"triple_r" <rre### [at] hotmailcom> wrote:
> Here's my thoughts on preferential weathering:
>
> 1) Go over the object with a coarse grid and store the radii.  Store these in an
> array.
> 2) Go over the array and calculate the second derivative or some other measure
> of curvature.  Wikipedia should have the Laplacian operator.  Just take r
> constant.
> 3) Of course sharp edges will only have a local effect, so filter the data with
> a Gaussian function.  Just calculate an average of every point with its
> neighbors, weighted by a Gaussian distribution.  Very simple.
> 4) Now go back over it with your algorithm.  Same type of grid, so if it's finer
> just interpolate the curvature onto the fine grid.  Areas with zero curvature
> get the original function.  Areas with high curvature get the full
> perturbation.

This is a very good scheme, and fits very well with the current algorithm. I
will play with it over the weekend and see how it goes...

Thanks for the pointer. :)

Post a reply to this message