Quote:
Originally Posted by Placidus
Thanks, Tim.
The cause of the magenta or red rings is the much greater stretching of [SII] and (to a lesser extent) [OIII] compared with the ubiquitous H-alpha.
We've got a lot to learn, but we're aware of six approaches to dealing with the resultant red or magenta stars.
(0) Ignore them. As that famous Caribbean astrophotographer Bob Marley so aptly put it, in his textbook Confrontation, "Dey say it's just a part of it.". On close inspection, many Hubble and ESO shots have magenta stars. This approach works best if you have a space telescope and tiny stars to start with.
(1) Remove the stars entirely. This involves trying to guess what is "under" the star. A bit like trying to repair an inkblot on a tablecloth by examining the pattern of the cloth and replicating it. Fred (BassNut) and others are good at this. Identify the star mathematically, then use multiple linear regression or the like to interpolate what would have been there. It works really well on smooth nebulosity, but very badly if the star concerned is right over some complex HH jet or shock front that is important to you.
(2) Identifying the stars and desaturating them to white. This is like bleaching a beetroot stain on a tablecloth. It's what we tend to do. It's just a cosmetic fix, but it doesn't require one to pretend to know what is under the star. The first limitation (not so bad) is that one's star-finding algorithm has to be able to find at least 20,000 stars, and to not imagine any, or you'll be left with all the faint ones, or have holes punched in your nebulosity. We try to avoid conspicuous grey rings by just partially bleaching, so the magenta is still there, but no so visually distracting.
The second limitation is that very bright stars take up a huge amount of real estate once they've been stretched to show [SII], and ultimately there's some overlap with the nebulosity. One can then choose between either (a) grey rings in the nebulosity, or (b) choosing to put up with red rings around the brightest stars. We think that's a good scientific solution, but they do stand out like sore thumbs.
(3) Taking RGB exposures of sufficient length that the star in its natural colour will be at least as big and fat and bright as the [SII] after stretching. One then registers the RGB shot to the narrowband shot, and takes the brighter of the two images. That's fair enough, and honest, but for something like say the Norma Supernova Remnant, where the SNR is incredibly faint, it's actually good-bye nebula. The stars overwhelm. It's also a lot of hard work for what is basically a cosmetic fix.
(4) A combination of (1) and (3). One guesses what is under the star and replaces it with a fake guessed background. One then puts smaller, less stretched, RGB stars more or less as place-holders, or eye-distracters, so that you don't see quite so badly how the star removal went.
(5) The final method (which again Fred mentioned to me) is to identify and remove the stars in the red and blue channels only. This works especially well in [OIII] because there tends to be less sharp detail in [OIII] and one doesn't have to guess so hard what to put there instead. One now re-identifies the (now much smaller) star, and bleaches it to white. That method should have the advantages of all and the disadvantages of none. We're yet to master it.
Best,
Mike
|
Nicely summarised
...lastly (6) you can't make everyone happy so do what
you like
