Non-uniform scale in jME

If it weren't true I might be offended by such comments…instead I'll just sit back with my afro and watch the jabs fly.  :mrgreen:


Honestly, they are not jabs.

Meaning they're true, so their not jabs, or meaning they're meant in good fun and jab is too negative a phrasing?  :slight_smile:


hehe, darkfrog, you're just trying to keep the fire going aren't you???  :wink:

renanse said:

Matrix4f = 8+64bytes or 72bytes  (am I missing how you got 76?)

+4 bytes pointer to that matrix (included these bcs we need less fields with matrixes)

Just trying to figure out what you meant…I'm innocent of stoking the fire…honest.  :roll:


Ok, I had a look into the Eberly book: Wild Magic 3 supports non-uniform scale in Geometry only. This way world rotation and scale can still be read. But this would not be enough to read any .3ds file.

And we discovered a major point (at least for us) against not being able to read rotation separately: physics! ODE needs a rotation being specified as Quaternion and does not support scale.

For us this problem can only be solved by converting the meshes before using them (either in max or while loading).

So: is non-uniform scale needed elsewhere? For me the costs of the solutions I have seen are quite higher than the benefits :frowning: