3D Avatar Animation Optimization in Metaverse by Differential Evolution Algorithm

Metaverse is an online platform that combines a wide range of technologies. It integrates the physical and virtual worlds to connect with other users in real-time. The application and development of the metaverse platform are growing in a broad range. For such platforms, an avatar represents the user to explore the metaverse platform. Therefore, the avatar, a 3D digital twin representing the user's expression and action in the metaverse, is essential to metaverse development. Creating a realistic 3D avatar animation is the next level of avatar development. The rigging step, an integration process of bones and joints to an object, influences the avatar animation. This study shows the concept of vertex weight value optimization adapted by the Differential Evolution meta-heuristic algorithm in the rigging process to create a realistic avatar animation. The fitness function represents how much a mesh driven by the corr. vertex weights is offset to a goal position under the influence of multiple joints. The proposed methodology is investigated for the challenging use case of animating face expressions, and shows the fitness function value of representing an animation could be minimized by 54.5%, from 1.49 of the initial design down to 0.68.