Dark Scale Enhancement Algorithm

[budguinn]

Hi all,

I've been trying a few tuts, and the Dark Scale Enhancement one is giving me fits.

I've done the color space deal.
Then the ATrousWaveletTransformm, but the AWT that I do doesn't match anything from the Tut....in fact it doesn't seem to do much of anything, let alone remove all color....I've tried to match the onscreen steps, and settings, but to no avail.  I've tried this with three different color images.

any ideas.....the on screen tut is from a different version than my current 1.06..could this be the cause?

Once I'm successful on this step it says that I must "To isolate the dark structures efficiently, the result from the previous step is subtracted from the image with PixelMath.  An image without bright small-scale structures is obtained after this operation (Figure 3). ""

How is this accomplished?....by applying the AWT to a "clone" of the original and then subtacting in PixelMath?.....the instructions aren't clear at all.

[Juan Conejero]

Hi Bud,

Yes, that tutorial was written around an older version of the ATrousWaveletTransform tool. You can find it under the Compatibility category, as ATrousWaveletTransformV1.

The DarkStructureEnhance algorithm has been implemented by PTeam members Carlos Sonnenstein and Oriol Lehmkuhl as a JavaScript script in PixInsight. You can run it from the main menu: Script > Utilities > DarkStructureEnhance. It is a wonderful algorithm and a very nice implementation.

[Nocturnal]

For a while I ran that script on all my nebula images. It works quite nicely. Not sure why I stopped using it. Different processing sequence I guess.

[budguinn]

thanks guys for the quick response....I'll go back and give it a try again..

bud

[RBA]

You can run it from the main menu: Script > Utilities > DarkStructureEnhance. It is a wonderful algorithm and a very nice implementation.

I've always found that DSE often enhances already existing dark structures a bit too much, and the 'amount' parameter doesn't adjust for JUST that, so adjusting that parameter is not the solution.

You'll remember I added two sliders to control that from the script, but the idea, already implemented (you even helped me refine it a bit) was rejected by Carlos. Oh well..

Not complaining, just sharing what I thought about it at the time (and still am), since I don't think I ever did...  

[Carlos Milovic]

DISCLAIMER: The other Carlos, not me

[RBA]

DISCLAIMER: The other Carlos, not me

Ah yes, sorry!!! The other Carlos!

[Jack Harvey]

Not sure how practical this is as I do not know how to program but a Fade tool with Real time preview, which would let you fade the results of the previously applied process would be very handy.  SO if you applied DSE, or Curves and after the application felt you were too heavy handed you could use a slider to fade it by a certain amount?

[budguinn]

Or......let's see.....hmmm......what else might work......hmmmm

I've got it ....how about a layers function with a "variable" mask....to eliminate the "heavy handed" areas?

[RBA]

Not sure how practical this is as I do not know how to program but a Fade tool with Real time preview, which would let you fade the results of the previously applied process would be very handy.  SO if you applied DSE, or Curves and after the application felt you were too heavy handed you could use a slider to fade it by a certain amount?

What those two sliders did, AFAIR, were to adjust the histogram at the time of running the script. That allowed you to control how dark the dark areas will get. You can of course run the tool to extract the mask, then adjust the mask and apply it, and do other things as well as you mention. And that's fine. My only beef with this is that I found that most of the time the areas that are already dark get "enhanced" a bit too much, so an adjustment at the time of running the script seemed a good thing, as it avoided what I saw as a common issue when running the script, that's all.

[Niall Saunders]

If you have applied the process to a Preview of the image, then all you need to do is adjust the sliders of that process and reapply. The effect is instantaneous. I don't see what the problem is

Take your image, grab the 'side tab' and drag'n'drop it just below its current location, and a full-image preview window will be created. MAKE SURE YOU THEN CLICK ON THIS NEW PREVIEW TAB TO MAKE THE PREVIEW ACTIVE. (If you can still see the green 'preview outline rectangle' then you are still looking at the ORIGINAL 'parent' image).

Pick your process of choice. Pre-apply a mask to the image if you wish - the mask is ALSO automatically applied to the Preview as well

Set the process parameters, and apply the process (drag and drop the triangle, to the Preview window, or just click the square, or hit key F6 - the latter two options apply the process to the HIGHLIGHTED image, obviating the need for the drag'n'drop inconvenience). The Preview image is altered - the original image remains unaffected.

Change the process parameters. Hit F6 again. The data from the ORIGINAL image is re-processed and the result appears in the Preview image. In other words you will NOT see the second process being applied OVER the results of the first process. You can even apply a totally different process - it will STILL use the original image data, modify it in some way, and then display the result in the Preview window.

And, you can use the quick Undo/Redo button that is available on the toolbar - which ONLY works on a process applied to a Preview (unlike the Ctrl-Z/Ctrl-Y buttons that work on a 'live' image, but which require you to move the mouse between two different buttons on the toolbar). In other words you can 'blink' between the 'before' and 'after' states with a stationary mouse and single clicks.

And, if you 'like what you see' you have two choices:

1.) Drag the preview side-tag to the workspace background to automatically create a brand new image, and then carry on processing (I really love this feature, as I can save images and processing histories at different stages of my workflow)

or

2.) Click the 'live' image tab on the window you were working with (you will see the Green 'Preview' outline reappear) and just re-apply the process - except, to the 'Live' image this time

Further, if you really want to be able to 'fade' a process - why not just create, and use, a full-frame mask comprising a 'single shade of grey'? PixelMath will do this in an instant. Use something like $T = k, where k is a value between 0.0 and 1.0; and apply the PixelMath to your image, using the 'Create New Image' option, to leave the original image untouched, but the new image having the same physical properties as the original one. (((By the way - this is just one of my 'thought experiments' - I have NOT tried it, because I have never yet had a need to)))

Cheers,

[Carlos Milovic]

Just "k" as expression

[Niall Saunders]

Carlos,

Have I just learned 'something new'?

You mean that, in the expression line where I might have typed something like $T = $T * 0.5 (to 'halve' the intensity of an image), I can simply type 0.356 to have PixelMath create a uniform image, where all ADU values were set to 0.356?

Now, THAT is a useful shortcut.

So simple I might not have to delete anything from my overcrowded brain in order to memorise it

Cheers,

[Carlos Milovic]

In fact, to halve the image intensities, you just need to write "$T*.5"

[Niall Saunders]

Yes, sorry, I knew that, but was trying to express myself more clearly

Cheers,