taby said:
It looks like the quaternion fractal

Yeah, i can see the typical revolutions.

Btw, i often use just points for visualization. It's faster, so you can tune parameters and see a higher resolution result within interactive rates.
I use the density or mass gradient for shading, so a form of voxelization is still needed, though.
This image has 22k points:
 

taby said:
P.S. I can now make the Mandelbrot set for any dimension n, where n is a positive integer. It's not just n = {2, 4, 8, 16} anymore. I consider this a breakthrough, if it's not already well known.

You mean you can make a 3D mandelbrot as well?
Sounds hyper interesting :D Pls post some images…

Yeah, I thought it would be new if 3D, but really, it's the same thing as the quaternion Julia set where Z.w = 0. Nice thought though!

The difference between vector and array is that array by default will consume the stack while vector the heap. The array is going to allocate in place while the vector is going to allocate on the heap. Which or what combination would be more performing is going to depend on a lot of stuff. If you are particularly interested in this topic, look up, data oriented design, for insight

We ended up doing 3D slices of the 5D Julia set. The following image was created by our teammate Paul.