When Strong Scene Reconstruction Hits a Wall: The Dynamics Problem

Like a photographer who can capture a perfect still-life but freezes when asked to film a waterfall—static skills don't automatically translate to understanding motion and change.

This means current AI can't yet reliably watch a video and understand *how* things move and deform, which is essential if we want machines to predict, simulate, or manipulate the physical world.