|Probablistic fill and a split.|
Currently, the split employs one or more fractally-perturbed line segments to divide a land mass in two. The resulting groups of plates are then shifted and slightly rotated. A slightly different split could be employed instead, which I shall term a hinge split. The same type of fractally-perturbed line segment(s) will be employed for the splitting, but the points where the split line crosses the plate boundary into "ocean" will be noted. One will be chosen as the origin for rotation, and one of the plates resulting from the split will simply be rotate slightly, rather than shifted and rotated about the centroid of the plate or land mass.
We'll see how those changes work in practice after I've had a chance to implement them. After all the main initial plates/landmasses have been split and laid out, they'll serve as the basis for division of the entire map into plates. The initial plates will retain their existing shape, but the rest of the map will grow new oceanic plates using probablistic filling. Perhaps larger continental plates that already exist may be split in two without being moved. These new plate configurations would then be use to generate some coarse elevation changes - mountains where plates collide, rifts, island chains where some oceanic plates collide, etc. That's probably a good week or two out, though.
* George E. P. Box wrote that"All models are wrong, but some are useful." My ultra-simplistic model might make a geologist scream, but I'm hoping it will rapidly produce acceptable land mass shapes and layouts which can serve as the basis for terrain generation.