"Bald Eagle" <cre### [at] netscapenet> wrote:
> This looks like just the thing to solve Ari's little project from a few years
> ago:

Wow, that's a monster thread. Lots to read there. Yes, I thought about stone
walls before, and even tried it out briefly, but ran into my non-square-grid
issues and didn't go back to it yet.

> Also, I'm sure if there was pebble size control in different regions

This would be great, but has implications for the grid efficiency. Either a
custom grid, with all the brute-force slowness that entails, or future work I
think!

> You're doing this by checking the possible bounding by 4x4x4 = 64 surrounding
> cells?  Or is it really 5x5x5, so 125....  :O

Yep, I'm currently building a grid, perturbing positions within their grid
voxel, then looping over the nearest 124 (2 in each direction) neighbours to
calculate each cell. If there was some way to efficiently find all the seeds
that influence the current cell before getting down to the intersections and
trimming, I could drop the grid approach and use any old input seed set. I
suspect that such a broad approach is either vexingly non-trivial, or maybe even
impossible :( But I will continue to ponder on it :)

Post a reply to this message