About Usual math functions in G'MIC-QT

What is the difference between argmaxabs() and maxabs()?

Feature Request : What about variation of narg? I would like to have nargnonzero to count all of the number of nonzero value within a vector? That’s one variation. The recent commit I did reminded me that maybe this should be a thing.

Another question, what in the world is cbrt()?


I got that after looking for it in google. I prefer to use x^(1/3) myself, but now I know.

cbrt() may be faster than x^(1/3) and calculates cbrt(-27) to -3, wehereas x^(1/3) for x = -27 should give a NaN

I discovered cbrt() recently too. This is what @David_Tschumperle told me in a PM

This computes the minimal of the absolute values of image pixels. You could achieve something similar, e.g. with:

$ gmic img1.jpg img2.jpg +abs min[-2,-1]

but this means duplicating the images, applying the abs, then the min, while absmin does that in a single shot, without need to duplicating the image data.

So I believe we can think of

in the same way: abs then argmax and abs then max.

I had to read it multiple time to get it, but thank you for the clarification. The feature request part haven’t been typed about though.