In this paper, we present and discuss the procedural technology developed i
n the framework of the Image Metropolis project, to model landscapes at reg
ional scale. We call it "IMAGIS" to symbolize the alliance, in one single i
nformation system, of the two most powerful visual media for the representa
tion of a territory: the image and the map.
IMAGIS is an interface between the 2 D world of plans and the 3 D world of
perspectives. It implements a set of procedures specifically designed to tr
ansform landscape forms described in a GIS quite synthetically by 2 D entit
ies, into much more elaborate objects in a 3 D space. It is intended for cr
eating broad landscape models, from which photorealistic images can be comp
uted with Atelier de Modelisation de l'Architecture des Plantes (AMAP) soft
ware.
AMAP is a modeling workshop allowing users to generate and visualize a larg
e number of plant models at various growth stages. AMAP was created from th
e outset for the needs of agronomists who seek to simulate very accurately
the architecture and growth behavior of plants, in a natural or a designed
environment. It is now quite established in the realm of landscape architec
ture.
IMAGIS is aiming at a different community of users: the planning and GIS or
ganizations who need to visualize regional development plans for impact ass
essment, communication and decision-making purposes. Although, it may be ap
plied in a wide variety of contexts, its built-in features makes it particu
larly suitable for suburban planning.
The first, part of the paper sets the rationale for the development of IMAG
IS and focuses on issues of perception and representation of broad landscap
es. The second part provides a short description of AMAP's features and int
roduces IMAGIS' new functions, showing how they widen the application spect
rum of AMAP from the plant to the land cover. Finally, some preliminary con
clusions from Imago Metropolis are presented, that set-up the agenda for fu
ture researches and developments in the field of landscape procedural model
ing. (C) 2001 Elsevier Science B.V. All rights reserved.