I think there’s a bug somewhere in PlaneRotationGizmo in 4.2 alpha.
From 4.1 to 4.2, mesh.scaling is now slightly affected by rotation gizmo.
Adding @Cedric our Gizmo Guru
I’m taking a look!
I get scaling values of 1.0000001360647142 instead of 1.
That’s the float epsilon due to computation. Nothing I can improve here. Do you get greater/different values?
No, same small values here , but why does it affect scaling at all?
I know it was changed to also support non-quaternion rotations, is this related?
I just remembered i made a local version for 4.1 to support non-quaternion rotations,
using quat.toEulerAnglesToRef() when restoring it.
found the snippet aswell, https://www.babylonjs-playground.com/#31M2AP#123
Gizmo code changed a bit the last 2 months to support various types of mesh, nodes, camera,…
So, the math behind also change and changing operation order can result in variation in that range.
For example, division is famous for having less precision that multiplication. When I say ‘less’ it means having difference of 10e-6 maybe, so no worries