Now, I do actually think that's a pic of a garnet.  But it's still fun to recall that a rhombohedron can be derived from a cube by stretching or squashing along a 3-fold rotation axis (of course, that has to be happening internally and not just morphologically), and a rhombic dodecahedron, similarly stretched or squashed, becomes the stubby hexagonal prism (or is it two trigonal prisms?) capped by a rhombohedron that in part make Yinnietharra dravites so interesting (besides their huge size).  Stretch or squash various isometric crystals along different symmetry elements and you can derive the other systems and forms.  It's the elegance of how the crystal systems are related... :-)