Quote:
Originally Posted by marc4darkskies
Holy cow Ivo! You're a fan of sharpening though!!  Granted this is probably a quick & dirty demo, but I'd be careful deconvolving this data. The stellar profiles don't need to be reduced IMO. |
I disagree - this data at this resolution is very, very soft, as is to be expected. Deconvolution in this particular image is able to recover a good bit of real data. As a matter of fact, as a validation of the decision to deconvolve, one can now make out diffraction patterns around the brighter stars.
Surely you're not going to seriously argue that this;
http://startools.org/download/Tutori...eforeDecon.jpg
is better than this?
http://startools.org/download/Tutori...AfterDecon.jpg
Quote:
|
In fact, I'm noticing halos around stars.
|
Yes, you are absolutely correct. I didn't take the time to create a proper star mask for the de-ringing (I just used a quick auto mask), leaving some stars unprotected (for example Eta Carina is fine, but some other smaller ones didn't make it in my mask
). I tried to do better (time allowing) in the second data set Bert posted.
Quote:
|
On top of that you've also done wavelet! Consequently it has a classic overcooked look that zooming in doesn't appear to mitigate.
|
I don't think you understand what wavelet sharpening is or how it works?
I stated that I left the smaller scales alone and merely increased the prevalence of the larger structures (which the deconvolution cannot have touched). I fail to see how that automatically means an 'overcooked' image?
Maybe you don't like the local histogram equalisation or optimisation?
Overcooking, to me, means showing features that aren't there (artefacts) or hiding features that
are there through the incorrect use of filters. What do you define as overcooking?
Quote:
|
I'm also noticing that you've rendered almost invisible (and maybe even clipped?) the faint extents of nebulosity that are screaming out to be revealed in this wonderful data.
|
Nope. By choice. But you're free to make a different choice. The automated scene (AutoDev) stretching routine comes up with the best possible curve by homing in on the curve that generates the maximum amount of detail
for a specific scale. It guarantees that all detail of a specific size (I chose smallest) is maximally visible within the constraints of a global stretch. It allows you more artistic freedom in choosing what the important feature is in your image.
By the way, I challenge you to clip your data in StarTools (except when explicitly allow or in the Layer module) - you won't be able to do it
Quote:
|
All of the techniques you mention may be applicable to this image but, and notwithstanding personal taste, they need to be done in a selective and controlled way to maintain some kind of natural look.
|
No. Just, no. There is no 'natural' look in astrophotography and anyone who claims they know what it is is a fraud. Strong words, but I'm passionate about this.
Personally, I think you yourself could stand to gain quite bit with your images by getting 'with the program' so to speak, as I don't believe you are getting the most of your data with the tools available these days. If StarTools isn't your cup of tea, PixInsight is also a fantastic way to get more from your data than is possible with a generic application like Photoshop.
Ah image processing... Rogelio Bernal said it best when he said "There are as many schools of astrophotography as there are astrophotographers".
Cheers,
Seeing I have a CCD now, got to try me some narrowband, IF the OSC CCD will allow MUCH of it (SXVR-M25c) - probably not much I guess? 
