Same tracking pass as the skeleton study, kept on top of the video instead of isolated from
it. Scrub the timeline: the pose moves because the person in the clip moved, not because a
model was told to.
Source clipserve 182732
Coverage90 / 90 frames
ModelMediaPipe Pose Landmarker
Duration3.0 s @ 30 fps
How to read it: every joint position is a per-frame estimate from the video
pixels, no sensor involved. Same underlying tracked skeleton as the isolated Skeleton Study —
here it stays pinned to the person it came from.