Screen space shape manipulation by global structural optimization

Existing structure-aware shape manipulation frameworks have greatly facilitated the shape manipulation procedure. However, they still require users to possess the necessary 3D modeling experience since the manipulation, e.g., positioning shape handles, is performed in 3D space. For novice users without such experience, this procedure remains tedious. On the other hand, performing manipulation in screen space is more natural to novice users, especially with 2D input and display devices. This paper presents a novel interactive framework for manipulating 3D man-made shapes in screen space. Users only need to edit abstract proxies of input shapes in 2D to obtain desired deformed shapes without considering their 3D geometries. This is achieved by first analyzing the structures of the shapes and then estimating the deformed shapes with a structure-preserving optimization method. Our efficient algorithm can produce real-time shape manipulation results as previews, thus avoiding a trial-and-error editing procedure. To verify the effectiveness of our framework, we conducted a user study that contained four shape manipulation tasks. In this study, we compared our framework with a baseline framework that requires users to edit shape’s geometry in 3D space and does not provide previews for the manipulation results. The statistics and the user feedback confirm that our framework is more intuitive and efficient for shape manipulation.