I was using depthmaps as mask for some experiments, but the depthmaps are not always suitable out of the box so I use x_cut to get a visual feedback result which I then use as mask.
Now that all works fine but I would prefer a softcut rather than a cut, and there is no x_softcut only a gcd_softcut
However, I found a way around the absence of x_softcut that I can use:
What I do is I use +x_cut on the grayscale image (a depthmap in my case), then I assign the native variables im and iM (min and max values of the image) to my own variables to basically get the results (or should I say arguments) of the x_cut selection stored as numbers (AFAIK it doesn’t give results in numbers/variables natively right?), then I use ‘remove’ to get rid of the image created by x_cut, then I apply gcd_softcut with the values from the previously created variables, et voilà, I have my visually approximated softcut.
I have no idea if there is an easier/fancier way to achieve this, but it’s what I thought of so far.
So although I can get by without an x_softcut I feel GMIC should have it.
That of course is just one guy’s opinion, who just coincidentally would have a use for it, but how often do people use x_cut or an imaginary x_softcut? is it worth the effort to add it? Dunno.
I mean the approximation trick is OK, although I don’t see the falloff of course, but when you see the x_cut output you sort of know what to expect from a softcut enough to use it.
But on the other hand, if you see the extensive list of x_ commands it feel like GMIC ought to have a x_softcut.