Sh2-132 Data for Processing--Please show me what teh data has to offer

[Rodd]

I have been over this data to the breaking point and I just can't get it to look decent.  Please, someone feel free to process the included calibrated FITs files and let me know whether the data is capable of being processed into a respectable image.  If so--I will keep plugging away.  But if not, I will need to get a better handle on acquisition.  My big problem is it is not repeatable.  The attached JPEG is the best I have been able to do with 3 colors.  But after 50 attempts, only 1 came out like this.

Images

1) Calibrated Ha

2) Calibrated OIII

3) Calibrated SII

4) Combined-Non linear but no non linear processing

5) Fist non linear step--star reduction

6) My best final result.

Ha.fit

OIII.fit

SII.fit

Start Non Linear.fit

Fist nonlinear step Star Reduction.fit

[Rodd]

Note the very different look compared to many of the bicolored images and Hubble Pallete images posted.  What am I doing wrong?

[Rodd]

Here is a crop of the head for easier inspection.

[graemlourens]

Hi Rodd.

Happy to chip in also in this thread but not as a competition, more to just see if maybe i can help you out to get to a point where you can be happy with your image

First thing i saw (before even starting to process) is that you have clamping going on on your OIII master. The staralignment process (when choosing wrong algorithm or clamping threshold) can produce those black pixel artefacts around the images. 

What staralignment algorithm did you use?

And just for clarification: those masters have already got NR applied or are those the raw master stacks? (looks extremely smooth to me )

Update: Trying to get a LUM master is very tricky as you're OIII-data seems 'strange', something is wrong with the stars. It even looks as if the stars came from binned data. You didn't bin 2x2 anything here in this data right?

Kind regards, Graem

[Rodd]

21 minutes ago, graemlourens said:

Hi Rodd.

Happy to chip in also in this thread but not as a competition, more to just see if maybe i can help you out to get to a point where you can be happy with your image

First thing i saw (before even starting to process) is that you have clamping going on on your OIII master. The staralignment process (when choosing wrong algorithm or clamping threshold) can produce those black pixel artefacts around the images. 

What staralignment algorithm did you use?

And just for clarification: those masters have already got NR applied or are those the raw master stacks? (looks extremely smooth to me )

Kind regards, Graem

Wow--thanks, I must take a closer look after alignment.  Never knew what clamping was for.    I used Auto (lancos-3) and yes--there has been noise reduction in the linear state Mure Denoise, and 2 wavelet layers removed using MSLT with a linear mask, and a touch of TGVdenoise with mask.  I am pleased with the smoothness prior to any non linear processing.  Also--remember, each stack has 7-9 hours of data in 30 min subs--so noise is low anyway.  I figured if I could make them as noise free as possible it could only help

[graemlourens]

3 minutes ago, Rodd said:

Wow--thanks, I must take a closer look after alignment.  Never knew what clamping was for.    I used Auto (lancos-3) and yes--there has been noise reduction in the linear state Mure Denoise, and 2 wavelet layers removed using MSLT with a linear mask, and a touch of TGVdenoise with mask.  I am pleased with the smoothness prior to any non linear processing.  Also--remember, each stack has 7-9 hours of data in 30 min subs--so noise is low anyway.  I figured if I could make them as noise free as possible it could only help

Strange. So if you didn't bin anything, then its strange to me why the auto algorithm should cause clamping if HA & SII were done with the same settings.

Try to sort that out so that the stars in the OIII-Master do not have this clamping, but then i'd be happy if you could post the raw masters RIGHT after stacking. No NR done at all, nothing. Only like that i can effectively play with the data and try and get out whats there, knowing i have the fully raw data.

I would suggest you to always use 'Blink' to look through the data after staraligning. Especially when you're handling binned data (where you are 'upsampling' to 1x1) the algorithm is a crucial decision as this can cause bad artefacts on the resulting data.

Kind regards, Graem

[Rodd]

43 minutes ago, graemlourens said:

Strange. So if you didn't bin anything, then its strange to me why the auto algorithm should cause clamping if HA & SII were done with the same settings.

Try to sort that out so that the stars in the OIII-Master do not have this clamping, but then i'd be happy if you could post the raw masters RIGHT after stacking. No NR done at all, nothing. Only like that i can effectively play with the data and try and get out whats there, knowing i have the fully raw data.

I would suggest you to always use 'Blink' to look through the data after staraligning. Especially when you're handling binned data (where you are 'upsampling' to 1x1) the algorithm is a crucial decision as this can cause bad artefacts on the resulting data.

Kind regards, Graem

I cannot upload files when I quote a response.  So right now the FITs files are uploading and I will attach them and submit a response in about 20 minutes.  Just to let you know as you will not be notified of a quote since I can't quote and upload.  I reintegrated the OIII and there was no clamping issue--starange is right.  Then again, I have been over this data so many times that I am literally swimming in files--dozens and dozens, so it is possible that I uploaded the wrong one before.  Anyway--I appreciate the help.  Lately my posts are not getting very much response, so I REALLY appreciate your assistance.  Give me 20-----Rodd

[graemlourens]

No worries Rodd.

I like challenges  But you'll have to be patient until tomorrow probably evening. I think though that we'll continue discussions on processing in private messages, so we can keep the thread clear for other people to post their versions.

Kind regards, Graem

[Rodd]

Graem--I can't post images or files in these threaeds when I quote a resonse.  Here are the files.  I included a mure denoised stack for each filter--I love MD and I think it works wonders with the right settings.  Let me know if you think I am using the right settings (take a look and tell me if the mure denoised images are cleaner).  The ones with no noise reduction are straight out of calibration-aligned-integrated.  Then I ran MD.  I think the MD ones are superior.

 

1 Ha Fits no noise reduction

2 Ha Fits with Mure Denoise Script

3 OIII no noise reduction

4) OIII Mure denoise script

5) SII no noise reduction

6) SII Mure denoise script

Ha No noise reduction.fit

Ha Mure Denoise.fit

OIII no noise Reduction.fit

OIII Mure Denoise.fit

SII No noise reduction.fit

SII Mure Denoise.fit

[graemlourens]

Raws are great this time! you must have done some funky staralignment with the previous OIII version

I'll experiment tomorrow evening and post what i can cook up (btw i'll be using the tonemapping approach)

Kind regards, Graem

[gnomus]

Hi Rodd

I hope you don't mind me pitching in on this one, now that Graem is involved.  But I downloaded your data just to see what it was like.  It seemed pretty fabulous to me.  I've only just started having a play at Hubble Palette myself and I have a lot to learn.  I don't know what sort of result you were looking for, but I was able to get something I was happy with pretty quickly.  This was done mostly in PixInsight but I went into Photoshop at the end for one or two procedures.  I don't know the object so am not sure what I should be shooting for.  

Anyway, for what it is worth, here is my version.  One could ramp up the blues and oranges differently, of course [Edit: Incidentally, I started with your Mure Denoised files]:

[gnomus]

This may be a bit wilder, if you prefer that sort of thing:

[Rodd]

33 minutes ago, gnomus said:

Hi Rodd

I hope you don't mind me pitching in on this one, now that Graem is involved.  But I downloaded your data just to see what it was like.  It seemed pretty fabulous to me.  I've only just started having a play at Hubble Palette myself and I have a lot to learn.  I don't know what sort of result you were looking for, but I was able to get something I was happy with pretty quickly.  This was done mostly in PixInsight but I went into Photoshop at the end for one or two procedures.  I don't know the object so am not sure what I should be shooting for.  

Anyway, for what it is worth, here is my version.  One could ramp up the blues and oranges differently, of course [Edit: Incidentally, I started with your Mure Denoised files]:

Great image. After Graems insight on alignment artifacts, I am starting to get something  more like yours--not quite there yet.  I don't understand what it is I am doing wrong--I go through the accepted PI workflow--only 6-7 tools and I get something different.  

[Rodd]

1 hour ago, graemlourens said:

Raws are great this time! you must have done some funky staralignment with the previous OIII version

I'll experiment tomorrow evening and post what i can cook up (btw i'll be using the tonemapping approach)

Kind regards, Graem

What is tone mapping?

[gnomus]

Just now, Rodd said:

Great image. After Graems insight on alignment artifacts, I am starting to get something  more like yours--not quite there yet.  I don't understand what it is I am doing wrong--I go through the accepted PI workflow--only 6-7 tools and I get something different.  

Thanks Rodd

In looking at your first image what struck me was the size of the star top right - it was quite large -  and the edges of the nebula - quite grainy.  It seems to me that these are tell-tale signs of overstretching.  I do this too and am trying to rein it in.  I have just had to accept that I am going to 'leave something in there' if you know what I mean.   Once I compiled your image and did a restrained stretch - watching all the time for star bloat, then I was left with a very green image (from all the Ha, of course).  A quick blast of SCNR with the 'Protection Method' set to Minimum Neutral got me very close.  After that it was just tweaking really.  I hope that helps - but I am aware that I am just a beginner.     

[Rodd]

All sounds so familiar--everything but the results of course.  Thanks for you help.  At least the data is not garbage.

[gnomus]

The data is far from garbage.  I am increasingly coming to the view that the 'art' is primarily in the stretch.  Get this wrong and it's an uphill (and losing) battle from then on.  Keep a constant eye on the background and the stars.  Stop well before things start to get 'messy' and accept that you have probably found the point at which the data is going to give all that it is going to give.  I think it is better to end up with something pleasing, even if you were not quite able to bring out the faint puff of gas in the corner that Sir Arthur Streeb-Greebling was able to conjure in his masterpiece: maybe Sir Arthur had better skies, better gear, better skills or (just) more subs than you do.  

[Rodd]

37 minutes ago, gnomus said:

The data is far from garbage.  I am increasingly coming to the view that the 'art' is primarily in the stretch.  Get this wrong and it's an uphill (and losing) battle from then on.  Keep a constant eye on the background and the stars.  Stop well before things start to get 'messy' and accept that you have probably found the point at which the data is going to give all that it is going to give.  I think it is better to end up with something pleasing, even if you were not quite able to bring out the faint puff of gas in the corner that Sir Arthur Streeb-Greebling was able to conjure in his masterpiece: maybe Sir Arthur had better skies, better gear, better skills or (just) more subs than you do.  

Everyone says not to use the STF and histogram transfer in PI for the stretch.  But I can never do better than what the STF can do--never.  Everytime I try, it degrades the image.  Maybe that is my problem--the stretch.

[Rodd]

41 minutes ago, gnomus said:

The data is far from garbage.  I am increasingly coming to the view that the 'art' is primarily in the stretch.  Get this wrong and it's an uphill (and losing) battle from then on.  Keep a constant eye on the background and the stars.  Stop well before things start to get 'messy' and accept that you have probably found the point at which the data is going to give all that it is going to give.  I think it is better to end up with something pleasing, even if you were not quite able to bring out the faint puff of gas in the corner that Sir Arthur Streeb-Greebling was able to conjure in his masterpiece: maybe Sir Arthur had better skies, better gear, better skills or (just) more subs than you do.  

How did you get the tiny tiny stars in the very center of the lions head to be white?  I can't even get them to show up on  a star mask so I can't protect them-or desaturate them

[gnomus]

2 minutes ago, Rodd said:

Everyone says not to use the STF and histogram transfer in PI for the stretch.  But I can never do better than what the STF can do--never.  Everytime I try, it degrades the image.  Maybe that is my problem--the stretch.

I think that we may have had this discussion before.  I find the STF stretch horrible and way over the top.  If that is what you are doing, then I think you may have found the issue.  What do you do when you do this stretch?   Give me a few minutes,  Rodd and I will post some screen shots ....

[gnomus]

OK - this is my STF of your data - it's revolting - (more screenshots to follow)

[gnomus]

I'm about to start stretching your image - note the two numbers in the boxes I've highlighted.

Here I have started the stretch - I am watching what I am doing in real time preview.  I have not applied the stretch yet - I am going to zoom in on the curve.... (more photos to follow)

[gnomus]

OK.  I have increased the zoom factor in the highlighted boxes by moving my mouse over the lower window and scrolling my mouse wheel (or you can use the arrows in the windows).  Now I can see what I am doing and I have finer control over my middle slider - I have not yet touched the black point slider and I will never touch the white point slider (more photos coming):

 

 

[gnomus]

I have applied the stretch shown above and reset the sliders.  Now I am going to balance up the RGB histograms.  Note the numbers in the right hand boxes.  You can zoom them too.  I will normally just use the arrow key to zoom the horizontal part of the graph.   I now click on the R, G or B button to adjust the R, G or B curve and try to line all the histogram peaks up.  here I have already lined up the Red and Green by bringing the Red black point slider in and I am now doing the same for the Blue:

Once that is done, I switch back to RGB/K and bring in the black point slider until I just start to clip the data (clipping out 1 pixel is certainly OK).  (More to come, but we're nearly there)

 

 

[gnomus]

OK - so I applied all those changes (by hitting the square) and the image got a bit darker, so I reset everything again and hit the magnifying glass symbol to reset every zoom level back to 1.  I now do my final stretch.  I am looking at the stars and the background.  Note along the bottom, when I hover the mouse pointer over an area of background, I have luminescence values for RGB around the 0.1 mark.   That seemed to work with your data - with mine I often have to stop around the 0.08 mark.  But I think the main guide should be visual.  Nearly there...