Bloom Removal Script?

[jeffweiss9]

I've been contemplating buying an NABG camera for the big boost in QE but that would be contingent on good bloom removal s/w and was hoping someone would come up with a good script for PI for doing this.  I know it's been discussed in this Forum previously but I haven't seen anything emerge here and would like to stay within PI to do this. 
  Ron Wodaski's algorithm is considered perhaps the best from what I've read and he has offered the source without write-up for someone to turn into a PI script at:

http://pixinsight.com/forum/index.php?topic=1300.msg6616#msg6616

Any of our great script writers here or PI Development Team members want the challenge?

Thanks,
Clear skies,
Jeff

P.S. As another seed idea to those discussed previously in this forum, I like the idea of taking some short exposure subs so the bloom removal tool at least knows where the stars are located in the obscured segment, as well as where they are NOT located, which would supply a nice constraint.

[georg.viehoever]

Here is a method that already works with PI: http://pixinsight.com/forum/index.php?topic=5023.msg34770#msg34770 . Maybe you give it a try.
Georg

[jeffweiss9]

Georg-
 That's a great start but I don't think it's necessary that you have no information whatsoever so that this is pure painting:

Quote from: georg.viehoever on 2013 January 08 17:37:16
   " Deblooming as discussed here specifically adresses the problem of overflowing CCD cells, see http://learn.hamamatsu.com/articles/ccdsatandblooming.html. There is typically no useful information that you can recover from bloomed pixels, with any technique whatsoever. You just dont know if there is a star behind a set of bloomed pixels, or if if is background."

What about the 'advanced technique' in Maxim that uses some short-exposure subs so it doesn't have complete lack of information about what is behind the bloom; in particular, whether there are stars there or not.

In any case, since this would have to be performed on all the subs, it would have to be a batch script that could mostly run by itself with a few hand inputs where it needed human guidance.

-Jeff

[Carlos Milovic]

Short-exp subs may be used in PI, with the HDRCombination tool to remove blooming. You'll need a large amount of those frames, to reach an acceptable noise level in the background.

[jeffweiss9]

Carlos-
  I was thinking it might be done with a small number of short-exposure subs if you are just worrying about where the stars are, or are not.   Although granted that would not give documentary accuracy, it might be better than no information on them.
-Jeff

[georg.viehoever]

Ji Jeff,

HDRCombination will substitute data from darker images into regions that are overexposed- such as those that show blooming.  The effect may therefore be similar to what MaximDL's 'advanced technique' does (I dont have Maxim, so I dont know in detail how it works).

Did you consider observational techniques such as rotating the sensor? They guarantee good data even for regions where 50% of the images have blooming.

Georg

[jeffweiss9]

Hi, Georg-
  That's a good idea and I think that's what the pros usually do.  Among several problems for me to do that is my use of an internal guide chip in my SBIG camera, not to mention that the lens-to-chip distance of the field flattener that I use isn't long enough to fit any automatic rotator I know about.    But I guess I could get a piggyback guide scope with more choice of guide stars and I do know of another field flattener that should give a longer lens-to-chip distance with my scope, although it also reduces 0.75.  It would be nice not to have to redo my whole system to take advantage of the high QE.
-Jeff