How to deal with dark holes leftover after flat calibration?

[joelshort]

I have a set of 24 light frames and I captured them in two groups with two different flats groups as well.  Unfortunately for some reason the second group, about 8 images, have a dust bunny black hole in the lights but the flats don't have it so it doesn't get corrected. 

How does one remove these black holes if the flat doesn't correct it?

[pfile]

buried in this presentation by david ault is the pixelmath mojo to create circular masks... your dust spot is not perfectly circular but you may be able to get away with creating a circular mask and then multiplying the exposed dust bunny by some small amount using pixelmath...

http://trappedphotons.com/files/Presentations/Removing%20Internal%20Reflection%20Halos.pdf

rob

[joelshort]

I tried to do something like what David Ault suggests but using a range selection to mask all but the dust bunny.  I was hoping I could do something with the individual subs before stacking but both David's method and my range selection method would need to be done after stacking.

[msmythers]

Joel

Here is another idea. At what stage you use it is up to you. I have two images A and B. B has a dark spot. I use the inellipse function in pixelmath to create a mask for the dark spot. I use convolution to soften the edges as needed. I then clone B and apply the white spot to the B clone using pixelmath. I then make some clones of the A image and of the B clone. I need more then 2 images for ImageIntegration. Using ImageIntrgration I combine the images. Using theses parameters I have rejected the white spot in the B clones while averaging the rest of the image set.



Mike

[joelshort]

Interesting Mike.  I don't quite follow all that but I'll try to figure it out.  Thanks!

[pfile]

i think if you could come up with the right pixelmath expression (similar to what david showed - essentially only multiplying by some factor within the bounds of the circle) then you could use ImageContainer to apply it to all of the subs. i can't remember if using a mask during ImageContainer would work, so you might need a "self contained" pixelmath expression that essentially includes the mask.

rob

[vicent_peris]

Hi Joel,

Not a solution for this image... But, if you use a BIG dithering pattern while acquiring the image, it can mitigate these artifacts because they will be easier to reject.


Best regards,
Vicent.

[gcardona]

You can try to adjust the Pixel Rejection parameters, but if that doesn't work, I would stack and then on the integrated image use CloneStamp to manually fix the blemish(es).

Gabe Cardona

[vicent_peris]

This could be also of help:

https://pixinsight.com/examples/NGC6914-CAHA/index.html#Multiscale_Image_Integration


V.

[mmirot]

Very clever. I wonder if it will work with halos.

[joelshort]

This could be also of help:

https://pixinsight.com/examples/NGC6914-CAHA/index.html#Multiscale_Image_Integration


V.

Regarding the multiscale image integration, from the example in the link above, how do I integrate just scales 1-2 pixesl and 4-16 pixel scale?  I don't see how to do that in ImageIntegration. 

[vicent_peris]

This could be also of help:

https://pixinsight.com/examples/NGC6914-CAHA/index.html#Multiscale_Image_Integration


V.

Regarding the multiscale image integration, from the example in the link above, how do I integrate just scales 1-2 pixesl and 4-16 pixel scale?  I don't see how to do that in ImageIntegration.

This is going to be a new functionality in ImageIntegration.

V.

[vicent_peris]

Very clever. I wonder if it will work with halos.

I don't think so... I guess those halos are always in the same place.

V.

[joelshort]

This could be also of help:

https://pixinsight.com/examples/NGC6914-CAHA/index.html#Multiscale_Image_Integration


V.

Regarding the multiscale image integration, from the example in the link above, how do I integrate just scales 1-2 pixesl and 4-16 pixel scale?  I don't see how to do that in ImageIntegration.

This is going to be a new functionality in ImageIntegration.

V.

Ah, I see.