Rémi Bèges

No improvement on perfs. Benchmarks to be written

Cell noise was using a bad hash function. Must be replaced by a native float/double method.

The functions returns values outside [-1,1], and seems also to have some weird behavior for some values of lacunarity. EDIT : FBM uses 2DSimplex as source, and the FBM noise's...

They contain non unitary vectors, once fixed it might improve a lot global apparence of all Perlin based noises.

Investigate why there were 4 dummy feature points in the old source version

See https://aftbit.com/cell-noise-2/ Only euclidean distances for now Need manhattan and chebychev