Right now I'm working on rings, which are fairly trivial (it's just a 1-D fractal, pulled into a circle) except for the annoying detail that I have to render them in two pieces, front and back, in order to have them encircle a planet. I'm still thinking about the best way to create the planets' surface maps; definitely have to at least experiment with a cube map and see if it's even plausible to go in that direction.
Funny thing, though. I was searching around the Web to see if anyone had been playing with cube-mapped spheres, especially in the context of drawing fractal planets. I found lots and lots and lots and lots of webpages, progress reports, and forum discussions from people working on fractal planet rendering engines. Some of them quite awesomely sophisticated continuous level of detail systems, too, which supported smoothly panning from interplanetary space down to a ground level! I wrote a very basic CLOD landscape engine way back in my first game dev job (nowhere near as advanced as the ones I see out there now) so I know it's a complex task -- and making it run not just smoothly but on a planetary scale is an impressive achievement. But as I looked over all these pages, one thing stuck out about all of them.
Every single one of these projects, without exception, went nowhere. The progress reports become farther and farther apart. The forum discussions just stop. The web pages vanish. The project always, always dies. I have never ever heard of a single completed indie project of any type or scale that had, as a feature, sophisticated fractal planet rendering. [*]
I wonder why that is. My current, tentative theory is that if an engine sports such advanced planet rendering, it tempts the developer to sick, stupid levels of ambition for the project, levels of ambition that no single human being could ever come close to fulfilling, and thus the project fails.
[*] There is Orbiter, but I don't think that has fractal planets in it: you can fly down to the planet, but it's just texture-mapped flatness.