too many stars in the image

[John_Gill]

Hi,

I did some imaging last night and after pre-processing there are a billion stars covering the entire image and the nebula (mmh, sometimes we complain about too few stars).  So I now want to remove just the smallest stars without destroying everything.

I ran the script "Extract Wavelet layers" then selected Layer01 and binarized the Layer01.  Then used Pixel Math "Image - Layer01" but the result is not great.

Is there a better way and should I do this in the begining or end of the workflow?

Look up
John

[oldwexi]

Hi, have a look at my video which shows how to remove stars using the process
DefectMap. This could maybe also help.

http://www.werbeagentur.org/oldwexi/StarRemoval_DefectMap_english.mp4

Thats the basic idea, you can change it by making the starmask only for smaller stars, or/and
blending the original with the star removed/reduce image using PixelMath, 
that gives more or lesser star reduction.
PixelMath Expression i use:
(imagewithstars * 0.7) + (imagewithoutstars * 0.3)
mixes 30 % of the "withoutstars" image into the "imagewithstars" image.
etc...
Gerald

[John_Gill]

Thank you Oldwexi, I will try your method but I will need to fiddle as I would like to keep 30-50% of the stars and get rid of the small stars.

Look up
John

[msmythers]

John

Here is an example using the StarMask tool and the DefectMap tool. I also show the same mask using MT but only using an amount of 0.80. I didn't like the results at anything higher for this image.

This is extreme I think. You would also need to do some cleaning, smoothing or blending for a final image.



Mike

[John_Gill]

Thanks Mike, I will try this as well.  I had the following idea ---> During Image Integration you can tweak the Sigma High and Low to remove satellite trails, cosmic rays etc.  Would that process remove small stars or would it create other issues?

Look up
John

[wimvb]

I wouldn't try removing stars with image integration. For small and faint stars to be removed, you will need to set the rejection parameter so low that you also start clipping detail in your image. You'd probably end up with a noisy image that contains faint star residuals.
One of the methods already described here, a star mask based on few layers (less than the default 5), or MLT to isolate the first 4 - 5 layers and binarize/convolute that, should do the trick.