Hey, thanks for the reply,
Ok, so yeah I would have to extract multiple rotation / scale / translation matrix from a any matrix then. But I think it’s not feasible on certain situation.
I mean, since
Rotation is :
[cosX, -sinX, 0]
[sinX, cosX, 0]
[0, 0, 1]
Translation :
[1, 0, x]
[0, 1, y]
[0, 0, 1]
Scale
[X, 0, 0]
[0, Y, 0]
[0, 0, 1]
How could I get a combination of the above basic transform with this simple matrix ?
[1, 2, 0]
[3, 1, 0]
[0, 0, 1]
(I took only 2d coordinates because I only need 2d, and it simplify the problem.)
I’m considering taking vertices of the mesh directly and transform their position, but would it be heavy on the performance ?
Thanks !