Efficient 3D content authoring framework based on mobile AR

Many applications for generating and authoring 3D content are dependent on interaction and visualization on a 2D screen. There have been various applications using 3D interaction to overcome the limited interaction in 2 dimensional space. Those applications, however, require special hardware [13] for interaction or marker-based techniques for scene recognition and tracking. Both approaches make the applications less intuitive and uncomfortable especially when one creates and manages 3D objects in the content. In order to overcome these limitations, in this paper we propose a novel application that creates 3D content such as 3D animations and 3D movies. In our method we make use of Augmented Reality (AR) techniques and hand gesture recognition in 3D AR space. To obtain the specific information of an open set, we use real-time object recognition and marker-less tracking based on the generic randomized forest (GRF) method, which facilitates the use of mobile devices like ultra-mobile PCs (UMPCs) for generating 3D content. Since in general it is difficult to generate realistic 3D content on the mobile devices in real-time, we apply a server-client model to our application so as to separate content construction and scene generation. Based on the techniques mentioned above, we develop an interactive and easy-to-use AR application using a natural user interface (NUI) to make 3D content for not only experts in the related fields but also for general users; we show the novelty and the efficiency of our proposed approach via making a short 3D animation.