Why Did the Wrong Person Drive My Product-Demo Motion Reference?

Identify the Intended Motion Performer

A merchant might record a salesperson demonstrating a coffee machine while a customer stands closer to the camera. Although the salesperson performs the commercially important action, the customer may occupy more of the frame.

Runway's Kling 3.0 Motion Control guidance recommends a single performer. When multiple people appear, motion selection follows the person occupying the largest screen proportion.

This rule belongs to the documented Runway integration. It should not be presented as a universal rule for every Kling interface or other motion-control service.

Before submitting a reference, identify the intended performer and examine whether another person dominates the frame. A correct product image or character identity reference does not resolve ambiguity about whose movement supplies the performance.

Decision branch: If the intended performer is not visually dominant, evaluate a clearer single-performer reference before attempting motion transfer.

Check Continuity and Action Visibility

A useful driving reference should show the intended action continuously enough to understand the movement. Runway recommends a continuous single shot with the performer visible, avoiding disruptive cuts, excessive camera movement, obstruction, and overly rapid action.

Vidu's official Motion Control guidance separately favors a single front-facing performer and clear reference material. That supports the general editorial decision to prioritize an observable performance, but it does not establish Runway's largest-screen-proportion selection rule for Vidu.

For product demonstrations, visibility includes the hands and the object being manipulated. Cropping away a foreground bystander may reduce performer ambiguity, but an aggressive crop can remove the product contact that makes the action understandable.

If a clean crop cannot retain the performer's relevant body movement and product interaction, obtain another recording or redesign the shot. Do not assume any crop will force correct performer selection.

Illustrative Exercise: Obstructed and Clean Demonstrations

Illustrative exercise — fictional merchant references, not executed Runway or Vidu operations, generated videos, or physical demonstrations.

Input A shows a salesperson behind a counter operating a coffee machine. A customer in the foreground occupies most of the image and intermittently blocks the salesperson's hands.

Input B shows one salesperson from the front, with the coffee machine and both hands visible during the demonstration. The shot remains continuous.

For A, the team should not assume the product operator will supply the motion. Evaluate whether a revised reference can remove the ambiguity without hiding the actual demonstration.

For B, the performer is easier to identify, but the generated interaction would still require independent review.

Decision branch: Replace an ambiguous reference when the intended performer or hand-to-product action cannot be observed reliably.

Separate Performer Selection From Motion Accuracy

Review the reference in two stages. First, determine whether the intended person is clearly the motion source. Second, determine whether that person's movement adequately communicates the required product action.

A correct performer selection does not guarantee that a hand grips a handle, presses a switch, or releases an object at the intended moment. Those interactions require separate visual inspection.

Keep the original reference and a short action description together in the production handoff. Identify the performer, the relevant body movement, and the expected product contact.

If the wrong person is selected, correct the driving reference rather than repeatedly changing the target character's face. The face reference determines a different part of the task.

Stop when the available footage cannot clearly identify the intended performer or preserve the essential demonstration. Choose another reference or a different production approach instead of treating repeated generation as a reliable correction.

Illustrative Exercise — Motion Performer Selection and Reference Suitability

These are illustrative exercise rows. Save a copy and record your own observations; the example is not an executed acceptance test.

Download review worksheet (CSV)