Stores the target pose only. previousPosition / previousRotation are the
finite-difference history of a running simulation, and §79 keeps simulation
state out of a static scene definition ("physics state, animation state, and
replay data must be separate optional sections"), so the history is not
written; on load it is seeded to the restored pose (capturePrevious), which
is the same state a freshly authored target has and differences to zero
velocity rather than to a spurious spike.
Scale is omitted when it is identity (1, 1, 1) so a document written
before the channel still round-trips. A present scale restores; absence
leaves the default. The physical side of a §19 blend is identity
(decision, 2026-09-06).
The
PoseTarget(§19, §42) serializer.Stores the target pose only.
previousPosition/previousRotationare the finite-difference history of a running simulation, and §79 keeps simulation state out of a static scene definition ("physics state, animation state, and replay data must be separate optional sections"), so the history is not written; on load it is seeded to the restored pose (capturePrevious), which is the same state a freshly authored target has and differences to zero velocity rather than to a spurious spike.Scale is omitted when it is identity
(1, 1, 1)so a document written before the channel still round-trips. A present scale restores; absence leaves the default. The physical side of a §19 blend is identity (decision, 2026-09-06).