repair stars

[Rkonrad]

I have been trying to use the "repair" algorythm specifically, "residistribute - core is average location". In both the 1.3 and most recent 1.4,the computer just hangs for hours even if only one star is selected. I have a pentium i3 with 23.5 gigs of memory.

Thanks Richard

[admin]

Hmmm...
I'm having trouble replicating this.
Are you sure you are using a star mask where the stars are selected (e.g. small green blobs) rather than de-selected (e.g. "cut out")?

[Rkonrad]

Yes If I, I'm masking out a single small star. Perhaps I can send you a unlisted YouTube video for you to see?

Richard

[admin]

Hi Richard,

Masking out the star you want rounded is very much the opposite of what you want; you are now trying to round the whole image, rather than the star, hence my question;

Are you sure you are using a star mask where the stars are selected (e.g. small green blobs) rather than de-selected (e.g. "cut out")?

Can you confirm this is what you are doing?

If you still have the problem, despite masking in the star you want to round, could you provide the image and mask?

Thanks!

[Rkonrad]

Sorry I think I didn't word that right. Here is an unlisted youtube video showing you the problem. StarTools just continues trying to heal and freezes. I'm using the latest alpha. Version 1.3 seemed to work this time! I tried it earlier but it also froze. Nevertheless here is the video (the image is not mine but Eric Coles data of cave nebula).
Thanks!

Richard

https://youtu.be/rMqpsbk0r0w

[admin]

Hi Richard,

Thank you very much for going through the trouble of uploading that YouTube video.
I can't quite explain what I'm seeing (and having trouble to replicate it). Would you be able to upload the mask and image for me (Dropbox/GoogleDrive or similar)?

Thank you!

[Rkonrad]

Thanks Ivo. These are the links:

https://www.dropbox.com/s/e0sbe592x3iod ... .tiff?dl=0
https://www.dropbox.com/s/zayk2dtxtgqur ... .tiff?dl=0

This is a different file but it happens whenever I try it.

Richard

[admin]

Hi Richard,

Many thanks again for uploading.

I have been able to replicate the issue and implemented a fix (will go into the next release).

I didn't realise how big the "blob" was you used in the mask until I got my hands on the the mask. The issue is that such big areas (contiguous blobs of pixels) are not suitable or appropriate for star rounding (you have a lot of other smaller/fainter stars caught up in the large area that extends way beyond the star). The application didn't really freeze, but was choking on the amount of calculations required for something that was meant to be quick. I have now implemented an upper limit, gradually reducing precision for such large areas, so at least the application keeps functioning.

For now, the "workaround" is to be careful with what you put in the mask, though, really the result of putting such large areas in the mask is very likely not usable in the first place.

Thanks again for uploading and creating the YouTube video! It's much appreciated.

[Rkonrad]

Thank Ivo,

I was trying to, based on my understanding of the module, to spread out the central region of the star (which is too pinpoint for my taste) to a more average location - bigger area less bright in general. Is that the correct use?

Thanks!

Richard

[admin]

Thank Ivo,

I was trying to, based on my understanding of the module, to spread out the central region of the star (which is too pinpoint for my taste) to a more average location - bigger area less bright in general. Is that the correct use?

Thanks!

Richard

The repair module is meant to gather a star's light (as indicated by the mask) and redistribute it in such a way that it radiates it evenly. For this it first removes the star's "blob" completely by substituting a plausible background, then resynthesises the star using the pixels from the blob. It does this by either re-arranging or displacing them. The "Core is average location" is meant to indicate that the new position of the star's core should be the average of all the "blob" pixels' x and y coordinates. "Core is brightest pixel" instead finds the brightest pixel in the blob and makes its location the new x and y location of the new star.

I really like the test image by the way! Nicely processed IMO.