PixInsight equivalent of "Enhance DSO" from Carboni tools

[tnkumar]

Is there something similar to this in PixInsight?

[Alejandro Tombolini]

Hi

An easy way would be with a starmask to protect background and apply Morhologial Selection to stars.

Saludos. Alejandro.

[Josh Lake]

Morphological with a star mask will reduce stars with incredible control.

Local Histogram Equalization does a fantastic job of enhancing features, again with a great amount of control. See Harry's: http://www.harrysastroshed.com/LHE.html

[Juan Conejero]

Is there something similar to this in PixInsight?

Short answer: no. PixInsight is completely different from the application(s) where that tools live. We don't have magical boxes like "enhance xxx", "colorize yyy" or "apodify_zzz". This is real software and real image processing. As Alejandro and jlake have pointed out, we probably have tools implementing image processing algorithms and techniques that can be used to achieve results similar to what you want to do.

[tnkumar]

Thanks all for explaining how I can use PixInsight tools for achieving the desired results.

[Josh Lake]

To be fair, Carboni's actions are doing some similar things, albeit in a total 'black box mode'. I opened the action for Enhance DSO and Reduce Stars and noted that it essentially: creates a star mask; runs the Minimum filter repeatedly (ten runs of 3 iterations!); does some curves and noise reduction.

As a PI user, I of course don't think the 'one action for any image' strategy makes any sense, but you could theoretically step through this action one step at a time and see what it does.

The reason I was interested in Carboni's PS actions was from Tom Davis' presentation at NEAIC 2011 on capturing and enhancing dust. He basically finishes an image, then runs this Action multiple times (and goes on a coffee break, it takes 10 minutes per run), and that brings out the dust. His before/after shots were impressive, but I know that the technique is absolutely possible in PI... I just don't know how.

Here are Tom's images with this 'blind processing' technique: http://tvdavisastropics.com/astroimages-1_000024.htm
His list of published images is staggering: http://tvdavisastropics.com/astroimages-1_000045.htm

[Nocturnal]

Tom is a legend for sure.

[Juan Conejero]

The problem is not with what a particular magical box does or not, but with the concept of magical image manipulation.

finishes an image, then runs this Action multiple times and that brings out the dust.

This is an excellent example of what I mean. I read this sentence and the following questions arise immediately in my mind: Why? How? When does it work and when doesn't it? Through what mechanism is it supposed to work for this particular task? What is this 'action' doing with the image? How can I be sure that the 'dust' is real and/or correctly enhanced if I can't explain what is happening and why? (not that I'm saying that these particular results are not correct; I'm speaking in general terms)

I can put you examples of dust that is the result of stretching the image with straight histogram manipulations and dynamic range compression techniques. A couple of them:

http://astronomiapampeana.com.ar/foto/77/NGC-6727-Dust-and-Gas-in-Corona-Australis.html
http://pixinsight.com/tutorials/NGC7023-HDR/en.html

Both techniques---histogram stretching and HDR compression---are perfectly documented and described in the literature, including our documentation and many tutorials. So the visible dust in these images is the result of perfectly justified and repeatable procedures. Moreover, these procedures are based on relatively simple principles and algorithms; there is no magic at all, just science and technology.

So instead of magical recipes, we pursue a scientific approach to image processing. For example, when you say:

Morphological with a star mask will reduce stars

you are talking about a well-known and perfectly documented image processing technique: Grayscale morphology, which is an extension of mathematical morphology applicable to continuous tone images. I can explain you why an erosion filter reduces the extent of small-scale bright structures such as stars. You can find good descriptions on the Wikipedia and on fundamental textbooks such as R. Gonzalez.

Local Histogram Equalization does a fantastic job of enhancing features

Zbynek Vrastil's LocalHistogramEqualization tool is an excellent implementation of the CLAHE algorithm (Contrast-Limited Adaptive Histogram Equalization). Again, CLAHE is a well-known image processing algorithm that can be found in the literature, including a nice description in the Wikipedia.

And I could continue. Even if the results of a magical box can be nice, I will never agree with the underlying idea. Magical processing is essentially boring because it teaches us nothing at all; it doesn't allow us to grow. Our mission with PixInsight is the development of astrophotography through image processing culture, so we are against the concept of 'magic', irrespective of the results achieved.

Sorry for the lengthy story, but the topic is important and I am particularly sensitive to these controversies today

[Josh Lake]

Believe me, Juan, you are preaching to the choir, at least when it comes to me! The very fact that all Photoshop processes were black boxes with a very arbitrary feel is exactly what drove me to PI, and I'm here to stay. I used to be one of those, "Processing in PI, finished in Photoshop" guys, but as I learn, I find that I don't need to go to Photoshop for anything anymore.

My response was in part to elicit this response from you -- some links and methods that would allow me to figure out how to duplicate the 'dust enhancement' in PI (with science!) and out of PS (with 'magic'!).

Thanks. 

[ruediger]

The best replacement for the asked astronomy tools "Enhance DSO and reduce stars" function IMHO is "ExponentialTransform" with Function PIP, followed by a star size reduction with star mask (contours) and morphological selection.

I don't fully understand the discussion here. I still use Photoshop with all kinds of action sets for postprocessing of astropics. What's the difference to PixInsight's highlevel scripts like "MaskedStretch", "DarkStructureEnhance", "ClassicStarMask" and so on? These are black boxes to the end user which are only fully understandable if someone digs into the source code. The ExponentialTransform has basically only one useful slider (order). It does magical things, but is even more blackboxed than a photoshop action that unfolds to a large list of understandable and customizable single steps.

Rüdiger

[Nocturnal]

That's fair Ruediger. Plenty of apparent magic happens in some of the PI processes. With scripts you can peel back the layers and see what's going on (DSE as an example) but you can't do that with processes. Of course Juan is always willing to explain what each process does and what the source literature is.

[Carlos Milovic]

Touché I'm planning on writing the documentation for ExponentialTransform this month. There is an old tutorial online that describes the whole process with a lot of detail. http://www.pixinsight.com/tutorials/obsolete-LE/exp-cmilovic/en.html