OptionalatmosphereFraction of the actor's own color that survives the scene's fog at the
camera depth of the actor's world root: 1 at the lens, falling toward 0
with distance, from sceneFogTransmittance over the scene's fog.
Screen coordinates alone do not say whether the subject standing there is fully visible or barely separable from the haze, so a pose-conditioned host reads this to know. Present only when the scene declares fog.
The actor's named joints projected to the frame, in
resolvePoseKeypoints order, with off-frame joints flagged rather than
clamped.
Id of the performing actor's scene node.
One performing actor's projected joints in one output frame of the pose-keypoint sidecar.
planPoseKeypointSidecarwrites one record per performance whose motion, skeleton and scene node resolve, in the shot's performance order. A pose-conditioned generator readskeypointsas the actor's skeleton on screen andatmosphere, when present, as how visible the actor is through the scene's fog.