@Shalvenay did you check questions on this site already? Like gis.stackexchange.com/questions/14731/… or gis.stackexchange.com/questions/160492/…

@Midavalo (btw I don't mean that negatively toward you, more that I didn't make note of the places for later recall, so of course someone asked me where those places were) :)

@ChrisW looking at those and some of the links off those -- I think things are getting closer -- but not quite to where I'm going with autogenerated terrain -- or does it really matter with a heightmap over a moderate-island-sized area whether the heightmap pixels represent a given lat/long increment, or a given x/y increment?

@Shalvenay That depends on where on the globe said island is. And how big the globe is since you've got a fictional world. If you replicated the earth for your globe, a one degree island is 69 miles (ish) wide at the equator and at 40 degrees north it's only 53 miles wide. At 80 degrees north it's only 12. Your issue is you're starting with flat, projected data (x/y pixels). You're trying to un-project this data back into a sphere(ish) if you're using lat/lon (angular measurements).

So either your world has to be flat, or you have to realize that your square pixels from the terrain generator aren't really square once you wrap them onto a spherical shape. Which is what that first question is getting at. I saw an interesting video on youtube the other day on the topic of projections, sort of a quick introduction that basically describes what you're trying to do... only in reverse. :) youtube.com/watch?v=kIID5FDi2JQ

@ChrisW yeah -- un-projecting is a good word for what I'm trying to do