I saw this on here: https://www.food4rhino.com/en/app/octopus?page=1

I am extremely interested in how the author achieve such effect with Octopus.

In here it is said that:

An input curve is divided into n points P with a Tangent vector T each. A neural network is evolved by Neat towards a fitness function which in one value sums up the deviation of the tangents of the sample points from the tangents of the network points.

Can someone point me to the right direction how to do this? any tutorial? hints?