Lack of red in M33 spiral arms.

[oldannie]

Have only just started using Pixinsight for anything other than basic pre-processing and am about to start again with my attempt to process the Triangulum Galaxy.  Is it possible to tell from this LRGB  jpeg image whether the lack of colour other than blue is down to my data or poor processing or both!!  11x900sec exp each of red, green and blue and 16x900sec exp Ha, Chroma 3nm.  Takahashi 106 and Atik 460.

[DaveS]

There's a heck of a lot of green there. I don't have Pixinsight but I ran your JPEG through AstroArt 8 using the "Attenuate Single Colour" tool to reduce the green and got this:

This was just a quick process to try and get rid of the green. Could also do with a Gradient Reduction. I think the equivalent in PI is SCNR Green but not sure.

Your data has a lot of potential.

[City9Town0]

Yep... SCNR green gives

[oldannie]

I think something went wrong in saving this as a jpeg .  I've only just looked closely at the image I posted and there is a mass of green in which  wasn't initially there as I did use  SCNR and the result I got was very similar to the two above so sorry about that not sure what went wrong. It's almost looks as though I've added green!!  My main concern was the lack of colour in the stars in the arms.

Thanks

Annie

 

 

[JPROSS]

How are you doing the colour calibration in PI, and how are you adding the Ha data? E.g. are you just using the colour calibration process and adding Ha to the red channel or something more exotic?

I'm not sure if it's just another issue of exporting to jpg but it looks like there are stacking artefacts that need cropped on the bottom, with some funny looking (light pollution?) gradients across the image too. I don't know precisely what an uneven background would do to the colour calibration process but I can't imagine it is good. Have you run the individual channels through DBE before combining?

[oldannie]

I think confusion has set it on my part as there was no Ha!!  Just 16x900sec Luminance.  So used to doing Ha.  You're right about the need for cropping.  I took the images over three nights and don't have the benefit of plate solving . I didn't apply DBE to the RGB  until after channel combination which is what I normally do but will try applying it to the separate channels and then on the combined rgb if necessary.  I have now applied photometric colour calibration before stretching and then a saturation boost using curves but with a luminance mask so my image is less washed out - am new to Pixinsight so it's very much a work in process!!  I have some Ha rgb data of the Crescent nebula waiting to be combined but haven't worked out how to do it yet.

Here is my latest version - jpeg.  

Thanks very much for your comments.

Annie

[JPROSS]

That's coming along rather well now Funnily enough adding Ha might bring out more of the red/pink you were looking for!

I usually apply background correction to the individual channels before combining - my reasoning is that light pollution/moonlight etc gradients will change over the night so RGB and L can all have quite different backgrounds to remove. The more complicated the background gradient the harder it is to model and remove, so I think it makes sense to try and tackle the individual channels rather than trying to model the whole thing at once. You can always run ABE or DBE again on the color composite to clean up any residual colour gradients.

You probably already know about them, but when I started with Pixinsight I found the tutorials at https://www.lightvortexastronomy.com/tutorials.html very helpful.

[wimvb]

What workflow do you use? For me the following works.

Preprocessing of L, R, G, B to create 4 masters. I use one and the same sub during image registration of all subs. This means that the 4 master images are aligned identically.

• Dynamic crop of 4 masters
• RGB channel combination
• DBE on the RGB image
• Background neutralization
• Photometric Colour Calibration
• Chroma noise reduction
• Arcsinh stretch of the RGB image. (This is the best colour preserving stretch method by far. Usually SCNR Green is needed to suppress a green cast.)

Then I process the L master and combine it with LRGB combination.

There's really no need to do DBE before channel combination unless you have complicated and persistent gradients.

[oldannie]

• New Members
•  0
  •  3
• Location: Aberdeenshire

Posted 1 hour ago

That's coming along rather well now Funnily enough adding Ha might bring out more of the red/pink you were looking for!

I usually apply background correction to the individual channels before combining - my reasoning is that light pollution/moonlight etc gradients will change over the night so RGB and L can all have quite different backgrounds to remove. The more complicated the background gradient the harder it is to model and remove, so I think it makes sense to try and tackle the individual channels rather than trying to model the whole thing at once. You can always run ABE or DBE again on the color composite to clean up any residual colour gradients.

You probably already know about them, but when I started with Pixinsight I found the tutorials at https://www.lightvortexastronomy.com/tutorials.html very helpful.

 

Thanks, it's much improved I think.  I was thinking if there were enough clear nights I might be tempted to collect some Ha data! I do agree with you about using DBE on each channel for the reason you mention so will try that out.  I have looked at Light Vortex but have been using Adam Block's video tutorials on Pixinsight - Fundementals.  They are excellent and you can follow what he is doing using your own data.

[oldannie]

Having not long been using  Pixinsight I haven't yet created a basic workflow.The challenging thing about Pixinsight is that it's not formulaic and having only just started it's quite confusing. Lots of choices depending on what your data is like.   However like you I use WBPP initially together with cosmetic correction and  use the same luminance file for all subs.   I have been using DBE after channel combination but will try it before.  Have begun to use MureDenoise before DBE and before channel combination but not sure how effective it is yet.  At some point I use photometric colour calibration and SCNR.  ArcsinhStretch is also a process I have used.  I have experimented with MaskedStretch on Luminance master using cloned images and blending using PixelMath a la Adam Block.  Have yet to get to grips with masks which seems important.  I hope however that shortly I'll be able to at least create a basic workflow - hope it's as clear as yours.

Annie

[ollypenrice]

Two things stand out, firstly the dominant greens and secondly the excessive stretch. Inevitably a hard stretch of middle and high brightnesses reduces their colour intensity. I would go much easier on the stretch and preserve more colour and more contrast in the spiral structure.

I always find that a careful DBE (not too many markers, even spacing over the chip, none close to the galaxy) gets my background sky neutral. However, a dose of SCNR green is often needed to subdue the greens higher up the range.

I would not expect Ha to solve you red issue (if there is one. M33 is not very red.) What it will do is find interesting structiures in isolated HII regions in the arms.

After DBE and SCNR green I prefer to go into Photshop rather than torture myself in PI. 

Olly

Edit: Plate solving may be a minor convenience but Tom O'Donoghue and I did a 36 panel mosaic over 4 years without it. No big deal.

Edited November 18, 2021 by ollypenrice

[oldannie]

I take your point about the dominant greens .  I did apply SCNR but think I have overstretched it so back to the drawing board for that . I also think I could be a bit more careful with DBE though I do take care not to have too many markers  and to avoid stars and the galaxy itself - try, try and try again I think🙂  The Tif version is better, not so much green but when I tried to save it in Photoshop as a jpeg it would only save as jpf? which I then couldn't open on my laptop - any thoughts about that Olly?  Am afraid I've never really got on with photoshop so happy to continue to try with Pixinsight - enjoying it at the moment 🤔  Yes, plate solving has not proved to be a major inconvenience and if I were to collect some Ha data it would be for the detail not the red.  Thanks very much for your response Olly .

 

Annie

[wornish]

I think olIypenrice hit the nail on the head "there is not much Red in M33".  I had a look at some of my images of M33 and there is very little Red.

[oldannie]

I agree - live and learn 🙂

 

Annie

[ollypenrice]

On 18/11/2021 at 12:44, oldannie said:

I take your point about the dominant greens .  I did apply SCNR but think I have overstretched it so back to the drawing board for that . I also think I could be a bit more careful with DBE though I do take care not to have too many markers  and to avoid stars and the galaxy itself - try, try and try again I think🙂  The Tif version is better, not so much green but when I tried to save it in Photoshop as a jpeg it would only save as jpf? which I then couldn't open on my laptop - any thoughts about that Olly?  Am afraid I've never really got on with photoshop so happy to continue to try with Pixinsight - enjoying it at the moment 🤔  Yes, plate solving has not proved to be a major inconvenience and if I were to collect some Ha data it would be for the detail not the red.  Thanks very much for your response Olly .

 

Annie

You're not trying to save it at 16 bit? I'm pretty useless on these IT details I'm afraid.

Regarding M33 colour, it's often presented with very dominant blues in the spiral but I think an honest processing simply gives a low colour intensity pretty well everywhere on M33. I went back to mine after looking at the Hubble Team's rendition along with a conversation about galaxy colour with Vlaiv and dropped the colour significantly, particularly in blue.  

 

[oldannie]

Yes Olly I am inclined to agree with you about falling into the trap of trying a bit too hard with the colour, especially blue and I do like the way you've presented colour in your M33.  I think it's back to the drawing board with mine.  And no I did try saving as an 8 bit!

Many thanks Olly.

 

Annie