NGC1333, me against you...talking about dust processing

[LorenzoM]

Hello, i use this post to present my raw data and to invite anyone to do better than me.
To this link you can download raw fit, calibrated with bias and dark.
Unfortunately i have not flats.
http://dl.dropbox.com/u/59547074/Public%20Fits/NGC1333.fit
Enjoy it and good work!!!
The frames was been acquired under dark sky using a sbig ST2000 and a William Optics FLT98 F/5; attached my version.

I would you like to use this post also to talk about acquiring and processing of subjects with dust regions, mixed with riflection nebulas.
My problem in processing this kind of subjects is related to the weakness of dust and lightness of reflection regions.
To bring out dust is often necessary to perform three or more stretch, however reflection areas and stars will be burned.
I can apply a mask and so to do more stretches in regions with only dust applying a little heavy noise reduction, but the difficult is in creating the mask because dust has not defined edges and there is the risk to create an evident gap between background and dust itself.
There are particular attentions in acquiring light frames?
So how do you procede to process dust? I would like to collect some experiences about processing.

[vicent_peris]

Hi Lorenzo,

Could you please upload your single frames? This artifact can be processed during image integration.


Regards,
Vicent.

[TeeJay]

Hello Lorenzo,

I am not very experienced with those dark Nebulas, but this is my attempt. I used a Mask from RangeSelection with this parameters:

Code: [Select]

P001.lowRange = 0.240000;
P001.highRange = 1.000000;
P001.fuzziness = 0.20;
P001.smoothness = 30.00;
No noise reduction because I don't linke the results...

CS TeeJay

[Raymond]

Hi Lorenzo
Here is my result ....

NGC1333
-------
Noise Reduction
1. MMT   layer 1: S(t=1.25,s=1.00,a=0.00)
   layer 2: S(t=0.70,s=1.00,a=0.00)
2. MorphTransf   
   morphol. median
   Iter: 3   amount: 0.35
   size 3; shape +

Stretching of image
3a. Masked Stretch
   Iter: 75
   target med: 0.080
3b. Masked Stretch
   Iter: 125
   target med: 0.080
4. HT   RGB/K
   shad: 0.0777
   midt: 0.0150

Star Mask
5a. ATWT deleting all layers above 6 (incl.) from copy of original

image
5b. HT on Mask
   shad: 0.07
   midt: 0.15
5c. ATWT deleting layer 1 (blurring)

Large Scale Mask
6a. PixelMath
   image(ngc1333) - 1.1*Star Mask
6b. CT (S-curve)
   I: 0.060   O: 0.045
   I: 0.270    O: 0.370

7. HDRMultiscaleTransf   with Star Mask applied (prot. stars)
   nb layers: 5
   x lightness mask
8. CT   with Large Scale Mask applied
   I: 0.085   O: 0.200
   I: 0.250   O: 0.430

Creation of Range Mask
9a. Lower lim: 0.50
    upper lim: 1.00   smoothn:7.50   x invert
9b. CT   I: 0.45      O: 0.25

10. ATWT with Range Mask applied
   layer 2: +0.030
   layer 3: +0.030
   layer 4: +0.040
   layer 5: +0.060
   layer 6: +0.060
   layer 7: +0.060
11. LocalHistEqual
   kern r: 26
   contr lim: 1.3
12. CT (S-curve)
   I: 0.10      O: 0.08
   I: 0.34      O: 0.40
---------------------------------------------------

and the image .....

[LorenzoM]

Thank you very much for taking up the challenge.
I will answer to every one!
First the image is just 3.5 hours, and was acquired under dark sky but with a lot of moisture so the signal is not really strong.
For this reason the processing with PI is quiet complicated, as i have just observed PI is an incredible software but just if you have a lot of signal, otherwise any tool amplifies the noise and finally it's difficult contain it with noise reduction.

Could you please upload your single frames? This artifact can be processed during image integration.

Yes i can, you can download single frames from here, frames are not calibrated so is included integrated bias and dark.
It's first time i hear image integration has an active part in processing, i am used to think it as a neutral step.
i am really courios...
http://dl.dropbox.com/u/59547074/Public%20Fits/NGC1333.rar

Thank you TeeJay for your attempt, in my opinion the result is not bad, but there is a little too noise so the very difficult step is precisely the noise reduction, so you have avoided the worst

Thank you too Raymond, i think you have a little exaggerated the the background is patchy, the image has not so much signal to allow the use of so much tools.
And stars have gone out of control. Can you post an highest resolution? jpeg compression has ruined a lot the image.
Calmly i will apply your steps, but trying without ATWT and LocalHistEqual.

[vicent_peris]

Yes i can, you can download single frames from here, frames are not calibrated so is included integrated bias and dark.
It's first time i hear image integration has an active part in processing, i am used to think it as a neutral step.
i am really courios...

Anything is neutral in image processing but our mind.

Will take a look.
Vicent.

[LorenzoM]

Ehi Vicent don't forget this post. You have thrown tha ball, now i am really curious

[Alejandro Tombolini]

Hi Lorenzo,

It is very difficult to achieve good contrast in this image.

There goes my attempt:



Details of processing

ImageCalibration, StarAlignment and ImageIntegration



Crop



DBE. First duplicate the image to see where to put the samples and then use the instance over the original image.





MaskedStretch and HT 1



MaskedStretch and HT 2



MaskedStretch and HT 3



Noise Reduction whit MMT



StarMask to protect stars and apply HDRMT twice





Range mask to noise reduction in the background



Invert the mask and ExponentialTransformation



Duplicate the image and use as mask to protect lights and apply LHE



Noise reduction once again with ATWT



Saludos, Alejandro.