Yi-Jheng Huang, Shu-Yuan Chan, Wen-Chieh Lin, Shan-Yu Chuang.
Computers and Graphics, Vol.54, 2015, pp.127-134.
Abstract
Transformable models are 3D models whose overall shapes can be changed merely by rotating or translating their component parts. A good example of this is provided by the robots in the movie Transformers, but transformable models also exist in our daily lives, in the form of foldable furniture, tools, and toys whose shapes can be altered to save space or enhance functionality. Nevertheless, the successful creation of a transformable model is not an easy task. Designers have to consider how to shape and arrange each component part so that the model can take on two very different shapes by only translating and rotating its parts. They also need to carefully plan a transformation process, including the transforma- tion order of the component parts and the type of joints between them, to ensure that the parts do not collide with each other and cause damage or even the failure of the entire system.
In this paper, we present a system that can generate a transformable model and its associated transformation motion. Taking two inputs? source model and a target skeleton, with the latter representing a user’s desired figure? he system first adjusts the target skeleton and embeds it into the source model. Then, the simulated annealing (SA) is utilized to optimally segment the source model into parts. Finally, the transformation of these parts is animated based on the results of a two-level motion-planning process. The experimental results discussed below indicate that our system has considerable promise for assisting novice users to create transformable models easily, as well as providing profes- sionals with feasible options for references.
The contributions of this paper can be summarized as follows:
(1) We introduce the transformable-model problem to the fields of modeling and animation.
(2) We propose an optimization-based segmentation approach to generate suitable parts for the construction of transformable models.
(3) We propose a two-level motion-planning process to animate transformable models plausibly.