I always hated structural biology when i was studying for my master's, now i feel thankful that these people exists when i need a structure so i don't have to solve it myself...
A big part of modern structural biology is computer-assisted folding prediction based on crystallography but mostly NMR data (and some black magic for sure). Maybe you've heard of www.fold.it, this is a collaborative project that crowd-sources possible protein structures proposed by users as part of a puzzle game. I don't know how useful these data are currently to solve real structures but i like the idea :)