I had a quick look at the paper and this scheme seems useful as well.
I'm willing to implement the general conversion functions for this
scheme as well if and when I have some time left.
But only after cubehelix is approved so I know what the format should be.
If you already have code to do it that would be even better of course.