NGC2403 in LRGB

[Tzetze]

Happy hangover day, everyone!

I got some decent luminance data on NGC2403 on Christmas eve while the clear forecast held out. Unfortunately, some thin cloud rolled in while gathering RGB subs. Of course, once I'd completed for the night the skies cleared up again. I thought I'd make do with what I'd managed for now and will hopefully acquire fresh RGB subs at a later stage.

The RGB channels turned out very noisy but just about usable. I'm not overly happy with the colour calibration this time around and will have another go at it so am considering this to be a work in progress for now.

Star bloat is quite a problem (0.7776" image scale). I'm thinking I may be able to control this more with the use of masked stretching and elliptical masks. Plenty to improve on here.

Spent a lot of time trying to nail the deconvolution and I think it eventually paid off in this case.

Lum: 23x, 900s, 1x1

RGB: 9x ea, 225s, 2x2

10" Newt at f4.7, EQ8, Atik 460ex, Astrodon LRGB filters.

 

Edited December 26, 2019 by Tzetze

[x6gas]

I like this a lot Mike.  It's a deceptively difficult target is this...

So it is pretty noisy, unfortunately, but much of it seems to be chromatic noise.  It makes it hard to get a good background colour and the background sky looks a bit purple...  I wonder what the Lum layer looks like and wonder if you could apply a gaussian blur to the colour layer to blur out most of the noise and get the detail from the Lum layer?

Well done, though... plenty of detail in the galaxy and you've got great definition which is hard for this target.

Merry Christmas!

Ian

[Tzetze]

Thank you Ian.

I maybe should leave this image as luminance only until I get a decent set of RGB but couldn't resist trying to do something with it. Going to give the colour processing another shot tonight. I did blur the RGB before combination but certainly could have been more careful with the LRGB blending. I'll try a heavier noise reduction and take more care on the combination.

Here's a full res of how the luminance turned out.

Edit: Well, I can see now with a fresh pair of eyes that there is a faint satellite trail still present. There's some considerable noise in the lum here too which would benefit from some further reduction.

I'll have to go back to the integration stage on this..

Edited December 26, 2019 by Tzetze

[ollypenrice]

Good stuff. I think the luminance certainly out-guns the colour so more RGB data would be worth having. The background sly seems a bit 'colour busy' yet the brighter signal seems bleached out a bit. How did you combine L with RGB? I think that might be the secret.

Olly

[Tzetze]

Hi Olly.

I used the Pixinsight LRGB Combination process, leaving the lightness, saturation and chrominance noise reduction sliders at their default.

I've reworked the image some. On the RGB image, I ran through the colour calibration process again but this really only needed a very fine adjustment. I performed a much heavier noise reduction using Multiscale Linear Transform on 6 layers beginning with a threshold of 8 for 6 iterations and finishing with a threshold of 1 for 1 iteration.

After a light histogram stretch, I ran through the LRGB combination process with a heavier leaning towards both lightness and saturation. Then some further RGB/K, L and Saturation curves adjustments while keeping the background well masked.

I may tweak the curves further, but for now I'm very pleased with what I was able to get out of this with such poor RGB subs. Of course, I'm more than happy to receive any pointers on the areas I've missed. 

 

[x6gas]

OK, so I will totally defer to Olly on this one (and, in fact, every one!) but....

The Lum layer looks a tad over stretched to me.  I'd be tempted to dial that back just a bit.  Then, if you have PS, I'd be very tempted to switch to that, give the RGB a bit of a Guassian blur, add the Lum layer with blend mode luminance and see what you get.

Your most recent post looks really good but I'm convinced you have excellent data here which has a bit more to give,

Edited December 26, 2019 by x6gas
Correcting you're (you are) to your (which means your)

[ollypenrice]

45 minutes ago, Tzetze said:

Hi Olly.

I used the Pixinsight LRGB Combination process, leaving the lightness, saturation and chrominance noise reduction sliders at their default.

I've reworked the image some. On the RGB image, I ran through the colour calibration process again but this really only needed a very fine adjustment. I performed a much heavier noise reduction using Multiscale Linear Transform on 6 layers beginning with a threshold of 8 for 6 iterations and finishing with a threshold of 1 for 1 iteration.

After a light histogram stretch, I ran through the LRGB combination process with a heavier leaning towards both lightness and saturation. Then some further RGB/K, L and Saturation curves adjustments while keeping the background well masked.

I may tweak the curves further, but for now I'm very pleased with what I was able to get out of this with such poor RGB subs. Of course, I'm more than happy to receive any pointers on the areas I've missed. 

 

I don't do this in Pixinsight but in Photoshop so I can't help but...

9 minutes ago, x6gas said:

 

The Lum layer looks a tad over stretched to me.  I'd be tempted to dial that back just a bit.  Then, if you have PS, I'd be very tempted to switch to that, give the RGB a bit of a Guassian blur, add the Lum layer with blend mode luminance and see what you get.

You're most recent post looks really good but I'm convinced you have excellent data here which has a bit more to give,

... I agree with this. I often push the L too hard myself then have to go back and re-do it. In Ps I add the L to RGB in partial iterations, reducing colour noise and increasing colour saturation between each one.

I think your rework is a very big step in the right direction.

Olly

[Tzetze]

1 hour ago, x6gas said:

OK, so I will totally defer to Olly on this one (and, in fact, every one!) but....

The Lum layer looks a tad over stretched to me.  I'd be tempted to dial that back just a bit.  Then, if you have PS, I'd be very tempted to switch to that, give the RGB a bit of a Guassian blur, add the Lum layer with blend mode luminance and see what you get.

You're most recent post looks really good but I'm convinced you have excellent data here which has a bit more to give,

 

1 hour ago, ollypenrice said:

I don't do this in Pixinsight but in Photoshop so I can't help but...

... I agree with this. I often push the L too hard myself then have to go back and re-do it. In Ps I add the L to RGB in partial iterations, reducing colour noise and increasing colour saturation between each one.

I think your rework is a very big step in the right direction.

Olly

Thanks both. I'll try easing back on the lum processing for another run through. PS advice taken on board although I'll admit am reluctant to use more than one processing application. It feels like enough to be trying to get a handle on Pixinsight for now. I'll look further into what's involved with Photoshop.

[x6gas]

6 minutes ago, Tzetze said:

 

Thanks both. I'll try easing back on the lum processing for another run through. PS advice taken on board although I'll admit am reluctant to use more than one processing application. It feels like enough to be trying to get a handle on Pixinsight for now. I'll look further into what's involved with Photoshop.

Oh no.  Pi is wonderful at some things but different software does different things better than others...

[Tzetze]

I guess there's always something new to learn. I'll make it a priority.  

[wimvb]

Great image already. Regarding lrgb combination, you have to remember that white doesn't contain any colour information. If you combine white L with colour, the output will be almost white without any colour. Anywhere where you want colour, the L will need to be less than 1 (intensity scale from 0... 1). Adam Block has a good tutorial about lrgb combination in pixinsight. 

https://adamblockstudios.com/articles/Fundamentals_LRGB

Edited December 26, 2019 by wimvb

[Tzetze]

25 minutes ago, wimvb said:

Great image already. Regarding lrgb combination, you have to remember that white doesn't contain any colour information. If you combine white L with colour, the output will be almost white without any colour. Anywhere where you want colour, the L will need to be less than 1 (intensity scale from 0... 1). Adam Block has a good tutorial about lrgb combination in pixinsight. 

https://adamblockstudios.com/articles/Fundamentals_LRGB

Ahh, thank you Wim. This is valuable information. Quality link too, wasn't aware of this resource. You've made something click into place.

[Tzetze]

So, I've reworked LRGB combination and things have improved somewhat. Certainly a less washed out appearance.

[MarsG76]

Even with the hurdles to jump over, you have a very nice result.. the star bloat might have been caused only by actual seeing.

[dannybgoode]

17 hours ago, Tzetze said:

So, I've reworked LRGB combination and things have improved somewhat. Certainly a less washed out appearance.

Lovely image. I’ve made a start on this target but it’s tricky from a light polluted garden. Just need much more data and for that I need a break in the clouds!

[Tzetze]

17 hours ago, MarsG76 said:

Even with the hurdles to jump over, you have a very nice result.. the star bloat might have been caused only by actual seeing.

Thanks very much Mars. I hadn't taken into account the effects of poor seeing.  :facepalm:

[Tzetze]

2 hours ago, dannybgoode said:

Lovely image. I’ve made a start on this target but it’s tricky from a light polluted garden. Just need much more data and for that I need a break in the clouds!

Thank you Danny. All the best with your own data!