Board Thread:Bug Reports/@comment-30645975-20170226013044/@comment-26172435-20170227095215

It's in use in five biomes and effectively compresses (the reverse of stretching) the surface in X and Z direction. This results in a look typical for heavily eroded limestone ('karstic') landscapes: steep and sharp edges and much variation of surface elevation over short distances. This works rather well within a biome with adjacent chunks using the same generation maths and logic. But not at biome transition zones where adjacent chunks are not generated with that parameter.

As far as I understand, the compression of XZ surface, as well as the stretching in vertical direction that happens in the Mountains of the Wind, is superimposed on each individually generated chunk after the surface has been determined with the same maths and logic that is used for any other chunk (based on and determined by the world seed). This is, in my view, the main reason why the match between surfaces of normal chunks and 'XZScale or StretchHeigth' modified chunks is so bad. The method now used by the mod basically creates an off-set in either horizontal or vertical direction on the seed-based generated surface.

Hence my thoughts and careful, nooby, recommendations to Mevans in that recent Facebook thread on the announcement of PB31.2 (iirc).