Deconvolution producing "doughnut" in galaxy core

[magnusl]

Hi!

I'm processing an image of M81, and I'm having problems with deconvolution and the galaxy core, and to a smaller degree with the brighter stars. What happens is the the core becomes a "doughnut", with a small dark centrum. This then causes problems in later processing, with Masked stretch and so on. The attached jpg is the very center of the galaxy core after a deconvolution with global dark 0.0120, 60 iterations, a star mask as local deringing. Overall this deconvolution looks good, except for this effect on the core and the brightest stars. Any ideas on what I might be doing wrong here?
The full xisf.image is found here.

https://www.dropbox.com/sh/fs744ujp8oed79d/AACfnztnt-E98N8km8Tv2dUBa?dl=0

It is quite noisy, I don't know if that matters - I live in a lightpolluted area and it is taken with a Canon EOS 350D.

Any ideas warmly welcome.

Magnus

[Alejandro Tombolini]

Hi Magnus, you may try with a mask to protect the galaxy core during deconvolution. Probabily the default setting in StarMask tool will generate a mask that include the biggest stars and de Galaxy core.
By the way, I think that the DBE you have applied can be improved so it shows a more homogeneous background.
Saludos, Alejandro.

[magnusl]

Hi!

Cool, thanks. So, I can see two options:
1. I produce a mask of the galaxy core, add it to the luminance mask that I use that protects the background.
2. Make a star mask that includes the galaxy core along with the stars, and use this for "local ssuport" for derining.

Which one would be better? I ask because I've tried both, with no success, but I could ezperiment more with the more viable option.

As for DBE: were you thinking of any particular tweak to DBE? Grateful for any suggestions!

Magnus

[Alejandro Tombolini]

Hi Magnus, just generate a starmask and use it protecting the image. (protecting de galaxy core and big stars).
I have tried Starmask tool over your you have shared (I understand that it have deconvolution applied) and the parameters showed in the attached screen-shot work fine to generate a mask that can protect the galaxy core during deconvolution.
If you are going to generate the mask over your undeconvoluted image, you may need to adjust some values on the Starmask process.

Regarding DBE, I would need to see the original image, previous to DBE to have a better idea of what parameters can be adjusted, but check if you are not placing samples over the stars.

Saludos, Alejandro.

[magnusl]

Hi!

Cool, I uploaded the integrated image (integrated_approved.xisf), pre DBE, in the dropbox folder so if you want to have a look, I'd be grateful.

I also noted that the "doughnut" in the galaxy core actually is present in the DBE-version. But not in the pre-DBE version. I did a new DBE, and this time no doughnut. Hrm. Is there any way a DBE process could result in such a doughnut?

Magnus

[Alejandro Tombolini]

Hi Magnus, there is no way DBE can generate that. Maybe you worked with DBE over an already deconvoluted image.
Attached is a quick DBE process that works quite well but maybe you can improve it even more.
You may also want to check again the images approved for integration and the integration parameters, so you may be able to avoid the stars' artefacts showed in big stars (see attached image)

Saludos, Alejandro.

[magnusl]

Hi!

yes, felt weird that DBS should produce that. So it's a mystery. I start all over again.

yes, those artefacts - what might those be? I really don't know what to look for.

Best,

Magnus

[Alejandro Tombolini]

what might those be? I really don't know what to look for.

Hi Magnus, I would first used Blink to inspect the images. You can select the registered images (the _c_d_r on BPP script) and see how are they in the corners. Note that the defect is not in the center and seems to be worse radially to the edges.
If you detect out of focus images or with elongated stars it is better that you exclude them from the integration.
You can also inspect the rejection maps to see what is been rejected an maybe adjust the rejection parameters. Make a try integrating the images without rejection and compare it with your actual result.
Hope this helps.   

Saludos, Alejandro.