Open marklundin opened 1 month ago
That could be useful for sure, recast's voxelisation process doesn't require high precision in a lot of cases. I would be interested to see performance implications for larger inputs.
(Aside: If in creating recast-navigation/playcanvas you come across some logic in recast-navigation/three you could reuse, you could consider creating common utilities in recast-navigation/generators or recast-navigation/core)
Just wondering if it makes sense to have a prevision specifier here on the key, so that you can collapse nearby vertices?
https://github.com/isaac-mason/recast-navigation-js/blob/39ccf74307359b58145923aa454101b632912a08/packages/recast-navigation-three/src/utils/get-positions-and-indices.ts#L59