Removal of undesired star reflections

[wbernhardt]

Hi,
when trying to work on my M 109 frames I discovered undesired reflections in rainbow color. The reflection - my assumption - do come from nearby star Phaecda, because hen rotating the camera by about 90°C, the observed structure also rotated and stayed with the same stars.
Now, how can I get rid of these reflections? With Pixelmath? But how?
Many thanks in advance,
Wolfgang

[GJL]

Hello Wolfgang,

Have the same problem with M 109. I think the sky isn´t clear in these days in Germany, there are some cirrostratus, transfering the light of the star.
Don´t have an idea to get rid of it, exept of Clonestamp tool. But this is a lot of work.

Gerhard

[Geoff]

Unwanted reflections are a hardware problem. Have you tried to identify any surfaces in your optical train that may be causing them? This should be the first step to take.
Geoff

[oldwexi]

Hi Wolfgang!
It depends... I assume the reflections are not at the same place in the 2 image series, therefore -
i would use the registered images and stack the one series and the other series separately.
Than i would calculate the difference with PixelMath which should left over the reflections.
Than use this difference image as an input for creating a mask with RangeSelection.
As these reflections are usually in one color channel only i would use the mask
to reduce with MMT the layers where the reflections are visible.

To give a detailed correct advice, the 2 stacks would be needed.

To use Clone Stamp is for PS ImagePainters only...

Gerald

[wbernhardt]

Thanks for all your advices.
Just one point: I only have a series of 32 frames taken at the same position, and two frames only with a different angle to see the difference (also taken the day after). The colors - as I mentioned - cover the whole rainbow space, ie it´s not in one channel only. Therefore, unfortunately I cannot subtract easily the colored noise.
I wanted to upload a picture, but it does not work although I used less than 512 KB in jpg format.
So, I wonder of how to get the stars and the galaxy separated from the background. One way I thought about was using a mask (for each channel) where I deleted all stars and the galaxy?
Of course, amending the tubus architecture would be one way, but now I have the frames...
Many thanks Wolfgang

[pfile]

hi, you can upload the fits file to google drive or dropbox or skydrive or really any web-based file host and the post a link here.

rob

[oldwexi]

O.k. Wolfgang,
so let us have the stacked FITS File. No JPG. The linear Fits file would be preferred.
To find the proper solution - if possible - the FITS file is necessary.
Gerald

[wbernhardt]

Dear all,
finally I managed to get a link for a fits-formatted picture in google drive:

https://drive.google.com/file/d/0B7rBCvEfrLd1eTlPTUhWdkt1Ync/edit?usp=sharing

Hope it works (it took quite a while to have it uploaded...)
Best regards,
Wolfgang

[GJL]

Hello Wolfgang, Gerald, Geoff

my image is looking very similar. Hope there is a good solution.

Regards   Gerhard

[Geoff]

Of course, amending the tubus architecture would be one way, but now I have the frames...
Many thanks Wolfgang

I am about to make myself unpopular.  You should not try to fix avoidable hardware problems with software.  If you have acquired some images and have found a hardware problem, then the correct method (in my opinion) is to throw away the images, fix the hardware problem and then reacquire the images.  Why work with crappy data?

[oldwexi]

Hi Wolfgang!
I have to agree with Geoff.
In your case i dont see any chance for me to remove these rays, which
cover the complete image, with SW. Usually the reflexions are in one color mainly which allows with
PixelMath to create masks for reduction of the reflections.
But in your case i am sorry this is far beyond my capabilities.
Please check your image train for reflecting surfaces.

Gerald

[wbernhardt]

Thanks, Gerald.
Its the first time I observed such reflections, but I will check the "hardware"...
Maybe I can sell or offer this picture for a little window part in a church...
Wolfgang