Big Movements or Small Motions: Controlling Digital Avatars with Single-Camera Motion Capture

Mingyang Su, Binlin Feng, Junfan Zhao, Haoqian Yu, Keyi Zeng, Xiu Li

Abstract: Digital human avatars provide users with more natural interactions in virtual reality, games, and specific scenarios and enhance user immersion. This study explores the ways of controlling a digital human avatar and the effects of different movement amplitudes on the user experience, using a monocular motion capture technology, Mediapipe, for whole-body and hand joint information capture and digital human control. The study tested the interaction with different movement magnitudes (small, medium, and large) by designing two game tasks - simple walking and complex gesture manipulation. The results show that small-range actions are easy to control and improve ease of use in simple tasks. In contrast, medium-range actions provide a higher immersion and engagement experience in complex tasks. These findings provide new insights into designing user-friendly digital avatar interactions that can help improve user satisfaction and the quality of virtual experiences. We expect these insights to guide the development of digital incarnation technology.