NGC2244 Reprocess with RGB stars (PI)

[CraigT82]

I have spent an absolute age redoing my Rosette image from a couple of weeks ago, this time processing completely in PI rather than APP and adding 1 more hour of 03.

I also replaced the stars in this narrowband image with RGB stars using starmask and pixelmath following a YouTube tutorial.  I only had 5x30s for each of the RGB channels so not really much of an improvement to the image overall, but the whole exercise has helped me learn quite a bit more about PI so worthwhile so the same. 

This is 5x1200s Ha and 5x1200s O3, with 5x30s each for RGB for the stars.

200/1200 newt, atik 383l mono. Baader 7nm NB and RGB filters.

(First image is the posted original)

 

 

[tooth_dr]

Very very nice Craig.  Great detail and contrast in the globules.

[Adreneline]

Very interesting Craig and well implemented. I know just how difficult it is to combine RGB stars with a NB image and make it look natural and not look as if the stars have been stuck on as an after thought.

I spent a lot of time recently doing exactly the same thing with my NB image of IC405 & IC410 and a lot of time debating with myself whether it was the right thing to do in the first place!

I was completely undecided at the end whether it was the right thing to do and I can't decide whether to do it again on another NB image. I will be really interested to see what reaction you get to your efforts and resulting image.

Thanks for sharing.

Adrian

[vlaiv]

I actually prefer nebulosity in first version. Like that you added RGB color to the stars - that is very nice touch.

Not sure what process you followed - but StarNet++ for star removal is good way to do it - make starless version, subtract from original to have stars only version - apply RGB color to that version (to Ha channel - stars tend to be tight in Ha). Combine starless versions into color image and then blend in stars with RGB color.

In any case - first rendition of nebulosity has more 3D feel to it - second one came out flat. That can happen if you push data too much. I understand that additional data gives you incentive to push it further - but sometimes it does not work as expected (as we tend to over do it). Here is example of what I'm talking about:

vs

Although signal in second is stronger - it just looks flat and lacks structure shown in above one.

Or perhaps this:

vs

 

[CraigT82]

9 hours ago, Adreneline said:

Very interesting Craig and well implemented. I know just how difficult it is to combine RGB stars with a NB image and make it look natural and not look as if the stars have been stuck on as an after thought.

I spent a lot of time recently doing exactly the same thing with my NB image of IC405 & IC410 and a lot of time debating with myself whether it was the right thing to do in the first place!

I was completely undecided at the end whether it was the right thing to do and I can't decide whether to do it again on another NB image. I will be really interested to see what reaction you get to your efforts and resulting image.

Thanks for sharing.

Adrian

Thanks for your comments Adrian, your Flaming Star and Tadpoles widefield is a corker 👍. So far im not too sure on the methodology of adding the RGB stars to NB images, there is always more than one way to skin a cat and I don't think I used the right method considering the very limited RGB data I had to play with, but bat this stage I'm just trying to learn as much as I can about DSO processing. It is waaaaay more involved and time consuming than planetary/lunar processing!

[CraigT82]

9 hours ago, vlaiv said:

I actually prefer nebulosity in first version. Like that you added RGB color to the stars - that is very nice touch.

Not sure what process you followed - but StarNet++ for star removal is good way to do it - make starless version, subtract from original to have stars only version - apply RGB color to that version (to Ha channel - stars tend to be tight in Ha). Combine starless versions into color image and then blend in stars with RGB color.

In any case - first rendition of nebulosity has more 3D feel to it - second one came out flat. That can happen if you push data too much. I understand that additional data gives you incentive to push it further - but sometimes it does not work as expected (as we tend to over do it). Here is example of what I'm talking about:

vs

Although signal in second is stronger - it just looks flat and lacks structure shown in above one.

Or perhaps this:

vs

 

Thanks Vlaiv.

I spotted that difference in the nebulosity myself after I exported the new image from PI and compared to the original, but at that point I had spent so much time on it I couldn't face going back to re-do it 😆.  I will keep coming back to this one and see if I can make a perfect blend of the defined nebulosity of the first with the RGB stars of the second. 

[Adreneline]

3 hours ago, CraigT82 said:

So far im not too sure on the methodology of adding the RGB stars to NB images, there is always more than one way to skin a cat and I don't think I used the right method considering the very limited RGB data I had to play with, but bat this stage I'm just trying to learn as much as I can about DSO processing.

I'm not sure the method I used is the right way but in outline this is what I did:

• Star aligned, dynamic cropped and background extracted the NB and RGB masters
• Combined the SHO data and carried out a partial stretch before removing the stars using Starnet
• Completed the stretch of the NB starless image, colour saturation, noise reduction, colour balance, etc.
• Duplicate the RGB image and use Arcsinh stretch very conservatively on one copy and knee-bend curve stretch on the other in Affinity.
• I then experimented by blending the SHO starless and one or other of the RGB masters using either PixelMath or Affinity using Lighten or Screen blend.
• Finally a little bit of colour balance and noise reduction in Affinity and Dfine.

As you say there are no doubt numerous permutations on the above plus other processes that could be applied. I quite like the colours Arcsinh stretch provides but it is too easy to push it too far and get what I feel is then a very pretty but unnatural look to the star colours. The knee-bend curve in Affinity lifts the star colours conservatively whilst at the same time pulling out an RGB image of the nebula which I have found can both add and detract from the image. My image of IC405 combined the SHO with the Arcsinh RGB image using a Lighten formula in PixelMath.

Having said and done all the above I'm still not convinced it is beneficial overall to put RGB stars into an NB image but that's just me over-thinking the problem.

Good luck!

Adrian