An arm can follow the motion reference until it disappears behind a bag, then come back with its wrist rotated the wrong way. Isolate that moment when deciding whether an image-led motion-control result can survive a cut into a close shot.
Build the test input
Film a permitted reference in which the right hand accepts a tote, leaves view behind its side panel, and comes back to grasp the short handle. Choose a character reference whose sleeves make the wrist outline visible. Mark the last clear pose, hidden interval, and first clear pose.
Run the proposed comparison
Use the same character image and reference take for each operation that accepts both. Keep the tote on the same side of frame. Fix the frame rate when reviewing exported files, and compare the first matching pose after the obstruction rather than identical timestamps from different clip lengths.
Decide what passes
Pass only if the correct hand returns on the correct side of the handle and the sleeve stays attached through the grip. A brief soft patch in full occlusion is less consequential than a wrist flipping in the first unobstructed frame. Save the frame that first makes the mismatch visible.
Act on the outcome
When the return fails repeatedly, use a new pose that keeps fingertips visible or cut to a separate insert before the grab. Compare the cost of those edits with rerunning the shot. Reserve wide takes that fail the close-up check for an edit where the obstruction still conceals the wrist.

