Very nice 🙂

[Tzetze]

On 18/12/2019 at 14:28, ollypenrice said:

I think the image is very good indeed. Colour and depth are there already. I can't help thinking you could get slightly sharper small-scale detail, though. But this is great.

Olly

Thanks Olly, I appreciate your input. I'm trying to get to grips with deconvolution. It's not proving to be easy, to say the least, but will keep plugging away at it.

On 18/12/2019 at 21:29, Stub Mandrel said:

Very nice 🙂

Thank you Neil.

[tony210]

Great image - I find deconvolution difficult and often end up with 20 images which I have to go back to a few days later to choose the one that's both showing more detail and aesthetically the most pleasing(they are certainly not the same thing).- Tony.

[dannybgoode]

Lovely image - a level of quality to aspire to   
 

Quick question - how do you rescale the L shot 1x1 and the RGB shot 2x2. Do you resize the L accordingly or is there another method? 

Edited December 24, 2019 by dannybgoode

[Tzetze]

1 hour ago, dannybgoode said:

Lovely image - a level of quality to aspire to   
 

Quick question - how do you rescale the L shot 1x1 and the RGB shot 2x2. Do you resize the L accordingly or is there another method? 

Thank you, Danny.

I can only speak for Pixinsight processing. Once separate L, R, G and B images have been stacked they need to be registered to each other using the StarAlignment process. A reference image is selected which the other images will be aligned/registered to. If the reference image is 1x1 and the other images are 2x2, the process will automatically upscale the 2x2s to match the reference image.

2 hours ago, tony210 said:

Great image - I find deconvolution difficult and often end up with 20 images which I have to go back to a few days later to choose the one that's both showing more detail and aesthetically the most pleasing(they are certainly not the same thing).- Tony.

Thanks Tony. That sounds much like my own experience with deconvolution. Raccoon eyes and noise are very troublesome to control.

Edited December 24, 2019 by Tzetze

[dannybgoode]

@Tzetze - thanks for the information. I haven’t delved into the murky world of Pixinsight but I’m going to investigate it