Sketched-based 3D Modeling and Animation

Peiying Chiang


We developed a 3D modeling system, called S3D, which constructs 3D models from the users’ 2D sketches. It has no special requirements on the input devices and provides a simple set of tools for 2D contour and skeleton drawing. Functionalities are provided in S3D to automatically improve the quality of the user-drawn sketches. S3D automatically recognizes the relationship between the contours and skeletons that users draw and based on which, automatically chooses the 3D modeling elements to fit the users’ drawing. A compact but versatile set of 3D modeling elements are provided which can be combined to model objects of a wide range of shape complexities including not only tubular objects but also man-made objects of irregular shapes. Moreover, 3D objects created with controllable elements are automatically embedded with a skeleton and users can define the skeleton structure by simple attaching operations. Enabled by the skeleton structure, the resultant 3D models can be flexibly manipulated. User evaluation shows that S3D is easy to learn and use and the 3D mode creation process with S3D is robust and simple. After a simple training, a casual user with only basic computer operational skills
may create a vivid 3D cartoon character in less than 20 minutes on the average.
demo mv4
Project Progress

This is an ongoing project so we cannot tell too much about it right now. This work will be finished in few months. Please check with us again later. Following shows some 3D objects created with our prototype:


Graphical User Interface

GUI prototype

2D contour 2D skeleton Estimation 3D object

Sample results

2D contours can be approximated with our deformable-cylinder, Ellipsoid, or Prism
Development tool
Java, Java3D




This is a continuous work of the 3D thumbnail system [1][2] that we developed previously.  The similar customized primitives and approximation process are re-designed for approximating the 2D contour 

1.     Pei-Ying Chiang, May-Chen Kuo, C.-C. Jay Kuo, “Feature-Preserving 3D Thumbnail Creation with Vox-based Two-Phase Decomposition”, Advances in Visual Computing, 6th International Symposium, ISVC 2010, Las Vegas, NV, USA

2.     Pei-Ying Chiang, May-Chen Kuo, Teri Silva, Milton Rosenberg, C.-C. Jay Kuo, “Feature-preserving 3D thumbnail creation via mesh decomposition and approximation”, Pacific-Rim Conference on Multimedia, Shanghai, China, Springer, 2010