MSCD
Abstract
This study examines how lightweight 3D motion reconstruction and domain-specific annotation can reshape remote Tai Chi instruction. Because Tai Chi relies on three-dimensional trajectories, center-of-mass control, joint coordination, and tacit embodied principles such as song, conventional 2D video often creates gaps between instructors’ judgment and learners’ bodily understanding. To address this limitation, the study develops a consumer-grade multi-camera workflow using Caliscope for calibration, Pose2Sim for skeletal reconstruction, and a Blender-based annotation system for visualizing joint orientation, balance deviation, and frame-specific feedback. Through an exploratory qualitative comparison of three teaching conditions—real-time Zoom instruction, pre-recorded 2D video review, and 3D reconstruction with interactive annotation—the study finds that annotation, rather than reconstruction alone, is the key pedagogical mediator. It externalizes instructors’ tacit spatial diagnoses, supports more principle-driven explanations, and helps learners locate errors through specific joint relations rather than overall form matching. However, the translation from visual understanding to bodily execution remains unresolved. The study contributes a Tai Chi-specific annotation framework and offers broader insights for remote embodied learning, movement pedagogy, and digital cultural transmission.

