Thoughts on Star Masks ...

[phcjpp]

Hi Guys,

I just wanted to hear others opinions on star masks. It seems to me that being good / very good at creating high quality star masks is really quite important. Many many of the steps used in image processing seem to need a high quality star mask. I am currently doing the helix nebula in narrow-band and have really struggled to get a good star mask off the Ha, SII, OIII masters even though they are superb quality. I have gone as far as to take some 5 min luminance images just to get a good star mask (it worked). Am I mad ? Doing something wrong ? or does this make sense ? Star mask tutorials always seem to be quite short and always work. Images like the Helix in narrow-band seem very hard to get a good mask from.

Thanks and sorry for the rambling!
Chris

[Riccardo A. Ballerini]

a real time preview would be a life saver!!
In order to protect the smaller stars I always find myself doing LOTS of tries!

[marekc]

I've always found star masks to be, far and away, the hardest thing about Pixinsight. They're streets ahead of everything else, when it comes to difficulty and complexity, IMO.

Two main things make them hard for me:

1) Lots and lots of controls. The idea of 'experimenting' or 'iterating' is pretty difficult, when there are so many things to adjust.

2) The names and descriptions of the controls make very little sense to me. The best way I can describe them is that they strike me as "non-physical". Reading the name or description of a control almost never gives me a clear idea of what the control does.

Something like real-time previews, or even just pictorial depictions of what the controls do, would be a godsend for me.

In fact, here is how I envision the ideal, user-friendly StarMask tool:  You have your image, and you click on several of the smallest, dimmest stars that you'd like to see masked. Then you click on several of the largest, bloat-y-est stars you'd like to see masked. Maybe even a few stars in between. The tool would take it from there.

It's a bit like autoguiding. Imagine you're about to start an autoguider calibration routine, in a program like PHD or Maxim. You have the option to click on a star, in order to tell the program "*This* is the star I'd like to use for the calibration." You don't use a series of sliders and fill-in boxes in order to specify the star you're interested in. You click on it. Done and done. Clicking on representative stars is, in my dreams, the ideal way to interact with StarMask. I'd much rather see this implemented (if it's even possible) than any number of new sharpening or denoising tools, however marvelous they may be.

Another example of the kind of thing I envision: When you're using PEMPro, there's a Calibration Wizard, which figures out your plate scale, your position angle, which axis is RA, etc..  The program takes an unguided 20-second exposure, and you click on the beginnings and ends of several star trails in the resulting image. Done and done.

I realize that StarMask is doing things that are much more complicated than the examples I've given above, but I can't help fantasizing about a more user-friendly way of using the tool.

- Marek

[phcjpp]

Interesting .... I envision a chapter in a book on it! There are a number of ways to get to a star mask - very much depends on the type of image I think. These can include starting with range masks to help the tool out etc. There is also the classic mask script. Currently, range mask followed by star mask is my favorite.

Chris

[TheMadOne]

Welllllll......... !!  First off IMHO the one of features of PixInsight that makes it stand head and shoulders above everything else I've ever used is the ease of which creating good masks for performing the different steps of processing the varying features and highlights of your images can be done.  Now given I've used Christian Buil's Iris processing application in the past with some success, Iris has the same capabilities to make wonderful masks for your processing steps, but to do so requires diving head long into the command line interface of the program where you pretty much have to setup all of the parameters and such for your mask yourself first before even starting to try to generate them.  Then it takes yet another series of typing in various commands to apply and use the masks.  Where as in PixInsight you can pretty mush set everything up from the GUI fairly quickly and have it all right there on hand instead of having to call everything back up with the CMD Interface to begin using it.

I find the star mask feature was at first a bit of trial and error to begin to understand how to best take advantage of it's powerful masking capabilities, but still fairly quick to get to the best mask for the run as compared to everything else.  You basically just have to throw caution to the wind and dive in a little aggressively at first.  Depending on the noise in the background I usually set my noise thresholds up in the .2000 - .4000 range, I have bright larger stars to mask out I set the scale up to 6 or 8 to generate masking large enough to cover those stars, set the Structure Growth large scale no more that a 2 usually 1, the small scale depending on the  smaller star structures to 3 - 4, always set the compensation to 1 and the Smoothness I usually turn way down to 6 or 8.  This will usually give me a nice good fit for all of the star structures large medium, small, and tiny.  Sometimes I may have to run the Noise threshold up a bit  more, but for most of the images I start with starting out aggressively and laying out the star masks works quickest and best.  Also, after I make any changes to the stars I always generate a new star mask, usually the exact same settings work well on any subsequent star masks, but I find if I morph, decon, sharpen, or in any other way make changes even subtle to any structures it I get better masking results with a newly generated mask.  At any rate I've found I can create and work with masks in Pixinsight so much quicker and more accurately then I ever could with Iris, or PS CS.


Mark Jordan

http://www.thestardeckobservatory.com

[cfranks]

I found  http://pixinsight.com/forum/index.php?topic=4559.0  to be very informative on Star Masks

Charles

[Tom OD]

Here are some screen shots I made after playing about with star masks.
Tom

[Tom OD]

Scale test screen shot

[Tom OD]

Smoothness Test

[marekc]

Hi Tom,

Thanks for posting those screenshots! That sort of thing is really helpful.

- Marek