First Attempt at M-42

[Rodd]

This image was plagued by 3 wicked messengers:  moonlight, wind and clouds.  Not to mention way, way below freezing temps.  This was my first attempt at a target that contains zones of vastly different luminosities, requiring multiple exposure times and knowledge of various processing techniques.  The first I have in spades (i.e. 933 odd subs), the second has yet to be bestowed upon me--but I learned allot about PixInsight during this project.  Please let me know if you know of any method of reducing background light without jeopardizing  the edges of the nebula:  I tried morphological transformation (worked pretty good at default settings), Background Neutralization (have to be careful not to eat away at the nebula), and Automatic Background Extraction (have to be REAL careful or the pic gets destroyed.   DBE does not work--memory errors--not sure what they mean.  The Mean FWHM of many subs was too high, but I can't use subframe selector due to a screen fitting issue--can't read the lists--so I have to manually review each sub, which I did not have the patience to do with over 900.  That is why, I am sure the Trapezium is not as resolved as it should be.  Darn it.

Red: 100x1sec, 60x2sec, 60x5sec, 60x20sec, 21x60sec, 10x240sec

Green: 100x1sec, 60x2sec, 60x5sec, 60x20sec, 21x60sec, 10x240sec

Blue:  100x1sec, 60x2sec, 60x5sec, 60x20sec, 21x60sec, 10x240sec

Televue np101is, SBIG STT-8300 with self guiding filter wheel, Badder filters, Astro Physics Mach 1 GTO

 

[toxic]

very nice indeed  Rodd

[Mr Spock]

Nice looking image. I'm sure some of our processing experts will be able to help out - that side of things is a bit beyond me.

[laudropb]

Very nice image and well worth all your bother.

[Rodd]

Thanks guys.  Its a long, bumpy road.

[xtreemchaos]

great image Rodd,, processing gives me a headache, it was well worth your hard work.well done, thanks for sharing,  charl.

[wimvb]

Great image.

As for selecting frames to process; have you used blink? I generally use this for frame selection. As you have lots of frames to select from, I'd suggest take each colour at a time. In blink you can deselect frames you don't want and save those that you do want in a separate folder. I find this very convenient.

As for DBE, I've never had memory issues with this.

Usually I set sample size to 15 and number of samples per row to 10 or somewhat lower. Tolerance to a high number (usually about 2). Then generate samples. I then inspect all samples inidividually by zooming in (1:2 or 1:1) and inspecting what the samples cover. If they are over weak stars, I move them; if over nebulosity, I delete them.

After that I try to bring tolerance down to a lower value (resize the samples, don't generate new samples).

In your image, samples would be around the edges and maybe one or two between M43 and the running man. (BTW, there is still some stacking residue at the bottom of your image. This will affect the DBE process if covered by samples.)

Correction is either subtraction (most of the time) or division, with normalisation checked. I keep the background model image for inspection.

As for stretching, I never do my first stretch so far as the STF stretch, but use the "manual way". After corrections, I normally do further stretching in small steps, using either histogram stretch or curve transformation.

Background colour can be reduced by doing a colourSaturation (bringing the slider down to desaturate), using a mask. In your image that would be extremely difficult to do.

 

Hope this helps. Good luck

[Rodd]

14 hours ago, wimvb said:

Great image.

As for selecting frames to process; have you used blink? I generally use this for frame selection. As you have lots of frames to select from, I'd suggest take each colour at a time. In blink you can deselect frames you don't want and save those that you do want in a separate folder. I find this very convenient.

As for DBE, I've never had memory issues with this.

Usually I set sample size to 15 and number of samples per row to 10 or somewhat lower. Tolerance to a high number (usually about 2). Then generate samples. I then inspect all samples inidividually by zooming in (1:2 or 1:1) and inspecting what the samples cover. If they are over weak stars, I move them; if over nebulosity, I delete them.

After that I try to bring tolerance down to a lower value (resize the samples, don't generate new samples).

In your image, samples would be around the edges and maybe one or two between M43 and the running man. (BTW, there is still some stacking residue at the bottom of your image. This will affect the DBE process if covered by samples.)

Correction is either subtraction (most of the time) or division, with normalisation checked. I keep the background model image for inspection.

As for stretching, I never do my first stretch so far as the STF stretch, but use the "manual way". After corrections, I normally do further stretching in small steps, using either histogram stretch or curve transformation.

Background colour can be reduced by doing a colourSaturation (bringing the slider down to desaturate), using a mask. In your image that would be extremely difficult to do.

 

Hope this helps. Good luck

Thank you so much.  I maybe using DBE incorrectly.  When I manually try to put Xs on the image I get memory issues--If I let the software place them I don't--but if I try and move them or delete them I do.  So far in all of my computer/software troubles (and they are abundant) in the end it has been operator error--so maybe that is the case here.  It doesn't feel like that though.  I recently cleared out a bunch of space on the hard drive, and I have 16gb ram--so that should be enough to run DBE on 1 picture--I can run ABE, Background Neut, and morph. Trans. OK.  In fact--I can run everything else I have tried with no memory issues.  

[wimvb]

It seems strange that DBE should give you memory issues. It shouldn't be that memory-demanding, as far as I know. I've only once managed to bring PI down on its knees, and that was when I tested a very unrealistic scenario involving integration of 200+ DSLR images. Otherwise it is extremely stable. Have you checked on the PI forum?

As for placing samples in DBE, I have noticed that different people have different approaches. Some use lots of small samples (like ABE), and others use few but large samples. I guess the method depends on image and preference. I usually get good results when using 7 - 10 samples per row with a size of 15 - 25.

I think that for DSLR images, if you take a small sample size, the large scale noise (mottle) may influence the result, so you would need larger samples. But you seem to be using a CCD, if I'm correct. I guess that sample size has to be large enough to average out noise with in it.

[ollypenrice]

The middle to upper brightnesses of the nebula are really splendid in this image. The core could do with attention and I'll link to a tutorial for Photoshop in a sec. However, there's a large scale vertical bandng thing going on which asks for attention. The bands appear in the red channel just short of the left hand side, down the middle and right over to the right, all very broad. I'm not sure what's causing them but I'd expect a carefully selected DBE to sort them out. Don't use too many sky markers. The fewest I've used is 5 and it worked a treat. Have you had this in other images?

To blend short and long subs I really don't think you can beat this method. http://www.astropix.com/HTML/J_DIGIT/LAYMASK.HTM

There's a PI tutorial on HDR wavelets for M42 but I'm afraid I don't like the result, myself. It looks too like every other Pixinsight HDR image.

Olly

[Rodd]

1 hour ago, ollypenrice said:

The middle to upper brightnesses of the nebula are really splendid in this image. The core could do with attention and I'll link to a tutorial for Photoshop in a sec. However, there's a large scale vertical bandng thing going on which asks for attention. The bands appear in the red channel just short of the left hand side, down the middle and right over to the right, all very broad. I'm not sure what's causing them but I'd expect a carefully selected DBE to sort them out. Don't use too many sky markers. The fewest I've used is 5 and it worked a treat. Have you had this in other images?

To blend short and long subs I really don't think you can beat this method. http://www.astropix.com/HTML/J_DIGIT/LAYMASK.HTM

There's a PI tutorial on HDR wavelets for M42 but I'm afraid I don't like the result, myself. It looks too like every other Pixinsight HDR image.

Olly

Thanks Olly--I think the banding might have something to do with the conditions--Bright moon in Orion, periods of  thin, hazy clouds, and stuff gusts-combined with teh use of color saturation  The 1 and 2 sec subs have significant sky glow--or haziness.  I have a version of this pick where I used Background Neutralization and morphological transformation to almost completely remove the banding and sky glow and give a nice dark sky, but the outer edges of the nebula were eaten away as well--I'm still not sure which I like better.  Regarding PI and HDRcomp and HDR wavelet Transform (?) --that's what I used to combine the different exposure time stacks and compress the core--The trapezium is visible but the FWHM is high due to conditions and below average guiding (not to mention a period of ill focus--I can't use subframe selector--the list it produces goes beyond the limits of my screen).  I don't really know what else to do with the core.  In fact, I have 3 10 min Ha subs I was going to use but I can't figure out how to compress the core--HDRComp and transform seem to only work when combining subs of different iteration times.  When I use them on the Ha stack, the core is left as a featureless patch.   A friend used Photoshop and compressed the Ha stack core beautifully, but no one has showed me how to do it in PI--I don't have PS--only PI, AIP4WIN and Nebulosity.  I have been concentrating on PI almost exclusively to try and learn it.  There is a contingent of PI folk who are vehemently apposed to PS.  I recently suggested using both to compliment one another and it nearly caused a brawl.  The PI forum won't post pics unless all processing is done in PI.  But I am starting to believe that the 2 together would be better than either alone.  I have a problem with a monthly subscription though so not sure what I will do.

 

Rodd