Moin,
> How did you create your own version, you simply knew what to do off - hand ?
well, “of hand” is not the best way to describe it
I struggle quite a bit when understanding the math I find somewhere - but I use this kind of project to learn new tools. When I learned Fabric, one of the first tools I built was a 3d-interpretation of Julia/Mandelbrot sets, simply setting one of the coordinates by the iteration count (which, usually, gets displayed as an indexed color). Building a software solution to this helps me in getting into the software (here: Houdini) *and* understanding the math - other people prefer bungee jumping or wild water rafting …
My “solution” does not use a SDF (I did not think of that, but that's the nice thing that you can learn new tricks all the time), but only sets points on a “cubed grid” (grids stacked over one another) or deletes them. When I created the image posted at the beginning of this thread, I had no idea about how to render something (not even talking about points/particles) in Houdini (not that I know much about that yet

), so the result may not be as visually pleasing. But *seeing* some result from your experiments is the best motivation, at least for me, so I didn't really care about “appeal”.
The best approach to math-involving stuff for me is to find sample implementations in code. I usually understand code better than math terms, so by “aligning” the code to the formula, I get an idea of what is going on - then I can take it from there, vary the formula and, at some point, write my own version.
For example: I created a gas/smoke solver in Fabric (from scratch). I had an idea of how to do it, but I couldn't make much sense from most of the formulae I found online (only some of them “clicked” with me, but I always thought I was missing something). Then I read the book by Jos Stam and found out that I wasn't missing anything at all, it's just that math-geniuses like to make things over-complicated to keep their caves clean from amateurs :-D So I build the smoke solver based on my understanding. And it worked very well.
So, in short: It's my way of learning tools: Trying to solve problems I don't fully grasp and by that “approaching understanding” and not caring too much about the “unbearable hurdles the tools put up for me” (people kept telling me that Houdini was impossible to learn. By “solving” problems like 3d fractals I can ignore that - and concentrate on the real issue).
I hope this, in some way, makes sense

Marc