Represents an SRT transformation.
An SRT transformation can represent a smooth rotation with fewer motion keys than a matrix transformation. Each motion key is constructed from elements taken from a matrix S, a quaternion R, and a translation T.
The scaling matrix \(S = \begin{bmatrix} sx & a & b & pvx \\ 0 & sy & c & pvy \\ 0 & 0 & sz & pvz \end{bmatrix}\) defines an affine transformation that can include scale, shear, and a translation. The translation allows to define the pivot point for the subsequent rotation.
The quaternion R = [ qx, qy, qz, qw ] describes a rotation with angular component qw = cos(theta/2) and other components [ qx, qy, qz ] = sin(theta/2) * [ ax, ay, az ] where the axis [ ax, ay, az ] is normalized.
The translation matrix \(T = \begin{bmatrix} 1 & 0 & 0 & tx \\ 0 & 1 & 0 & ty \\ 0 & 0 & 1 & tz \end{bmatrix}\) defines another translation that is applied after the rotation. Typically, this translation includes the inverse translation from the matrix S to reverse the translation for the pivot point for R.
To obtain the effective transformation at time t, the elements of the components of S, R, and T will be interpolated linearly. The components are then multiplied to obtain the combined transformation C = T * R * S. The transformation C is the effective object-to-world transformations at time t, and C^(-1) is the effective world-to-object transformation at time t.
- See also
- OptixSRTMotionTransform::srtData, optixConvertPointerToTraversableHandle()
