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

[Rodd]

So after all thias--you don't need background neutralization and color calibration?

[Rodd]

Should I shoot for 1.0 or is .8 better?

[gnomus]

OK so I applied that stretch and closed down the preview window and the Histogram Transformation process.  All I do now is bring up SCNR - change Protection Method to 'Minimum Neutral' and apply (in fact it even looked OK with the standard Protection Method):

Now I haven't done anything other than combine the stacks and stretch them Rodd.  I would normally do a bit of NR beforehand.  But you can see that from here it is tweaking Saturation and so forth - but gently does it.  Jings I hope that was helpful - it was a bit more work than I thought....

[gnomus]

40 minutes ago, Rodd said:

So after all thias--you don't need background neutralization and color calibration?

I would try Background Neutralisation before the stretch - it might bring the RGB closer together and you might not need to do so much clipping of individual channels.  But sometimes it doesn't always work - try it aand see - if it makes it worse undo it and go straight to Histogram.  I have not found Colour Calibration useful in NB images.

 

[Rodd]

6 minutes ago, gnomus said:

I would try Background Neutralisation - it might bring the RGB closer together and you might not need to do so much clipping of individual channels.  But sometimes it doesn't always work - try it aand see - if it makes it work undo it and go straight to Histogram.  I have not found Colour Calibration useful in NB images.

 

Well--WHeww! That was allot of work.  Thank you.  I see how the method is better.  I will start to try it.  As a last image for the 'Old" way--do you think this version is better than my previous attempts? 

 

[gnomus]

1 hour ago, Rodd said:

Should I shoot for 1.0 or is .8 better?

I think you should aim for what the eye is telling you.  Incidentally, apologies - the numbers are not 1.0 but 0.1 in the screenshots above (I am getting tired - I will correct the offending post).  Also, in the original post that did not make it clear that I get these numbers when I hover the mouse pointer over an area of background sky (I will Edit that bit too).  With my data (which probably is not as good as your 7+ hours per channel) I found that stars were starting to bloat and skies were starting to get too noisy at background levels much above 0.08.  But yours took 0.1 easily.  Please don't get hung up on these numbers - I only mention them because I tried an STF stretch and they were up over 0.2 - and looked horrible.

[gnomus]

55 minutes ago, Rodd said:

Well--WHeww! That was allot of work.  Thank you.  I see how the method is better.  I will start to try it.  As a last image for the 'Old" way--do you think this version is better than my previous attempts? 

 

The question is do you think it is better?  I think it is much better, but, to my eye, you have still overstretched (sorry if my 0.1 and 1.0 confusion misled you) and you have over-saturated things somewhat.  (Edit: Sorry, when I replied originally I thought you'd tried the new method - I didn't realise that this was still an STF stretch - that is why it's overstretched.)  Anyway I hope I have helped with the HistogramTransformation process.  You asked about reducing the star colour.  You could try a Star Mask and destaurate with that.  Or you could do what I do - go into Photoshop - bring up Filter-Noise-Reduce Noise and then reduce colour noise as so - it seems to work quite well:

  

[gnomus]

I think with STF as your 'standard' stretch, you have already completely ruined the image.  You are then embarked upon several really quite difficult tasks to try to repair the damage.  This will never be successful in my view. It is best not to mess it up in the first place.  (And I use 'mess' because the filters here will not allow me to use the word I want to use.. )

[gnomus]

32 minutes ago, Rodd said:

Well--WHeww! That was allot of work.  Thank you.  I see how the method is better.  I will start to try it.  As a last image for the 'Old" way--do you think this version is better than my previous attempts? 

 

Reproducing it here (with images) was quite labour intensive.  But doing it for real will take only a couple of minutes.  I think you need to abandon the STF as your 'go to' stretch.  You are putting yourself immediately in a position from which you can never recover - and giving yourself too much work.  I produced my processing of your data in around 30 minutes.  Obviously, if it was my data I would have spent a lot more time doing little fiddly stuff.  But your data was so good it really was a pleasure to work with - very easy.  Spending time over it should produce a really top class result.  Tiny steps though and the minute it goes wrong - take the processing back a couple of steps.  There is no point trying to fix a screw up with hours of repair work, when the screw up can be unwound so easily (often with just one 'undo' mouse click) from source.    

[Rodd]

16 minutes ago, gnomus said:

Reproducing it here (with images) was quite labour intensive.  But doing it for real will take only a couple of minutes.  I think you need to abandon the STF as your 'go to' stretch.  You are putting yourself immediately in a position from which you can never recover - and giving yourself too much work.  I produced my processing of your data in around 30 minutes.  Obviously, if it was my data I would have spent a lot more time doing little fiddly stuff.  But your data was so good it really was a pleasure to work with - very easy.  Spending time over it should produce a really top class result.  Tiny steps though and the minute it goes wrong - take the processing back a couple of steps.  There is no point trying to fix a screw up when the screw up can be unwound so easily from source.    

Thanks again--I will endeavor to never use the STF again, and I really appreciate your hard work.  I knew there was something I was missing.

[gnomus]

He he.  Now write it out 50 times; "I will never use STF again ... I will never use STF again....." 

Seriously, if I have been of help, it is my pleasure.

Good luck.

 

[whipdry]

Hi Rodd,

Hope you don't mind me processing your data, don't have PI so aligned & stretched in AstroArt, colour mapping & selectve colour in PS then I use the Ha for luminance as it's normally the smoothest.

Some like to combine the 3 sets of data for the luminance

 

Peter

[Rodd]

1 hour ago, whipdry said:

Hi Rodd,

Hope you don't mind me processing your data, don't have PI so aligned & stretched in AstroArt, colour mapping & selectve colour in PS then I use the Ha for luminance as it's normally the smoothest.

Some like to combine the 3 sets of data for the luminance

 

Peter

Yes--I am one that likes to use all 3 in a synth lum--That way all frequencies are in there.  Nice job.

[graemlourens]

Guys, i quickly go to sleep and the thread quadruples!  Honestly: Thx gnomus and Peter for chipping in. More eyes the better!

@Rodd To make things more complicated. I ALWAYS use STF for my initial histogramstretch. But i will always reduce the stretching by a far amount (but not the black point clipper). Its the same as doing it manually, but SFT is very good at deciding about the black clipping, without loosing data. Its a 5 second issue and you've got a good stretch in my opinion. It boils down to the same as @gnomus said though to never use the suggested auto STF for a stretching.

The reasoning is: STF will automatically try to use the full dynamic range (from nearly black to white in case of mono). The Problem is that it will do this to the whole image. So when you try and get even more contrast out for example the already bright nebula, there is no 'room' at the top any more to play with, and you're going to start overstretch. This is also a problem when you then try to reduce star size, as lots of stars could be oversaturated and then contour starmasks will not do you much good, as the core will not be reduce.

Tonight when i process your data (with tonemapping, you can google 'j p metsavainio's tone mapping technique' to see what it is) i will record my screen and post the video. So you (and everybody else) can see my way of processing your data. I think this could be interesting if also other people would share their whole processing run, so we could all maybe get some more insight? I like watching other peoples processes and then pick out what i would like to adapt.

Kind regards, Graem

[Rodd]

5 hours ago, graemlourens said:

Guys, i quickly go to sleep and the thread quadruples!  Honestly: Thx gnomus and Peter for chipping in. More eyes the better!

@Rodd To make things more complicated. I ALWAYS use STF for my initial histogramstretch. But i will always reduce the stretching by a far amount (but not the black point clipper). Its the same as doing it manually, but SFT is very good at deciding about the black clipping, without loosing data. Its a 5 second issue and you've got a good stretch in my opinion. It boils down to the same as @gnomus said though to never use the suggested auto STF for a stretching.

The reasoning is: STF will automatically try to use the full dynamic range (from nearly black to white in case of mono). The Problem is that it will do this to the whole image. So when you try and get even more contrast out for example the already bright nebula, there is no 'room' at the top any more to play with, and you're going to start overstretch. This is also a problem when you then try to reduce star size, as lots of stars could be oversaturated and then contour starmasks will not do you much good, as the core will not be reduce.

Tonight when i process your data (with tonemapping, you can google 'j p metsavainio's tone mapping technique' to see what it is) i will record my screen and post the video. So you (and everybody else) can see my way of processing your data. I think this could be interesting if also other people would share their whole processing run, so we could all maybe get some more insight? I like watching other peoples processes and then pick out what i would like to adapt.

Kind regards, Graem

Good morning Graem!  Yes, Gnomus took the baton and carried it a long way around the track.  I had a quick look at the tone mapping method (a tutorial).  I must say it seems like an awful lot of work.  The final image of this particular tutorial was a close in shot of part of the North American Nebula (Part of the Cygnus Wall I think).  While the method seems interesting, and rigorous, the final result did not seem very much different than images created not using tone mapping technique.  I suppose once you know how to do something, it becomes much easier and probably doesn't seem complicated--but there are allot of steps.  Do you feel its the better way?  Can't wait to see the image you come up with. 

[gnomus]

I'd be interested to see your tone mapping Graem.  I had a go at that tutorial.  I got into difficulty from the outset - with star removal.  With the suggested settings I was still getting large stars left in the image.  If I increased from the suggested baseline settings, I lost detail.  I suspect I may have overstretched my subs before trying star removal.  (it doesn't seem possible to do it with unstretched subs.). So if you are able to go over that in your video that would be great.

Steve

[gnomus]

Incidentally, Graem, your tip about using STF to set the black point seems a good one.  I assume that after applying the STH to the HistogramTransformation tool you just zoom in (as I did) and move the mid-point slider back to the right a bit until you get a preview you are happy with (again keeping a very close eye on stars and backgrounds).

[johnrt]

G'day Rodd,

I have had a play with your data and came up with this, I will also make some observations about your data.

• Firstly, your data is not junk. As pointed out, far, far from it. It is actually very good. I could have processed this in any way I wanted - zinging colours of the "Swag Palette", natural rosy tones, but I went with my own natural style - dusty, subtle and a little muted. But the lesson is none of your issues are with data capture, I can assure you of that.
• I looked at your files uploaded and I have to say you need to stop noise reducing your images before any other processing. You are wiping out fine detail and removing the natural crunch of the image. Remember - it is noise reduction not removal, every image has some noise, it's the difference between a posterised cartoon and a natural looking image. I cannot believe for a second reducing noise before you have done anything else is the best way to process. Look at the data and asses what it needs, don't process by the numbers. My upload has has no noise reduction at all. It doesn't need it.
• STF stretching - please, please, please don't use this to stretch. Don't use the Histogram tool either. Learn to stretch with curves, this is where you can craft the contrast and even star size as you stretch. STF is not the best judge of the black point, mid point or white point, your eyes are.
• Your data holds very nice detail, I was able to do huge amounts of sharpening on it, but only to the edges of structure, never the background, never the stars!
• I reused the Ha layer processed carefully as Luminance, the luminance must be as clean as possible and full of all the contrast you want to see in your final image. - adding OIII and SII is pointless unless there is specific structure you wish to retain (which is usually only ever the case in planetary nebula), otherwise you are just introducing noise and reducing contrast and detail.

To the image itself......

 

Anyway that's enough from me, I hope you find it of some use.

John.

[Rodd]

1 hour ago, johnrt said:

G'day Rodd,

I have had a play with your data and came up with this, I will also make some observations about your data.

• Firstly, your data is not junk. As pointed out, far, far from it. It is actually very good. I could have processed this in any way I wanted - zinging colours of the "Swag Palette", natural rosy tones, but I went with my own natural style - dusty, subtle and a little muted. But the lesson is none of your issues are with data capture, I can assure you of that.
• I looked at your files uploaded and I have to say you need to stop noise reducing your images before any other processing. You are wiping out fine detail and removing the natural crunch of the image. Remember - it is noise reduction not removal, every image has some noise, it's the difference between a posterised cartoon and a natural looking image. I cannot believe for a second reducing noise before you have done anything else is the best way to process. Look at the data and asses what it needs, don't process by the numbers. My upload has has no noise reduction at all. It doesn't need it.
• STF stretching - please, please, please don't use this to stretch. Don't use the Histogram tool either. Learn to stretch with curves, this is where you can craft the contrast and even star size as you stretch. STF is not the best judge of the black point, mid point or white point, your eyes are.
• Your data holds very nice detail, I was able to do huge amounts of sharpening on it, but only to the edges of structure, never the background, never the stars!
• I reused the Ha layer processed carefully as Luminance, the luminance must be as clean as possible and full of all the contrast you want to see in your final image. - adding OIII and SII is pointless unless there is specific structure you wish to retain (which is usually only ever the case in planetary nebula), otherwise you are just introducing noise and reducing contrast and detail.

To the image itself......

 

Anyway that's enough from me, I hope you find it of some use.

John.

Thanks John,  I like your dusty style.  So, you are not a fan of Mure Denoise?  That is meant to be used before anything.  I can see not doping teh other noise reduction like MSL:T with linear mask and TGVdenoise--but how about Mure Denoise?

[Rodd]

1 hour ago, johnrt said:

G'day Rodd,

I have had a play with your data and came up with this, I will also make some observations about your data.

• Firstly, your data is not junk. As pointed out, far, far from it. It is actually very good. I could have processed this in any way I wanted - zinging colours of the "Swag Palette", natural rosy tones, but I went with my own natural style - dusty, subtle and a little muted. But the lesson is none of your issues are with data capture, I can assure you of that.

Actually just took a long look at the image--WOW!! I really like it--the stars are amazing.  

[johnrt]

21 minutes ago, Rodd said:

Thanks John,  I like your dusty style.  So, you are not a fan of Mure Denoise?  That is meant to be used before anything.  I can see not doping teh other noise reduction like MSL:T with linear mask and TGVdenoise--but how about Mure Denoise?

It's not a case of being a fan of one method over another, if the data doesn't need it, don't do it. Your data doesn't need it.

 

16 minutes ago, Rodd said:

Actually just took a long look at the image--WOW!! I really like it--the stars are amazing.  

I used a version of tone mapping as described above, the stars are only from the Hydrogen Alpha applied in the luminance layer.

[Rodd]

12 minutes ago, johnrt said:

It's not a case of being a fan of one method over another, if the data doesn't need it, don't do it. Your data doesn't need it.

 

I used a version of tone mapping as described above, the stars are only from the Hydrogen Alpha applied in the luminance layer.

How about drizzle integration?  I feel this data could use it--the small stars are square and it is somewhat undersampled.  I have the Drz intergated stacks but I did not upload them because everything takes longer.  

[johnrt]

8 minutes ago, Rodd said:

How about drizzle integration?  I feel this data could use it--the small stars are square and it is somewhat undersampled.  I have the Drz intergated stacks but I did not upload them because everything takes longer.  

Are they really square? I hadn't noticed, and that is coming from someone who has just spent an hour looking closely at, and processing the data. So is someone casually viewing going to notice if they are? Not a chance.

Worrying about drizzle integration is scratching around for minimal improvements in SNR, when the basics are not yet correct. Learn to walk before you can run! You have the data at your fingertips to craft a fine image with just very simple processing steps.

[Rodd]

38 minutes ago, johnrt said:

Are they really square? I hadn't noticed, and that is coming from someone who has just spent an hour looking closely at, and processing the data. So is someone casually viewing going to notice if they are? Not a chance.

Worrying about drizzle integration is scratching around for minimal improvements in SNR, when the basics are not yet correct. Learn to walk before you can run! You have the data at your fingertips to craft a fine image with just very simple processing steps.

I believe you--I started processing without the STF--so far--I am amazed.  Hopefully I don't ruin it and will post an improved image.  You weren't kidding when you said there is a bunch of fine detail in  there-tiny details I never saw before.  I won't rush.  Thanks for all you input.  

[graemlourens]

5 hours ago, gnomus said:

Incidentally, Graem, your tip about using STF to set the black point seems a good one.  I assume that after applying the STH to the HistogramTransformation tool you just zoom in (as I did) and move the mid-point slider back to the right a bit until you get a preview you are happy with (again keeping a very close eye on stars and backgrounds).

Exactly (concerning STF) i mostly zoom to approx 20, i find that always sufficient.

I only use histogram stretching once (initial stretch, but a gentle one), and then do the rest with curves.

Hope to get my version still tonight. Was a long day, and still have work ahead of me!

Kind regards, Graem