Abstract: We present here the ?rst release of an SDK (Software Development Kit) for mobile devices supporting the animation of 3D talking heads: THAL-k. The SDK is constantly evolving and here we discuss the features of version 1.0. This library is thought as a support for all the developers wishing to build applications on smartphones or tablets including avatars to enhance the interaction functionalities. The main challenge we face is to provide developers with a complete SDK for the creation, customization and real-time animation of the models.
Authors: F. Sorrentino, R. Scateni.
THAL-k: TalkingHead Animation Library.
CG Libs Smart Libraries for Computer Graphics (poster).
Pisa, Italia, Giugno 2013.
Abstract: Curve-skeletons are compact and semantically relevant shape descriptors, able to summarize both topology and pose of a wide range of digital objects. Most of the state-of-the-art algorithms for their computation rely on the type of geometric primitives used and sampling frequency. In this paper we introduce a formally sound and intuitive definition of curve-skeleton, then we propose a novel method for skeleton extraction that rely on the visual appearance of the shapes. To achieve this result we inspect the properties of occluding contours, showing how information about the symmetry axes of a 3D shape can be inferred by a small set of its planar projections. The proposed method is fast, insensitive to noise, capable of working with different shape representations, resolution insensitive and easy to implement.
Abstract: Different poses of 3D models are very often given in different positions and orientations in space. Since most of the computer graphics algorithms do not satisfy geometric invariance, it is very important to bring shapes into a canonical coordinate frame before any processing. In this paper we consider the problem of finding the best alignment between two or more different poses of the same object represented by triangle meshes sharing the same connectivity. Firstly, we developed a method to select a region of interest (ROI) which has a perfect alignment over the two poses (up to a rigid movement). Secondary, we solved a simplified version of the Largest Common Point-set (LCP) problem with a-priori knowledge about point correspondence, in order to align the ROIs. We eventually align the poses performing least square rigid registration. Our method makes no assumption about the starting positions of the objects and can also be used with more than two poses at once. It is fast, non-iterative, easy to reproduce and brings the poses into the best alignment whatever the initial positions are.
Abstract: The number and quality of smartphones on the market has dramatically raised lately. Researchers and developers are, thus, more and more pushed to bring algorithms and techniques from desktop environments to mobile platforms. One of the biggest constraints in mobile applications is the fine control of computing power and the relative power consumption. Although smartphones’ manufactures are offering better computing performance and longer battery life, the mobile architecture is not always powerful enough. Furthermore, nowadays, the touchless interaction (e.g., the usage of voice commands) on mobile devices is particularly attractive. The device can also possibly answer to our questions (e.g., Siri-Speech Interpretation and Recognition Interface, which according to Apple is “the intelligent personal assistant that helps you get things done just by asking”). The use of talking avatars can improve the quality of the interaction and make it more useful and pleasant. Since avatars are static models, but the interaction requires dynamics, it is almost obliged to introduce avatars’ animations.
Abstract: Shape reconstruction from raw point sets is a hot research topic. Point sets are increasingly available as primary input source, since low-cost acquisition methods are largely accessible nowadays, and these sets are more noisy than used to be. Standard reconstruction methods rely on normals or signed distance functions, and thus many methods aim at estimating these features. Human vision can however easily discern between the inside and the outside of a dense cloud even without the support of fancy measures. We propose, here, a perceptual method for estimating an indicator function for the shape, inspired from image-based methods. The resulting function nicely approximates the shape, is robust to noise, and can be used for direct isosurface extraction or as an input for other accurate reconstruction methods.