Use of Semantics to Manage 3D Scenes in Web Platforms

Internet And 3D Scenes

Computer graphics have widely spread out into various computer applications. After the early wire-frame computer generated images of the 1960s, spatial representation of objects improved in the 1970s with Boundary Representation (B-Rep) modeling, Constructive Solid Geometry (CSG) objects, and free-form surfaces. Realistic rendering in the 1990s, taking into account sophisticated dynamic interactions (between objects or between objects and human actors, physical interactions with light, and so on) now make 3D-scenes much better than simple 3D representations of the real world. Indeed, they are a way to conceive products (industrial products, art products, and so on) and to modify them over time, either interactively or by simulation of physical phenomena (Faux & Pratt, 1979; Foley, Van Dam, Feiner, & Hughes, 1990; Kim, Huang, & Kim, 2002).

Large amounts of data can be generated from such variety of 3D-models. Because there is a wide range of models corresponding to various areas of applications (metallurgy, chemistry, seismology, architecture, arts and media, and so on) (DIS 3D Databases, 2004; Pittarello & De Faveri, 2006; SketchUp from Google, 2006), data representations vary greatly. Archiving these large amounts of information most often remains a simple storage of representations of 3D-scenes (3D images). To our knowledge, there is no efficient way to manipulate, or archive, extract, and modify scenes together with their components. These components may include geometric objects or primitives that compose scenes (3D-geometry and material aspects), geometrics transformations to compose primitives objects, or observation conditions (cameras, lights, and so on). Difficulties arise less in creating 3D-scenes, rather than in the interactive reuse of these scenes, particularly by database queries, such as via Internet. Managing 3D-scenes (e.g., querying a database of architectural scenes by the content, modifying given parameters on a large scale, or performing statistics) remains difficult. This implies that DBMS should use the data structures of the 3D-scene models.

Unfortunately, such data structures are often of different or exclusive standards. Indeed, many “standards” exist in computer graphics. They are often denoted by extensions of data files. Let us mention, as examples, 3dmf (Apple’s Quickdraw 3D), 3ds (Autodesk’s 3D-Studio), dxf (AutoDesk’s AutoCAD), flt (Multigen’s ModelGen), iv (Silicon Graphics’ Inventor), obj (Wavefront/Alias), and so on. Many standardization attempts strive to reduce this multiplicity of various formats. In particular, there is Standard for the Exchange of Product model data (STEP) (Fowler, 1995), an international standard for computer representation and exchange of products data. Its goal is to describe data bound to a product as long as it evolves, independently of any particular computer system. It allows file exchanges, but also provides a basis for implementing and sharing product databases. Merging 3D information and textual information allows the definition of the project’s mock-up. As a matter of fact, 3D information describes CAD objects of the project and textual added information gives semantic information on geometries. The main issues are the sharing and the exchange of the digital mock-up. The next section explains how we use a digital mock-up to create an information system with the help of the semantic included in geometric information. Information is exchanged and shared through a Web Platform.