This was caused by samples generated with at least one coordinate being 1.
I changed it to generate points [0,1) which actually is what Ebeida was already doing looking at the code now.
This is also more usable as these tiles fit neatly next to each other and random generation is almost always exclusive on the upper bound.
Fixes #29
This was caused by samples generated with at least one coordinate being 1. I changed it to generate points [0,1) which actually is what Ebeida was already doing looking at the code now. This is also more usable as these tiles fit neatly next to each other and random generation is almost always exclusive on the upper bound.