3D generation and animation of avatars

Abstract

The integration of Artificial Intelligence in 3D content creation has introduced transformative possibilities, yet significant barriers exist for non-experts seeking to leverage these advancements. This work addresses the accessibility and knowledge gap between 3D AI-generated content (AIGC) technology and end-users by developing an interactive tool for generating and animating 3D avatars. The tool combines advanced AIGC models with traditional 3D development workflows, creating an intuitive platform for seamless 3D content creation. A series of experiments were conducted to evaluate the performance and efficiency of state-of-the-art 3D generative models, including both text-to-3D and image-to-3D paradigms. The models were tested on identical rendering tasks, with outputs assessed qualitatively through human ratings and quantitatively by measuring generation times. The generated 3D avatars are further animated using Blender Python API, offering users a library of predefined action sequences and rendered animations. The tool not only simplifies the complex processes of 3D asset creation and animation but also demonstrates the potential of AIGC to save time and resources in the traditional 3D development workflow. This project represents a significant step toward democratizing 3D AIGC technology, fostering greater public interest and investment in this rapidly evolving field.