Page 1 of 1

Adjacent stars w/merged halos

Posted: Sun Oct 21, 2018 10:51 pm
by devonshire
Hi!

Richard's tip about using ROI (not always where you'd expect...) to influence Autodev's behavior has been most helpful to me, but there's one behavior that I can't seem to get control of. Wondering if anyone has found a way to beat this?

Consider the case of two adjacent stars, where at least one is strong and bright...

Autodev tends to expand and halo-ize prominent stars, and if the halo runs into another halo, they merge and form a blob of light with the cores of the stars visible within the blob. I can manage the degree of the problem with ROI, but haven't figured out how to stop the problem. Separate stars should stay separate...

Examples from M103 are attached. You can see that we start out with round, distinct stars in the stacked image, but post-Autodev (just taking defaults for this example), the bright ones have halos, and close, bright halos are merging (best example, upper right). I would think that the best thing to do would be to prevent Autodev from creating and merging halos in the first place, but I have no idea how to do that.

I did find a post describing use of the Magic module to shrink halos, and did test that. I can see how it would help with separate and distinct stars, but I did not see it un-merging the merged halos and restoring separate and distinct stars.

Ideas?

Thx!

- Bob

Re: Adjacent stars w/merged halos

Posted: Mon Oct 22, 2018 12:37 am
by devonshire
Oops...misspoke. Best example: Upper *Left*

Re: Adjacent stars w/merged halos

Posted: Tue Oct 23, 2018 4:19 am
by admin
Hi Bob,

The stellar profiles in your image are not really artefacts create by Autodev; they are simply attributes of your data and were formed during acquisition (through diffraction by your optical train for example).
The stars really do occupy the "halo" of others in your data. Aside from deconvolution, there is not much you can do about that.

The trick with AutoDev is to give AutoDev a good representation of the dynamic range problem you want to solve for (the "Region of Interest").

Giving AutoDev a full star (including its halo) will indeed optimize for that star. Giving AutoDev only part of the star's halo will tell AutoDev it's ok to ignore the lower part of the dynamic range that the darker part of the halo occupies. As a result you will get sharpish cut offs in dynamic range allocation, resulting in the "blobs" like the ones you posted.

Does this help?

Re: Adjacent stars w/merged halos

Posted: Tue Oct 23, 2018 2:56 pm
by devonshire
Ivo,

Thanks for the explanation of how Autodev is using ROI! I'd tried a number of different ROIs, but hadn't been able to achieve much more than differently-shaped halo blobs - or a too-sharp image with a much reduced star field.

However... I ran across a post you made re: deconvolution where you mentioned that even a small bin helped raise signal. So... I tried Load->Crop->95% bin->Wipe->Autodevelop w/ROI. The more aggressive bins are a bit strong for my taste, but the 95 was fine - and helpful.

I still have a halo, but it's much reduced and the cores of the stars are now more distinct. They do have rings inside them, but I think that's fixable downstream from this.

Results attached, w/ROI marked - thanks again!

- Bob