1 Web-based Shape Modeling with HyperFun Richard Cartwright ? , Valery Adzhiev ? , Alexander Pasko ?,? , Yuichiro Goto ? , Tosiyasu L. Kunii ? (?) BBC Research & Development, UK (?) The National Centre for Computer Animation, Bournemouth University, UK (?) Hosei University, Japan; author for correspondence: pasko@k.hosei.ac.jp (?) IT Institute, Kanazawa Institute of Technology, Japan We present a project on research and development of the high level language HyperFun for shape modeling using implicit surfaces and the more general function representation (FRep). An approach to collaborative Web-based shape modeling using HyperFun is described in detail. The presented EmpiricalHyperFun shape modeling system based on Empirical Modeling principles provides users with an unusual degree of mutual interaction through the Web. Key Words: Shape Modeling, Web-based Modeling, Implicit Surfaces, Function Representation, Empirical Modeling, HyperFun, Java, Applet INTRODUCTION Growing attention is paid recently to shape modeling and rendering using so-called “implicit surfaces” [3] and a more general function representation (FRep) (see the side bar) because of their well-known properties such as compact mathematical description of complex shapes, natural blending, covering both free-form organic shapes and mechanical parts with sharp edges, and a rich system of developed operations. The increasing number of applications facilitates the importance of methods and software tools for Web-based shape modeling. These applications include Web authoring, multi-user gaming, distant learning, data visualization, 3D retail environments in e-commerce, collaborative computer-aided design and engineering, and others. Many of these application areas have their own proprietary data representations of shape, making sharing of these shapes for collaborative engagement with a team of users a task for dedicated software with restricted distribution and applicability. There are several well-known protocols for exchanging geometric data based mainly on polygonal and parametric surfaces models, which are quite verbose. A more high-level and concise format for geometric data transmission would enable truly collaborative and creative shape modeling on the Internet through browsers. This can be achieved by employing shape models of a higher level of abstraction than is provided by polygonal meshes. Implicit surfaces and FRep are among the best candidates for solving this problem. This paper presents a project devoted to developing an open system architecture and Web-based shape modeling applications using HyperFun, which is a high-level programming language for specifying implicit surfaces and FRep objects. We first give motivation for and the general characteristics of the project, and then discuss in detail its Web-based modeling aspects such as a HyperFun to Java Bytecode runtime compiler, a simple HyperFun applet, and a more general collaborative modeling system called EmpiricalHyperFun (EHF). This system based on Empirical Modeling principles (see the side bar) allows for implementation of realistic behavior of shape models that are situated in multi-agent environments, providing open-ended exploration and experimentation with shared models. RELATED WORKS Most of the known implicit modeling systems are oriented towards specific subsets of objects and operations such as traditional skeletal models [15], convolution surfaces [11], distance-based models [6], or Constructive Solid Geometry (CSG) [4]. Although all these models are of the same mathematical nature, it is still not possible to exchange models between the systems, and therefore between the users. In this paper, we present a project devoted to developing an open system based on the more general function representation (FRep) - see the side bar. There are several well-known protocols for exchanging geometric data such as polygonal file formats (e.g., Autodesk DXF), Alias/Wavefront object files for parametric surfaces, PADL-2 for Constructive Solid Geometry (CSG), and STEP for the boundary representation (B-rep) and CSG. The Virtual Reality Modeling Language (VRML) [13] is an open standard that can be used as a means to distribute a shape to special Internet browsers that can be explored by a distributed set of users. VRML besides the description of the shapes contains additional information about the graphical scene such as light sources, background, texture, and other data. The recently proposed X3D standard [16] allows extensions with plug-in components for creation of standardized profiles of 3D Web clients. One of the X3D profiles fully supports the VRML97 specification. However, the expressive power of VRML is restricted by the polygonal representation of shapes and the set of additional predefined parameterized shape primitives such as a sphere or a cube. This provides only ever an approximation to the real shape model. VRML representations of shapes are typically very large, which is an obstacle to collaboration in the design Pre-final version before editing by the IEEE CG&A magazine editors, ?appeared in vol. 25, No. 2, 2005, pp. 60-69.
