Hi Mark,
1)
1. Should decon be executed before any of the steps in the above "b"?
You can do deconvolution after crop and after DBE, and particulary I would do it before MMT. However in some noisy images I have used MMT prior, to reduce some noise in the one pixel's layer.
2. Did you apply a luminace mask to the image prior to decon? You don't mention it.
No, I did not.
3. You did not use a star mask in this example?
No, I did not. But I suggest you to allways use a starmask to protect cores of biggest stars. When the image is linear a starmask made with the default values will work well for this, or perhaps could be necesary to increase "scale" to include largest structures if you have a big star for example.
4. You did not modify any values in Parametric PSF?
No. All default values.
6. What drove your decisions for the values in Wavelet Regularization?
Deringing value are by try and error. And the same for WaveletRegualrization
I suggest first work in a small preview, taking in consideration that the result will be different in the big image than in the preview, but this will allow you to quickly determinate an very aproximate setting and understand what happens when increasing o decreasing a value. Then, for example if you still get little dark rings when apply the process on the image, you will know that increasing a little "Global dark" it will be solved.
See what was the procedure in this case. I have exagerated the stretching, due the changes are very subtle
Original:
First select a number of iterations
The result has dark rings, then is needed to apply Deringing. I selected the values by default
The Global dark is exagerated and now we have bright rings. I select 0.01 in Global dark as is a value that usually works.
Now is better but still there are dark rings. To help with this now try Global bright and after a couple of tries, set 0.002 as good value:
But blinking with undo redo preview, can be seen that the backgroung is not ok, then I start checking the values of WaveletRegularization. I choose Poisson as noise model (try and error) and adjust layer 3 and 4
Now apply the process to the image, and it is expected that other adjustments are going to be necesary. (showing the preview but applied on the image)
There is "ways" and "shadows" between the stars and it would be necesary to adjust again Global dark, Global bright and may be Wavelet Regularization, but the easy way is to limit the number of iterations with "Coverange"
Regards,
Alejandro.
