Frederic Fol Leymarie

(Dept. of Computing Goldsmiths College, Univ. of London, UK)

Medial Representations for Driving the Architectural Creative Process

Medial representations of shape provide a powerful framework for the analysis and genesis of architectural forms, layouts, landscapes, cityscapes. In this paper we explore their potential use in driving the architectural creative process for 2D and 3D applications. We consider both the aspects of (i) the analysis of exisiting architectural layouts, and (ii) the genesis of novel ones. The archetypal medial representation of shape is the “medial axis” of Harry Blum (circa 1960) [4, 34].

Boundary elements, the outline samples of the objects of interest, are used as the source of a wavefront propagation, for which the quenching points constitute axial symmetries. In 2D layout applications the resulting medial axis (MA) takes the form of a graph which unites geometry and topology of the objects and the field they occupy in one single framework (Fig.1). In the following we first survey the topic of shape representation via medial structures and consider their specific use in architectural analysis and genesis. We then report of early works in extending such ideas for novel applications relevant to architecture. Nearly 50% percent of the world population now lives in cities (closer to 75% in industrialized countries), up from 4% in 1800 and 14% in 1900 [5].

The needs for the 3D modelling of large cities to support efficient urban information systems are present and growing, in particular in the domains of urban planning and management, civil protection, environment surveillance and crisis mitigation, and sensor network modelling. One key aspect that requires new technological development is in the management of reconstructed 3D scenery in a geographical context [30]. This in turns calls for a representation of 3D data permitting spatial queries going beyond today’s essentially 2D GIS (Geographical Information Systems) [29]. Spatial queries in a 3D graph structure have been proposed by Jiyeong Lee et al. to be based on the “straight” MA, i.e., a simplified medial axis restricted to the description of polygonal layouts [27, 28, 24].

One advantage of abstracting 3D volumes via a medial graph structure is to allow for a hierarchical organization of the data at multiple scales, which is supported by studies on human abstraction of geographic space [10]. In the system of Jiyeong Lee et al., a series of 2DMA for the communication network (e.g., the hallway structure), augmented to capture all useful horizontal connectivities (e.g., between rooms), are stacked-up and connected vertically to capture the spatial relationships between 3D entities of an entire building. Each augmented 2DMA represents a level in a building, i.e., typically a floor layout.1 This 3D GIS has been augmented to integrate a ground transportation system together with the hierarchical representation of buildings, in the context of emergency responses in times of crisis (such as due to natural disasters, fires, terrorism).

Frederic Fol Leimarie

(Computing Dept., Goldsmiths College)

Professor of Computing Frederic Fol Leymarie is codirector of the Post-Graduate program in Computer Games and Entertainment at Goldsmiths College. He previously directed the MSc Arts Computing.

He received his B.Eng. in Electrical Engineering, with honors in aeronautics, from the Polytechnic School of Montreal, his M.Eng. from McGill University (at CIM) in Computer Vision and Biomedical imagery, and his Ph.D. from Brown University.

His current research interests incorporate ideas from computer vision, together with the physics of waves and shocks and their modelling in modern mathematics via singularity theory. Frederic is also working on perceptual models grounded in geometry, based in part on Gestalt theory.

Frederic has initiated several “shape-based” projects mixing the Arts, Humanities, Social Sciences, and Computing, including CyberCity and CyberMonument, Digital sculpting (with the Mid-Ocean Studio), and Digital archaeology (co-founder of the SHAPE lab. at Brown University).