Star masks for Deconvolution

[m_abukhalid]

Ive read the tutorials referenced on this forum:
http://pixinsight.com.ar/en/info/articles/46/mask.html#StarMask
http://pixinsight.com.ar/en/info/processing-examples/49/barnard150.html

I have also been following the tutorial on using convolution and creating masks here:
http://www.lightvortexastronomy.com/tutorial-sharpening-fine-details.html#Section1

After a couple of hours of playing with star masks I have decided that Im completely lost so hopefully someone can answer some of the points that are confusing me. I am currently working on a widefield image. This image has been preprocessed and then I applied the DBE, background neutralization, Photometric color calibration and finally SCNR. Following that I have applied a level of noise reduction using MLT on the image luminance and chrominance which I believe has cleaned up the background nicely. The result can be seen in the link below. I would really like to bring out the nebulosity in the orion nebula (not much color there at the moment and slightly oversaturated). So the first step was to use deconvolution to bring out as much detail as possible.

https://drive.google.com/file/d/14on4DTBWNM-nyVzCoO4zle5qExf7qr2F/view?usp=sharing

Should convolution be performed prior to any noise reduction or is it ok to apply it after MLT has been applied?
When applying the luminance mask, is  there a lot of value in using range selection on the luminance mask and then combining with a star mask as indicated in the kayron tutorial?
I am struggling with the metrics to use for a good star mask. For deringing, should the smaller stars be excluded from the star mask? Would that mean that the faint stars would get fainter and appear like noise?

[ngc1535]

Should convolution be performed prior to any noise reduction or is it ok to apply it after MLT has been applied?

I suspect it would be best to do deconvolution before any noise reduction. The reason is because deconvolution has a kind of threshold for the noise floor through regularization.
 If you smooth things the wings of the star PSFs will be affected and this will alter the results. It will work- but the point is to get the best result (presumably).

When applying the luminance mask, is  there a lot of value in using range selection on the luminance mask and then combining with a star mask as indicated in the kayron tutorial?
I am struggling with the metrics to use for a good star mask. For deringing, should the smaller stars be excluded from the star mask? Would that mean that the faint stars would get fainter and appear like noise?

Looking at that tutorial- it appears the author is concerned with the dark ringing of small stars and he takes care to simply not include them by virtue of a mask he constructs.
 I think in general the global deringing parameter controls this very well for small bright stars. Perhaps with undersampled data where the stars are already the size of a pixel- you can justify excluding them because you cannot sharpen them more. As for the first part of the question- (black) clipping the luminance mask is certainly the way to go (either by a range selection or just by HistogramTransformation). This clipping acts as a threshold- you are saying that deconvolution will only be visible on structures that are above a certain signal strength.

-adam

[m_abukhalid]

Thank you... I think I am slowly getting the hang of it... Deconvolution is bringing out more detail and improving the larger stars slightly when necessary but I am disliking the pixel dots I am seeing on the smaller stars... I am starting to see why you would want to remove them from the mask...