I’ve seen a couple of people here talking about and showing maps for different seeds that look very similar at the big scale but with different details. It could of course just be a coincidence, but it seems to happen a bit too often and with too many random values at the same time, so it might be a bug.
The reason that I’m writing this is because it reminds me of a bug I had in a school project a few years ago.
We were generating an “endless” city but noticed that the random height of the buildings (not much more than boxes) were just a few different values, instead of anything between min and max.
For each building we generated a LocalSeed based on the WorldSeed and the X and Y coordinates of the building and used it for the random parts of the building. The problem was that the first two or three values that we got for each LocalSeed were limited to just a few possible values. I assume that this is because the random number generator use the last couple of generated values when it generates the new one, causing it to more or less fail the first couple of times.
We used the standard random number generator for C++, rand().
Solved it by simply not using the first three values.
I’m thinking that if you generate local seeds for each random vector that the simplex noise is based on, then your random vectors might not be as varying as think.