IKO - M81 & M82 LRGB+Ha - Processing Competition

[Laurin Dave]

Here's my second go.. this time with enhanced IFN. I did this using a masked curve. The mask for the IFN was created using Pixinsight's MLTransform (which isolated the large IFN structures) and Starnet (which removed the star residuals), the galaxies were protected using masks generated with PI's GAME script (which allows you to draw ovals).

Dave

 

Edited August 30, 2021 by Laurin Dave
Tweak to M81

[tango13]

Hi all,

this is my attempt.

Entirely processed with StarTools 1.8 alpha, with just a final touch (levels) in Gimp.

Hope you like it.

Ciao

[Spikey]

2nd go, this time I added the Ha. First time with Startools and quite impressed with it, especially for the price.

 

[Spongey]

A second attempt that's not quite as stretched, in retrospect the first version was far too bright.

 

Edited August 19, 2021 by Spongey

[billhinge]

On 15/08/2021 at 13:14, ollypenrice said:

@Grant  Pixinsight's Photometric Colour Calibration requires focal length and pixel size. Any chance of this being supplied? It would also be interesting to know the kit being used but you might have reason to withhold that information and it isn't necessary for the processing. Boy, I love the luminance!

lly

Any reason not to pull it from the image metadata, thats what I normally do?

[ollypenrice]

6 hours ago, billhinge said:

Any reason not to pull it from the image metadata, thats what I normally do?

My reason was pure ignorance! I'm only a PI dabbler. Thanks for this.

Olly

[teoria_del_big_bang]

My first attempt. still so much to learn  

 

 

Edited August 19, 2021 by teoria_del_big_bang

[The Lazy Astronomer]

Alright, so three new versions incoming (it's been very cloudy here this week)

1. Just a simple colour rebalance of my earlier attempt to make it a bit less purple:

 

2. Complete reprocess from scratch with hopefully a more accurate colour representation and improved/toned down IFN. HaLRGB composition, with both galaxies, the IFN and the stars processed independently and all layered back together. All done entirely in Startools 1.8.506 (full resolution this time - i.e. not binned):

 

3. And finally, for a bit of fun, the @vlaiv special (one for the pop art collage) 😁

Edited August 20, 2021 by The Lazy Astronomer

[bottletopburly]

7 hours ago, The Lazy Astronomer said:

Alright, so three new versions incoming (it's been very cloudy here this week)

1. Just a simple colour rebalance of my earlier attempt to make it a bit less purple:

 

2. Complete reprocess from scratch with hopefully a more accurate colour representation and improved/toned down IFN. HaLRGB composition, with both galaxies, the IFN and the stars processed independently and all layered back together. All done entirely in Startools 1.8.506 (full resolution this time - i.e. not binned):

 

3. And finally, for a bit of fun, the @vlaiv special (one for the pop art collage) 😁

Excellent processing ,

[markse68]

My beginners attempt- done in Siril and Gimp. I couldn't stretch it enough to see the ifn without it all going a bit weird and my stars are bloated as I don't know how to prevent that.

Mark

[Budgie1]

I did so much playing around with this that I can't remember the workflow! 🥴

All done in PixInsight and I think it went something like this:

• Add the Ha to the red channel.
• RGB combine
• Then add the luminance
• Mask for the stars + mask for the galaxies etc
• Play around with both until I got this:

Could I replicate it? No. 🤣

[powerlord]

Here's mine. I used Startool v8. RGBL and loaded the Ha as accents. Went though most of the workflow, playing around with it. After that, it went into starnet++ then into Affinity where I could play about with it some more, and combine the stars back. I was trying to aim for something 'natural' and clean. Pulling the reds out M81 and M82.

[MartinB]

Only recently spotted this competition and m away now for pretty much the rest of the month so this is a bit of a rush job.  This might be a blessing since my redo's rarely help the cause!  So I'm seeing the IFN as the most ethereal of chiffon scarves being languidly stroked away to reveal radiant beauty beneath.  OK I'll go and have a cup of tea but that notion is what has been in my mind in attempting to process this image.  

It's been nearly all PS for me.  I ran a deconvolution in PI but didn't like the result, it was like giving Salome acne.  I just used Starnet and Photometric colour calibration in PI.  I used the starless version  the swith the screen/inverted mask method along with some small curve tweaks to try to bring out the IFN.

 

 

 

[powerlord]

And one more (last one) - with the background IFN added back a bit.

More details about processing in Affinity:

I brought the no stars and with stars pics into affinity, and created a stars only by subtracting no stars from startools output.

I then worked on the no stars version, using affinity live filters - clarity filter on the galaxies, while masking off the rest, then combining that. I then seperated out the rest of the background and did a gussian blur on it a bit, to make those dark clouds more smooth and smoky, since the low SNR there made them quite noisy.

I boosted saturation a bit to bring out the reds mainly.

I then added the stars back, but at only around 60% opacity so that the galaxies are the main focus.

Edited August 21, 2021 by powerlord

[matt_baker]

Okay so I've spent a lot more time on this iteration, with the primary goal of conserving details in the cores without worrying too much about stretching the IFN out. 

As much as I like the IFN, I feel if it's stretched out too strongly, it just adds more noise than anything else to the image so I was subtle about it.

The definition of the core in M81 particularly I felt was much better than previous attempts, stars are better kept too and the addition of the Ha is how I like it.

Honestly with the amount of tweaking I did, I forgot the workflow but a general one would have been:

L:  Crop - DBE - Decon - TGV - MMT - ArcSin - Starnet - Curves - HDRMT - LHE

Ha: Crop - DBE - Selective subtraction with red channel using PixelMath to only highlight HII regions - TGV - MMT

RGB: Crop - DBE - PCC - SCNR - Addition of Ha using PixelMath - TGV - MMT - ArcSin - HDRMT - Combine with Luminance

HaLRGB:  Curves (saturation and colour balancing) - HDRMT - LHE - MLTSharpening - Star de-emphasis script - Export

 

I've also attached a TIFF since JPEG's can degrade the quality somewhat

 

Matt

 

v3.tif

[alacant]

Hi everyone

Thanks for posting the frames from the observatory. There's some lovely detail.

Not a fan of the FN stuff. The galaxies are so much prettier. So using only the H, red, green, and blue frames. There's a hint at the cloudy stuff, but hopefully without detracting from the beauty of the galaxies.

10 minute coffee-break processing using Ivo's latest StarTools v1.8.511; gotta love the new SV-Deconvolution module.

Thanks for looking and for the other interpretations. Love them all.

Edited August 24, 2021 by alacant

[The Lazy Astronomer]

8 hours ago, alacant said:

Hi everyone

Thanks for posting the frames from the observatory. There's some lovely detail.

Not a fan of the FN stuff. The galaxies are so much prettier. So using only the H, red, green, and blue frames. There's a hint at the cloudy stuff, but hopefully without detracting from the beauty of the galaxies.

10 minute coffee-break processing using Ivo's latest StarTools v1.8.511; gotta love the new SV-Deconvolution module.

Thanks for looking and for the other interpretations. Love them all.

Ooh l like the sense of depth in the core of Bode's 

[ollypenrice]

I thought the IFN and Bode's Loop were so good in the luminance stack that they deserved to be given priority. However, galaxy images need small stars for good comparative scale, so the processing priority was to try to haul out the Flux while going for the smallest possible stars. I used a set of custom Curves in Photoshop to lift up the IFN, de-starred the result in Starnet/Pixinsight and then used another set of custom curves for a star layer, added in Photoshop's Bend Mode Lighten.

I went easy on the addition of the Ha after stretching it very hard with a sharply rising custom curve. It was added to the red channel of the LRGB in Photoshop's Blend Mode Lighten.

Olly

 

[The Lazy Astronomer]

2 hours ago, ollypenrice said:

I thought the IFN and Bode's Loop were so good in the luminance stack that they deserved to be given priority. However, galaxy images need small stars for good comparative scale, so the processing priority was to try to haul out the Flux while going for the smallest possible stars. I used a set of custom Curves in Photoshop to lift up the IFN, de-starred the result in Starnet/Pixinsight and then used another set of custom curves for a star layer, added in Photoshop's Bend Mode Lighten.

I went easy on the addition of the Ha after stretching it very hard with a sharply rising custom curve. It was added to the red channel of the LRGB in Photoshop's Blend Mode Lighten.

Olly

 

And another one with such an interesting sense of depth in m81. Love your IFN as well, prominant but still very nice and wispy and natural looking - much better than anything I managed to do with it!

[bdlbug]

Thanks to the team at FLO and IKO for publishing the data - been a long time since I worked with LRGB_Ha data. Was fun going back into the PI and PS techniques, of which there are many.....
Took me quite a while to get the detail right with the Luminance and built it up into the RGB as HDR layers so I didnt over stretch the galaxies but also got the IFN into the mix without either being too much.

I also used a Pixelmath set of equations that are in the mastering Pixinsight publication by Rogelio Bernal Andreo to combine the Ha into LRGB - a first for me.

So here is the resulting image after my processing efforts over the past few evenings

 

Edited August 27, 2021 by bdlbug

[wimvb]

1st version (just to keep this post consistent):

Here's my attempt, easy on the IFN 

 

I think this will be my final version, more ifn

 

Edited August 28, 2021 by wimvb
more dust

[gorann]

In several of the images here I notice a curious error in the colour of a particular star situated at 8 o'clock in relation to the core of M81. Many people got it blue and in the data I found that there is a bluish halo-like area around it. But when I saw Vlaiv's version (small crop below) of the image I noticed it was yellowish, and Valiv made an effort to get the right star colours. So I looked up that star in Aladin SkyAtlas and it is HS85458, which is a F5 star so whitish-yellow. I wonder why the data easily makes us drift over towards blue on that star. Maybe it is the bluish area around it and maybe that has no connection to the star.

Cheers, Göran

Edited August 28, 2021 by gorann

[tomato]

Just when the science vs art debate on processing this data was starting to go quiet… ☺️

[wimvb]

 

1 hour ago, tomato said:

Just when the science vs art debate on processing this data was starting to go quiet… ☺️

 

Thankfully we know where the judges stand:

On 02/08/2021 at 17:11, Grant said:

It's completely subjective  but will be based on what image(s) we think do the best job of extracting the most from the data released and, look the 'prettiest' 

[vlaiv]

2 hours ago, gorann said:

I wonder why the data easily makes us drift over towards blue on that star. Maybe it is the bluish area around it and maybe that has no connection to the star.

That star is strongest in the image and it clips.

As such - it will easily get white color without any processing (max, max, max). Any color correction that comes after that is likely to boost blue more over other two components and that will produce bluish looking star.