Beyond Geometric Closure: Measuring and Diagnosing Mechanical Consistency in Visual Dynamics

Abstract

Visual accuracy, geometric validity, and agreement with applied forces are distinct requirements for visual dynamics. We introduce a diagnostic framework that connects reaction-free SI measurements to controlled interventions on coordinate readout, training integration, and actuator response. Reference replay calibrates the measurement path, while continuous scoring exposes progress against fixed engineering targets. With three seeds per setting, we evaluate KeyCLD and CON across 343 planar assemblies at 64/256 pixels and V-JEPA + AC and LingBot-VA on 100 shared spatial test assemblies. CON reduces image MSE by 37.6%/37.2% relative to persistence, yet neither planar model attains valid native geometry. At 256 pixels, calibrated RGB replay places forecast errors far above the reference measurement floor. Exact-state KeyCLD has identical geometric pass rates under Euler and RK4 training, yet its score rises from 52.65 to 64.30 as force and energy errors decrease. In both spatial pathways, CAD/IK reconstruction restores geometry while mean actuator residuals exceed five times the engineering targets. These diagnoses distinguish coordinate correction, numerical improvement, and persistent force-response errors, guiding targeted improvements to visual dynamics.

arXiv comment34 pages, including supplementary material; 16 figures and 38 tables. Code: https://github.com/BIGPAO123321/mechanical-consistency-visual-dynamics Data and spatial evaluation: https://zenodo.org/records/23448115 Trained models: https://zenodo.org/records/23448170 Pretrained models and continuation states: https://zenodo.org/records/23448172