Author Topic: Yet another m42 image  (Read 1870 times)

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Yet another m42 image
« on: 2018 March 07 13:22:09 »
Hi all,
I did some quick exposures of M42 (20-30 for each channel, L exposure was around 30s, used a set of shorter exposures to erase blooming with pixelmath) on my 402ME.

Starting to feel a little more comfortable with PI.

Any suggestions on improving this image? Looking at it now, it looks a bit blurry from the drizzled increase in resolution; I would downsample a bit (402ME chip is pretty small).

I was having trouble with my guider and I left it off (since the exposures were short), so I think there is a bit of movement in the stars; but overall things felt pretty smooth.  Stars had to be shrunk a fair bit, but I think it worked without too much artifact introduction.  The central stars are really hard to separate; is there a good reason why stars shouldn't be shrunk while the image is linear? (EDIT: just answered my own dumb question, you can't deconvolve if your stars don't reflect the atmospheric conditions of the acquisition).

Link to denoised L and  RGB (which was PCC'd):

https://drive.google.com/open?id=1xc4MHN6_fDUhbGMiEC7yIAH5_t5Skb_i
https://drive.google.com/open?id=1jQwHqoKSB2sCYb1MboLgMUoe0XVc9rGp



« Last Edit: 2018 March 07 22:32:57 by mmnb »

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #1 on: 2018 March 07 16:57:34 »
Tweaked contrast and boosted saturation.

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Yet another m42 image
« Reply #2 on: 2018 March 07 18:14:35 »
Hi

Here is a quick look at what I came up with. I never get to use deconvolution with my so here is my star mask and Decon settings also. The star mask was made from 2 separate star masks both having the Binarize tool applied and then Convolution after the StarMask tool. The masks were then combined with PixelMath. The rest is regular processing with HDR and so on. Like I said this is just a quick job. Didn't worry about stars or noise reduction.


Mike

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #3 on: 2018 March 07 18:49:32 »
That is much better than mine, more detail in the core.  I think I didn't pay close enough attention when applying deconvolution. Many thanks for the settings.

EDIT: Ah I remember I tried deconvoluting the RGB image without a star mask and was getting some psychedelic images.  I intended on seeing with just deconvoluting L was enough in the composite, but forgot to go back and try with a mask.
« Last Edit: 2018 March 07 20:58:13 by mmnb »

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Yet another m42 image
« Reply #4 on: 2018 March 07 20:14:22 »
I just did the deconvolution to the L as I don't see a point in doing to the RGB. Once you get more used to PI you can then start thinking about extracting the L from the RGB and combining that with the captured L for possibly more detail. First i would get much more used to PI and the tools though.


Mike

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #5 on: 2018 March 07 20:59:16 »
Ah ok, so your deconvolution for L was simply better.

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #6 on: 2018 March 07 23:41:30 »
As I try to replicate: what are the two star masks that are binarized?

Offline ChoJin

  • PixInsight Addict
  • ***
  • Posts: 106
    • View Profile
Re: Yet another m42 image
« Reply #7 on: 2018 March 08 01:25:46 »
are you sure you didnt push the denoise too much and lost details?

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Yet another m42 image
« Reply #8 on: 2018 March 08 03:25:30 »
mnnb
I created to star masks which covered the larger stars in brighter areas of the nebula and darker areas. This is the star mask for the deringing section of Deconvolution.

ChoJin
The detail was probably lost when mmnb was 'reducing' the star sizes. If no star mask or a poor mask was used that type of structure can happen easily. Here is the image with just a stretch and HDR applied with defaults. Then in the clone I apply Morphological Transformation with defaults but not through a star mask. Look familiar.


Mike

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #9 on: 2018 March 09 21:10:59 »
I'm still having trouble getting the same star mask.

The large stars are easy, but getting many of the smaller/dimmer ones without falling into noise is challenging in my hands.  In trying to pickup the small ones, I can't avoid picking up lots of stuff in the bright core.  Could you elaborate on mask construction and how you are judging it as "good"?

To be sure, the xsif files did not have their stars reduced. Just did Calibration,  Cosmetic Correction, Subframe selection, Registration, Integrate, Drizzle Integrate, Register channels, Crop/DBE/linfit, combine RGB and PCC.

Made an image with shorter exposures that same way where bright stars were not blooming, masked bloom and copied in linfit image via pixel math.  I can still see ghost noise where the bloom was, but this seems like a reasonable way to do it I think?
« Last Edit: 2018 March 09 22:34:43 by mmnb »

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Yet another m42 image
« Reply #10 on: 2018 March 09 22:17:01 »
Here is a posting I did last year that might help a little with making a star mask. https://pixinsight.com/forum/index.php?topic=10890

As far as how to tell what is a good star mask, it depends on what you are trying to do. For work on the stars like lowering brightness or masking stars during work on the background of an image I prefer very precise masks that cover just the stars and or some of the halo with the star. If I'm trying to work on color halos I might use the Color Mask script. When doing something like deconvolution I see that process more about trying to enhance the images clarity without creating other artifacts. This enhancement is a very slight improvement and only some images will benefit from deconvolution. There are many, many ways of sharpening an image in PI.

Since the clarity we are after is in the small scale areas and generally not bright objects, my opinion, objects such as stars don't benefit from Decon. Your star mask for deringing will be different then other types of star masks. The stars in the mask can be much larger then the actual star and they can be very bright in the mask. You still need the soften the edges but I like to have that happen outside the halo of the star. This helps prevent the dark rings and other odd artifacts around your stars that can happen with Decon. That small area around the stars that Decon is not working on is minor compared to having bad artifacts around stars.

One other thing, your optics seem off. I'm not an optics person so I can't say in want way they are off. I shoot mostly with a wide field camera lens and a very small refractor. This is why I rarely use Deconvolution. Dealing with the optics problems much like tracking problems will add to the complexity of processing images greatly.

Creating different types of star masks take time to learn. Sometimes it might take several attempts and even adding several masks together to get the right star mask. It all depends on what you are trying to do or protect. The same holds true for general masking also. PI gives us all the tools to create any mask we feel we need.


Mike

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #11 on: 2018 March 09 22:33:28 »
Many thanks for the thoughtful reply.

I'm a little paranoid about optics since i don't have a good way of telling if they are off vs. seeing (this is an AP 6 inch with a 402ME).  I am just about done getting automated focusing in between exposures, so that should help if it is the problem.  It could be guiding here, but (touch wood) guiding has been working pretty consistently for a while since (I am going to process a horsehead dataset next that had good guiding AFAICT).  Not too many other dials to turn that I can think of...more than happy to listen to suggestions from anyone.

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #12 on: 2018 March 14 18:08:49 »
Man. I haven't been able to replicate your mask yet (video was very helpful).

Original image, my mask, and you mask on top row (component small and big stars on bottom row).

Mostly using midtone mapping and truncation....I can't get the red circled stars in my mask without noise from the the nebula (similar brightness).  How were you able to get those guys?

Offline msmythers

  • PTeam Member
  • PixInsight Jedi
  • *****
  • Posts: 1176
    • View Profile
    • astrobin
Re: Yet another m42 image
« Reply #13 on: 2018 March 14 20:19:35 »
Hi

It looks like your using the large scale and maybe the small scale growth parameters. Try without those. I do use the compensation though. Here is what I suspect I did for one of the masks. I did not save what I did before. Every image is different and what settings work here might and probably won't be the right ones for your next image.



Mike

Offline mmnb

  • Member
  • *
  • Posts: 62
    • View Profile
Re: Yet another m42 image
« Reply #14 on: 2018 March 14 22:04:35 »
Ah, you did indeed say you were using binarize, but I thought you were talking about the option in star mask.
I was trying not to "cheat" by another processing step.  I think I should be ok then.  Really playing with the mask tool has helped me understand how to use it a bit better.