So I ported Ken Perlin’s exact implementation of improved noise over to blueprints with the only change being all the variables being limited to floats since doubles aren’t accesible via BP.
(Here’s the link to the java implementation by Ken Perlin: https://cs.nyu.edu/~perlin/noise/ )
So the code works fine (output between 0-1 with no negative values) until you get to coordinates of around 50ish when it starts spewing out values in the several millions (some positive some negative). Eventually it just outputs “inf” or “-inf”.
Is this a problem with using floats instead of doubles? I’m also open to it being a typo somewhere in my port but it works fine for smaller inputs. I also triple checke the “Grad()”, “Lerp()”, and “Fade()” functions and they all work as expected individually.
Any input into this matter would be a huuuge help. Thanks!