sep.extract parameters of unconvolved data when using filter_kernel

[cosmic-nori]

I am wondering how does the sep.extract() obtain object parameters of unconvolved data
(e.g., peak and tnpix parameters) when using filter_kernel. Could you please tell me about that?

I have not seen that a detected object with tnpix parameter = 0 so far.
But I guess it is possible that such an object can be detected.
Or, if we set minarea = 1 when using filter_kernel, is the criteria also applied to the unconvolved data? Thank you.

[kbarbary]

I believe how it works is: for each object, sep.extract stores member pixel information. For each pixel, both the unconvolved and convolved values are stored. Once all the pixels belonging to the object have been recorded, peak and tnpix are determined by going through the stored pixel values and doing various statistics on them.

minarea applies to the convolved data, I believe.

Yes, it should be possible to have tnpix = 0. You might be able to construct such a case by setting minarea=1 and having an object with a few pixels just below the threshold (such that one convolved pixel is just above the threshold).

[cosmic-nori]

Thank you for the clear explanation about sep.extract.
I understand that. After that, I tried to construct the case by setting minarea=1
and low thresh value of 0.5. Finally I could find objects with tnpix=0 as you suggested.
I was able to figure things out, thanks again!