I'm getting a bit out of my field since I don't know a lot about rendering (that is, modern hardware accelerated rendering; I actually know quite a bit about software rendering like Quake or Doom); however, I think a large portion of fancy graphics are already possible without Irrlicht changes, but just aren't taken advantage of. For instance, colored lighting only needs Minetest changes. Directional lighting from the sun is already partially possible, and I think that only a few Irrlicht changes are necessary to make it work better (but I'm no expert on that). Shaders are already possible, but we don't have many people who know how to use them, and no formalized API for them.
My main point is that we won't go to great extents to add large rendering features, like raytracing. That would take a lot of work with little benefit. I have no doubt that we'll add useful features, but just not huge ones. The core developers are already pressed for time, so we aren't going to add large Irrlicht maintenance costs as well by making large changes.
So, don't worry too much. I have no doubt that we'll get much nicer graphics by taking advantage of our Irrlicht fork, just not gigantic graphical features that require significant code changes.