Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Processing challenge--fix my M51!

  • Please log in to reply
50 replies to this topic

#1 Stelios

Stelios

    Fly Me to the Moon

  • *****
  • Moderators
  • topic starter
  • Posts: 6878
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 15 May 2019 - 03:46 PM

For anyone willing to help in this challenge, rather than starting with the raw masters, I thought I would give *as a start* the initial stretched LRGBHa image that I have:

 

M51uncooked.JPG     Link here to full FIT image (152MB)

 

This image has had *no* non-linear processing done, but has a fair amount of linear processing (Lum: DBE, Deconvolution, MLT denoise. Chrom: DBE and MLT denoise each channel, combine, Background Neutralization, CC then LRGB combine). It was stretched by application of STF and HT.

 

My initial processing effort resulted in *this* overcooked monstrosity:

 

M51overcooked.JPG

 

...which, however, illustrates that there's plenty of detail in the image. The trick that I hope you will be able to accomplish is to capture some of that detail *without* losing the faint nebulosity in the original.

 

If someone does process this file, it would be hugely appreciated if you would give the steps and basic parameters (for example, if you do HDRMT, how many iterations, how many layers, how much overdrive, did you use median transform or not?)

 

Thanks in advance flowerred.gif  to anyone who tries this! 

 

---------------
And for those who feel they *must* work with the original data, here's the original linear masters in a zip:

 

Link to zipped masters (217MB) (no DBE or ABE applied--note that the Ha master will look much better with just ABE)

 

Please let me know if the links work. I usually link from Dropbox, but this time I tried to link from OneDrive. Hope it worked! 


  • Jim Waters likes this

#2 TOMDEY

TOMDEY

    Soyuz

  • -----
  • Posts: 3546
  • Joined: 10 Feb 2014
  • Loc: Springwater, NY

Posted 15 May 2019 - 04:45 PM

Step right up; step right up! >>>    Tom

Attached Thumbnails

  • 78 whitewash Tom Sawyer.jpg

  • artem2 likes this

#3 bobzeq25

bobzeq25

    Hubble

  • *****
  • Posts: 15432
  • Joined: 27 Oct 2014

Posted 15 May 2019 - 04:56 PM

For anyone willing to help in this challenge, rather than starting with the raw masters, I thought I would give *as a start* the initial stretched LRGBHa image that I have:

 

attachicon.gif M51uncooked.JPG    Link here to full FIT image (152MB)

 

This image has had *no* non-linear processing done, but has a fair amount of linear processing (Lum: DBE, Deconvolution, MLT denoise. Chrom: DBE and MLT denoise each channel, combine, Background Neutralization, CC then LRGB combine). It was stretched by application of STF and HT.

 

My initial processing effort resulted in *this* overcooked monstrosity:

 

attachicon.gif M51overcooked.JPG

 

...which, however, illustrates that there's plenty of detail in the image. The trick that I hope you will be able to accomplish is to capture some of that detail *without* losing the faint nebulosity in the original.

 

If someone does process this file, it would be hugely appreciated if you would give the steps and basic parameters (for example, if you do HDRMT, how many iterations, how many layers, how much overdrive, did you use median transform or not?)

 

Thanks in advance flowerred.gif  to anyone who tries this! 

 

---------------
And for those who feel they *must* work with the original data, here's the original linear masters in a zip:

 

Link to zipped masters (217MB) (no DBE or ABE applied--note that the Ha master will look much better with just ABE)

 

Please let me know if the links work. I usually link from Dropbox, but this time I tried to link from OneDrive. Hope it worked! 

The best thing to upload is the unstretched stack.  Just unstretching this some is, in my eyes, a big improvement.

 

I personally have no need to stack (just busywork), but I want to stretch.

 

ChromCC.jpg

 

 


Edited by bobzeq25, 15 May 2019 - 04:59 PM.

  • elmiko, artem2 and Bretw01 like this

#4 Gipht

Gipht

    Surveyor 1

  • *****
  • Posts: 1611
  • Joined: 12 Nov 2016
  • Loc: Prescott Valley, AZ.

Posted 15 May 2019 - 05:27 PM

Thanks for letting me practice on your nice data.  May have made it worse!

Attached Thumbnails

  • Capture.JPG

  • artem2 likes this

#5 pfile

pfile

    Aurora

  • -----
  • Posts: 4872
  • Joined: 14 Jun 2009

Posted 15 May 2019 - 06:23 PM

here's a try at it; agree with bob that a linear file would have been better...

 

xpsm attached.

 

ChromCC_DBE_clone.jpg

 

 

Attached File  stelios.xpsm.zip   41.22KB   2 downloads


  • Stelios, artem2 and 42itous1 like this

#6 kathyastro

kathyastro

    Mariner 2

  • *****
  • Posts: 234
  • Joined: 23 Dec 2016
  • Loc: Nova Scotia

Posted 15 May 2019 - 07:02 PM

Mostly what it needed was unstretching.  I used Curves (multiple iterations), followed by sharpening with ATrous Wavelets, and then ACDNR.

ChromCC.jpg

 

I find that the STF / HT stretch seldom gives good results.  It is fine for a quick & dirty stretch for something like a star mask.  But stretching by hand with just HT usually gives me better results.


Edited by kathyastro, 15 May 2019 - 07:21 PM.

  • Stelios, John Miele, nofxrx and 20 others like this

#7 Joe G

Joe G

    Apollo

  • *****
  • Posts: 1481
  • Joined: 10 Jan 2007
  • Loc: Southern California

Posted 15 May 2019 - 07:35 PM

Stelios,

 

Some processing in PS.  Adjusted levels.  Selective Saturation adjustments.  Selective noise reduction.  A little bit of Astronomy tools local contrast enhancement and some unsharp sharpening.

 

Above it seems that you just pushed the sharpening and contrast too far.

 

Attached Thumbnails

  • M51 Stelios.jpg

  • Stelios, artem2 and 42itous1 like this

#8 f300v10

f300v10

    Vostok 1

  • -----
  • Posts: 183
  • Joined: 16 Sep 2018

Posted 15 May 2019 - 07:38 PM

I gave it a quick go in PhotoShop:

 

 

Attached Thumbnails

  • ChromCC-St-PS-sm.jpg

  • Stelios, artem2 and 42itous1 like this

#9 Stelios

Stelios

    Fly Me to the Moon

  • *****
  • Moderators
  • topic starter
  • Posts: 6878
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 15 May 2019 - 09:22 PM

Thanks to all who tried, I like your versions. If I had to pick frontrunners, it would be pfile and Joe G.--mainly because I prefer the brighter versions. 

 

I somehow failed to save the linear version *just* before the stretching! (I usually save a million files, but not this one). I will try to recreate it and hopefully post sometime tomorrow (unfortunately I have other things to take care of :))

 

More versions welcome, or you can wait for the linear file.



#10 Lite2

Lite2

    Messenger

  • -----
  • Posts: 489
  • Joined: 18 Jan 2017
  • Loc: USA, Indiana

Posted 15 May 2019 - 09:38 PM

Thanks for letting me practice on your nice data.  May have made it worse!

Feeling green? me too....



#11 pfile

pfile

    Aurora

  • -----
  • Posts: 4872
  • Joined: 14 Jun 2009

Posted 15 May 2019 - 09:47 PM

i think kathy did a good job recovering the contrast. i did use a little local histogram equalization but mostly left it alone.

 

rob


  • 42itous1 and kathyastro like this

#12 terry59

terry59

    Cosmos

  • *****
  • Posts: 8763
  • Joined: 18 Jul 2011
  • Loc: Colorado, USA

Posted 15 May 2019 - 11:38 PM

Thanks for making your data available Stelios

 

waytogo.gif

Attached Thumbnails

  • Image35.jpg

Edited by terry59, 15 May 2019 - 11:55 PM.

  • Stelios, elmiko, artem2 and 1 other like this

#13 Mert

Mert

    Fly Me to the Moon

  • *****
  • Posts: 5802
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 16 May 2019 - 12:23 PM

That is beautiful data, I took your masters into APP,

did combination in the tools menu and into Pix.

Didn't do much more but sharpening a tiny bit and

reduce noise ( ACDNR 1.7 ) there.

Attached Thumbnails

  • current-combine-RGB-tool-image-CN.jpg

Edited by Mert, 16 May 2019 - 12:23 PM.

  • Stelios, elmiko, artem2 and 1 other like this

#14 Bretw01

Bretw01

    Ranger 4

  • *****
  • Posts: 365
  • Joined: 13 Feb 2017
  • Loc: IL USA

Posted 16 May 2019 - 12:34 PM

Thank you Stelios for posting, it is a very nice image.

 

I used the masters; DBE on each - RGB combination using pixelmath - combined RGB with Ha using NBRGB script.

Noise reduction using TGV and MMT.

Initial stretch with arcsinh, followed with ACDNR.

More stretching done with HT - three passes followed with MMT for noise reduction.

A little star reduction done with morphology and a mask created with MMT.

 

Deconvolution on the LumMaster followed up with same process done on HaRGB file (minus arcsinhstretch).

A little sharpening with MLT. 

LocalHistogramEqualization at 14%.

LRGB combination.

Photoshop - vibrance layers used to remove color cast in background. Lab color to bring out the colors in the galaxy.

 

NBRGBCombination_PIPS.jpg


  • Stelios, BlueGrass, elmiko and 7 others like this

#15 Mert

Mert

    Fly Me to the Moon

  • *****
  • Posts: 5802
  • Joined: 31 Aug 2005
  • Loc: Spain, Pamplona

Posted 16 May 2019 - 01:57 PM

And a second version, maybe too much

 

Attached Thumbnails

  • current-combine-RGB-tool-image-V2-CN.jpg

  • Stelios and artem2 like this

#16 jerahian

jerahian

    Mariner 2

  • *****
  • Posts: 246
  • Joined: 02 Aug 2018
  • Loc: Maine

Posted 16 May 2019 - 02:01 PM

Very nice data, Stelios.  Thanks for sharing.  Here is what I did to achieve the following image:

 

stelios M51
 

I hope you like it.  To reproduce, I performed the following steps:

  1. TGVDenoise
  2. ABE
  3. ArcsinhStretch
  4. CurvesTransformation (a couple of times) - with fuzzy range mask
  5. Extract CIE L* - denoise & stretch for good dark and lightness
  6. ChannelExtraction - RGB channels
  7. LRGB combine with new extracted and processed L
  8. MMT - with fuzzy range mask
  9. CurvesTransformation (final touch-up) - with fuzzy range mask

I have attached each process' code to this post.  I did not include the code for the processing of the extracted L channel.

 

Attached File  stelios_m51_process.txt   7.47KB   8 downloads

 

-Ara


  • pfile, artem2, Bretw01 and 1 other like this

#17 OldManSky

OldManSky

    Viking 1

  • -----
  • Posts: 817
  • Joined: 03 Jan 2019
  • Loc: Valley Center, CA USA

Posted 16 May 2019 - 07:13 PM

Fascinating what a multitude of variations came out of this...:)


  • dswtan and Scott Mitchell like this

#18 sink45ny

sink45ny

    Surveyor 1

  • *****
  • Posts: 1584
  • Joined: 08 May 2014
  • Loc: Pennsyltucky

Posted 16 May 2019 - 08:57 PM

ChromCC

  • artem2 likes this

#19 evan9162

evan9162

    Surveyor 1

  • *****
  • Posts: 1569
  • Joined: 20 Sep 2013
  • Loc: ID

Posted 16 May 2019 - 09:10 PM

Being overly stretched, the highlights are very compressed and lose a lot of contrast.  More difficult to preserve the luminosity gradient to the core while showing more dim areas (like the star plumes)

 

ProcesS:

- A few curves adjusments to de-stretch

- MLT for NR in L and Chroma

- ArcsinhStretch

- Curves for some contrast

- HDRMultiscaleTransform (which flattens the tonality a bit too much in the core, but that's a tradeoff I suppose)

- More curves

 

 

m51proc-2.jpg


Edited by evan9162, 16 May 2019 - 09:16 PM.

  • Stelios, artem2 and 42itous1 like this

#20 Stelios

Stelios

    Fly Me to the Moon

  • *****
  • Moderators
  • topic starter
  • Posts: 6878
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 16 May 2019 - 09:20 PM

Thank you Stelios for posting, it is a very nice image.

 

I used the masters; DBE on each - RGB combination using pixelmath - combined RGB with Ha using NBRGB script.

Noise reduction using TGV and MMT.

Initial stretch with arcsinh, followed with ACDNR.

More stretching done with HT - three passes followed with MMT for noise reduction.

A little star reduction done with morphology and a mask created with MMT.

 

Deconvolution on the LumMaster followed up with same process done on HaRGB file (minus arcsinhstretch).

A little sharpening with MLT. 

LocalHistogramEqualization at 14%.

LRGB combination.

Photoshop - vibrance layers used to remove color cast in background. Lab color to bring out the colors in the galaxy.

 

attachicon.gif NBRGBCombination_PIPS.jpg

I totally like this version--I think possibly the arcsinh stretch did better than my STF-HT attempt. You seem to have preserved the faint luminosity while bringing out much of the detail. 

 

Just to be clear--for luminance, you stretched first, then used MLT and LHE. Was the stretch STF-HT? 

 

Also, LHE "at 14%" --what does that mean? LHE has kernel radius, Contrast Limit and Amount. 



#21 Stelios

Stelios

    Fly Me to the Moon

  • *****
  • Moderators
  • topic starter
  • Posts: 6878
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 16 May 2019 - 09:35 PM

I continue to be amazed at the variety of the versions! Thank you all very much for your interesting, beautiful, and educational attempts!



#22 Bretw01

Bretw01

    Ranger 4

  • *****
  • Posts: 365
  • Joined: 13 Feb 2017
  • Loc: IL USA

Posted 16 May 2019 - 09:36 PM

I totally like this version--I think possibly the arcsinh stretch did better than my STF-HT attempt. You seem to have preserved the faint luminosity while bringing out much of the detail. 

 

Just to be clear--for luminance, you stretched first, then used MLT and LHE. Was the stretch STF-HT? 

 

Also, LHE "at 14%" --what does that mean? LHE has kernel radius, Contrast Limit and Amount. 

Luminance - deconvolution first, followed by NR using TGV and MMT. Then a small stretch with HT followed with ACDNR, then two or three more small stretches with HT.

 

After all stretching was done, LHE was applied. Kernal radius and contrast limits at default, amount 0.14. This means that LHE is blended at 14% with the original at 86%. I do this with the preview, and adjust the percentage until I like what I see. 

 

Lastly a little sharpening With MLT.


  • 2ghouls likes this

#23 Kevin Ross

Kevin Ross

    Mariner 2

  • -----
  • Posts: 241
  • Joined: 28 Mar 2018
  • Loc: Traverse City, MI

Posted 16 May 2019 - 10:40 PM

Since everyone else is using PI, thought I'd offer a processed version using APP and PS, just for something different. smile.gif

 

It's a little red, maybe I added too much Ha to the red channel.

 

M51-combine-RGB-image-lpc-cbg-St.jpg


  • artem2 likes this

#24 evan9162

evan9162

    Surveyor 1

  • *****
  • Posts: 1569
  • Joined: 20 Sep 2013
  • Loc: ID

Posted 16 May 2019 - 11:48 PM

And here I started from scratch with the masters (details upon request)

 

 

 

m51proc-4.jpg


  • Stelios and artem2 like this

#25 Traveler

Traveler

    Mercury-Atlas

  • *****
  • Posts: 2852
  • Joined: 19 Aug 2007
  • Loc: The Netherlands

Posted 16 May 2019 - 11:49 PM

Very nice thread. Like Kathy's version the most (till now ;-)).

But boy, what a difference can make different processteps...what is right?


  • kathyastro likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics