Mix3D: Assembly and Animation of Seamlessly Stitched Meshes for Creating Hybrid Creatures and Objects

Symposium on Spatial User Interaction (SUI)(2022)

引用 0|浏览7
暂无评分
摘要
Hybrid creatures are a part of mythology and folklore around the world. Often composed of parts from different animals (e.g., a Centaur), they are also increasingly seen in popular culture, such as in films, video games, print media, clothing, and art. Similarly, hybrid fictional entities composed of parts from different objects are widely visible in popular culture. Thus, modeling and animating hybrid creatures and objects is highly desirable and plays an important role in 3D character design and creation. However, this is a challenging task, even for those with considerable prior experience. In this work, we propose Mix3D, an assembly-based system for helping users, especially amateur users, easily model and animate 3D hybrids. Although assembly provides a potentially simple way to create hybrids, it is challenging to extract semantically meaningful segments from existing models and produce interchangeable edges of topologically different parts for seamless assembly. Recently, deep neural network-based approaches have attempted to address parts of this challenge, such as 3D mesh segmentation and deformation. While these methods produce good results on those two tasks independently, they are not generalizable to human, animal and object models and are therefore not suitable for the task of heterogeneous component stitching as needed for creating the hybrids. Our system tackles this issue by separating the hybrid modeling problem into three automatic and holistic processes: 1) segmenting semantically meaningful components, 2) deforming them into interchangeable parts, and 3) stitching the segments seamlessly to create hybrid models. We design an user interface (UI) that enables amateur users to easily create and animate hybrid models. Technical evaluations confirm the effectiveness of our proposed assembly method, and a user study (N=12) demonstrates the usability, simplicity and efficiency of our interactive user interface.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要