When I first set out to make my Prodigy map, I sketched out a rough topographical map, picked a starting point, and then used Alexis' procedure (outlined here) to randomly generate the elevation of all the hexes in the map.
Along the way, I ran into a couple of problems, which I outlined in my previous post. I wrote it back in December, but forgot to post it until now.
As the map neared completion, I ran into my latest problem with the procedure: since each hex is created by a single generating hex from the previous iteration, terrain trends proceed in a mostly linear fashion, iteration to iteration. So, my terrain features ended up being mostly straight lines (this is especially apparent with regard to the lakes that appeared in the bottom half and right side of the map).
My problem with this is that geologic features are shaped in zones, not radiating lines. And my approach, while well-suited for a smaller scale endeavor, ended up not being quite good enough.
This brings me to the map you see above, which is the finished map overlaid with large (3 hex tall) hexes. My thought was to apply a layer of obfuscation to the map, taking the information within each large hex to calculate 3 things: the mean hex elevation (taking into consideration, but weighting less, the hexes which are but partially contained within the larger hex), the minimum hex elevation, and the maximum hex elevation.
I've now recorded that data for each of the larger hexes and my next (and final effort for the base map) is to use this to populate all of the smaller hexes from their parent hex (the larger one). The math is quite simple: the mean hex elevation for the hex remains unchanged. Excel then rolls 6d6 and subtracts 21 for each of the internal hexes. If the difference is positive, then that difference is multiplied by the difference between the mean hex elevation and the max hex elevation, divided by 16 (so that no hex will be exactly at the max height). If the difference is negative, then that difference is multiplied by the difference between the mean hex elevation and the minimum hex elevation, divided by 15.
Now, any hex's elevation has been produced not by one hex but by all of the hexes surrounding it. I think it will also lead to an exaggeration of geologic features - my relatively flat parent hexes will produce even flatter children, but my mountainous regions, where elevation changes wildly, will be even more varied.
When I finish, I will post the result and reflect on whether or not the procedure was worth the extra effort.