Hi again,
So I have been continuing to think deeply about the StarMask process (and I should probably stop!)...
1. I think I found a partial answer to my "bug" report. I noticed in an old screen capture of the StarMask tool. It used to have just a checkbox for "growth." This appears to have applied growth at all scales and you needed to have it checked to get any small scale enhancement. Maybe this is why you cannot set Large Scale growth to "0" and see anything happen to small scale growth? It is a legacy of the old version?
2. The hover descriptions describe the growth as increasing "mask protection." Taken literally I do not think this is what is happening. The growth is happening to the detected structures. In the final generated mask (assuming inverted is not checked as per the default) this is making larger white/gray structures which is doing the opposite of providing mask protection in the meaning of what a mask does.
3. The section titled "Mask Preprocessing" really REALLY confused me for a very long time. The preprocessing is directed to the target image. The hover description did not help since it states these are "Options applied to the initial image." However the subject is Mask... so initial image here (grammatically) means "mask." The hover description for the histogram section unfortunately didn't help either. It states: "The value is part of the histogram transformation applied to build the initial structure detection image." However, the verb "applied" is missing its subject!!
For a long time I honestly thought that "Mask Preprocessing" was adjusting the initial mask based on the generated result from the above.
So...
Perhaps the title of the section should read "Target Image Preprocessing."
Then the hover description that these are "Options applied to the initial target (or input) image."
Finally the other hover description could say "The value is part of the histogram transformation applied to the target image in order to build the initial structure detection image."
------------------
Juan has highlighted the concept that preprocessing the TARGET image (or a copy of it anyway) is an important skill to master when creating a good star mask (e.g. using HDRMT). This preprocessing section is just another example of it. This concept then gets lost in the language. I entirely understand that this product is created by authors who speak many languages. Is there a repository for these kind of suggestions where the labeling and descriptions fundamentally hinder initial understanding of how something works? My guess is these things are fairly easy to adjust in the PI code?
-adam
