3

I saw recently that Simplex noise(3D and higher dimensions) is patented... A substitute for simplex noise exists to avoid(only a lawyer can tell) the patented parts, namely Opensimplex. But I am not a lawyer so I don't want to risk anything...

Back to square one with the older noise algorithm, namely Perlin noise and it's directional artifacts.

Can anyone think of an algorithm that eliminates or at least reduces the generated directional artifacts from Perlin noise in 3D?, i.e. I need an algorithm that corrects the final generated result from Perlin noise. It doesn't matter if this algorithm is slow, because everything is only generated at startup!:)

karl88
  • 341
  • 2
  • 13
  • 1
    I created OpenSimplex. ---- I'm not a lawyer, but I extensively looked at the Simplex patent and reasonably believe OpenSimplex hits none of the claims. ---- Part of the reason for creating OpenSimplex was to create a visually-isotropic noise function that doesn't hit any of the patent claims, and part of it was also to create such a function that produced a smoother less-bubbly appearance than Simplex noise. – KdotJPG Jul 29 '16 at 19:51

1 Answers1

3

Since multi-octave Perlin noise is generated by adding octaves of noise together, directional artifacts can be reduced by rotating each octave by a different (random) amount.

You can also add multiple noise planes (each rotated separately) together at each octave, but this will change the appearance of the noise.

Here's an article by Ken Perlin about improving the appearance of Perlin noise.

samgak
  • 23,944
  • 4
  • 60
  • 82
  • Thank you so much! Rotate each octave works perfectly! This article seems very useful for my purpose. Thanks again. – karl88 Jul 23 '16 at 17:07