Could pixel rejection be applied for removing minimal blooming?

Max

Hello, since I use an old but rock solid SBIG ST10XME for imaging, I often have issue about star blooming. I know there is no alghoritms in PI to remove it but I have noted that maybe it is possible to use pixel rejection to manipulate minimum blooming effects. I have tried some pixel rejection setups but without to be able for a complete removal. Anyway, in my experience, Winsorized Sigma seems to work better than GESD. Of course I had to use extreme pixel rejection values. I have upload 4 screen shots. In the 1st there is the default GESD setup while in the 2nd the suggested/default Winsorized setup: the latter seem to do best results. The 3rd image is a GESD with extreme modified values and the 4th the Winsorized one: also it seems that Winsorized gives better result with slim blooming than the GESD. Anyway better than the default values. Is there a way for complete removal manipulating these parameters? It seems I am close to the goal but I no longer have room for action.

Thanks to all

Max

Max

Adam, I am doing several attempts to try to fix this problem and applied your suggestions about using Large Scale Pixel rejection. Since my 28 shots of NGC2336 are taken before and after meridian flip, the calibrated images show the blooming Up and Down after the registration. So I thought to apply your processing technique separately to the images with Down and Up blooming. The result is very close to the goal: the blooming seems to be gone  but I am not able to remove the black contour also increasing a lot the Growth (low) parameter as you suggested (see image in attach).

Thanks again

Max

BlockHead

Max,

If you used the Clone Stamp tool... unfortunately it will not write pure zeros and this is what happens.

There are two solutions.

1. On the images that you already black on... you need to force the values that are near black to be black ("0"). You can do this with binarize as long as these values are less than your sky values (likely true).

2. Alternatively...instead of painting the blooms..you use the GAME script to draw little rectangles. This will create a mask..that you apply your your (registered) images and then apply "0's" using Pixel Math. This is the best solution and it works 100%. 

-adam

BlockHead

Your data might be an excellent example for me to demonstrate this...

Max

Thanks Adam, I will try the second option hoping to be able to do it. If you want I can send you my data: 28 calibrated-cosmetized light frames. Let me know.

BlockHead

Yes! That would be great.

If the result is clean for you... it would be a cool example. 

-adam

Max

I should send data (about 300 mega) via Mail Drop. I need you email address (mine bermassi at yahoo dot it)

Max

Adam, maybe it's better to wait for your lesson, I'm not able to reproduce your suggestions about Game script. My version is 1.6.2 and it's different respect the version of the video (it seems only Curve editing is performed...). Also I do not find the basic lesson of Game Script into "Fundamentals" but I see only the "game Update" for satellite trails.Have some difficulties to set Game preferences.

Max

BlockHead

Hmm... I believe I demonstrated GAME in QuickStart under scripts.

However, I do need to do another lesson on it- to put it all in one place.

You can also see me using it in real time if you watch PixTV Episode 1. 

My address is ngc1535

at

caelumobservatory

dot

com

Max

Hi, sent via mail drop. Many Thanks

Max

Max

Hello Adam, I continued the attempts to remove the minimal blooming from my images and now I was able to follow your suggestions. Then results seem to be quite good (see image) but there are some considerations:

- The process is very tedious to do. After I did the GAME mask I had to apply it to all 28 images and I do not know a method to automate this process

- Also I applied  Pixel Math process manually to all 28 images 

- Than I saved all 28 images one by one

I'm sure there is a procedure to automate all, a sort of Photoshop Action to save time.

Another aspect is the kind of blooming. This result was possible because the blooming was not terrible and  took only one side (UP or DOWN, it depend by pre or post meridian flip imaging); but if the star is more brilliant the blooming could appear in both side (UP and DOWN) so I think the integration could not remove the black spots after PM. What do you think?

Best

Max

BlockHead

Hi Max,

1. You can use an Image Container to automate much of this tedium. You would load the images in the image container. Then you force the mask to be applied to each image in the image container. Finally you create a pixel math expression of "0" to put zeros in all the images. This also saves all the images in an output directory. There is a trick of applying a mask all images in the image container. I will need to look that up again. Let me know if this is important. But know it is possible and it would have saved you *alot* of work!

2. You are correct that if and artifact (those pixels) appear in a majority of the images... this selective rejection method will not work (no rejection method really). 

3. To be clear... Image Integration isn't "removing" the black spots... it is averaging all the values in each pixel stack that are not rejected. If the majority of pixels are rejected... then it will not work. 

-adam

BlockHead

By the way..just looked at the image before and after... very well done sir!!

If you don't mind..I would enjoy highlighting your good efforts!

May do so on Facebook? And..would it be ok to use your name?

I just want to give you credit for your excellent use of technique.

-adam

Max

Hi Adam, tanks for your support and nice comment. Of course You can do everything you consider useful at any platform or social group. No need about credits, I only followed the Master's suggestions...

Also I will be grateful if You will demonstrate the trick of applying a mask to many images..

Max