re: modeling at a bigger size - just a note on that, it can also not be too good to create things at a really big size as well, like 100000 units across your object as well.

For that case the reverse problem happens and instead of the fitting tolerance being too loose the fitting tolerance can be too tight in relation to the object and generate overly dense results. When that happens it makes for large file sizes and slower calculation times.

I am trying to move more of MoI to a "relative tolerance" method where instead of targeting a fixed tolerance value it instead uses a fraction of the object's size as the tolerance. That can help make things adapt to either large or small sized objects better.

But that is kind of an ongoing process moving pieces over at a time to use that system, right now some things use it and some things don't, so it tends to be best to build things more in the size of maybe around 1 to 1000 units in numeric size (just the numbers themselves are what matters, not what the unit system is, the unit system is just a label).

- Michael