A clip can imitate a sideways dolly using a flat pan. That may satisfy a loosely written prompt but miss depth needed for a convincing walk past furniture. Test relative parallax using points at different distances rather than judging the clip solely by how far it moves.
Build the test input
Use a designed corridor scene with a near black doorframe, a middle shelf, and a small clock at the far end. Request a gentle lateral move with the clock staying in frame. The near edge should travel farther in the image than the clock. Keep the geometry simple enough to trace.
Run the proposed comparison
Use the same crop and start still across candidates with image-to-video input. For each export, mark the three points at the start and at two later frames. Compare displacement as a fraction of frame width so different file sizes do not change the score.
Decide what passes
A flat slide moves every marker by approximately the same fraction. A believable shift keeps the depth ordering with larger displacement in the nearer plane. If a marker is generated away or changes its shape until it cannot be tracked, note a geometry failure rather than forcing a measurement.
Act on the outcome
Select a model suitable for the shot's depth requirement. If no run demonstrates clean separation, ask for less displacement or animate a deliberate layered scene in post. A modest movement with the right depth will often cut better than a wide move with poster-like geometry.

