Surface Block Generation Chance Depends on World Height

[mrgrim]

I noticed this while seed hunting for worlds with large amounts of obsidian rocks. The higher the world height, the less likely top layer obsidian would produce surface stones. This can be seen at https://github.com/anegostudios/vssurvivalmod/blob/81e52e9a61eabd6aa2131fb255f072feaf9901f1/Systems/WorldGen/Standard/ChunkGen/5.GenDeposits/Generators/DiscGenerator.cs#L359.

int surfaceY = heremapchunk.RainHeightMap[lz * chunksize + lx];
int depth = surfaceY - posy;
float chance = SurfaceBlockChance * Math.Max(0, 1.11f - depth / 9f);
if (surfaceY < worldheight - 1 && DepositRand.NextFloat() < chance)

The hard coded 9f I believe is the problem. Since everything is percentage based, taller worlds are penalized in this calculation. This constant should be scaled to either sea level or world height.

[mrgrim]

I've been doing a little more digging, and I suspect the sea level is being set after the deposit generators are initialized. This would explain even better the behavior I'm seeing especially with obsidian. The deposit variants are loaded in the AssetsFinalize stage of the GenDeposits ModSystem, but the world height is set in the StartServerSide of the GenTerra ModSystem.

GenTerra's execute order is later, but it's unclear to me if mod initialization is "interleaved". As in, is AssetsFinalize called for all mods before any mods StartServerApi is called?

I'm not sure if it's safe to delay initializing deposit variants or move the world height setting to an earlier phase.

[mrgrim]

I've created a new issue for what I described above, #5386. The surface rock generation and deposit generation issues are separate things.