Academic journal article Cartography and Geographic Information Science

Automating Landscape Illustration with Pen and Ink Style Rendering

Academic journal article Cartography and Geographic Information Science

Automating Landscape Illustration with Pen and Ink Style Rendering

Article excerpt

Introduction

The art of perspective landscape illustration has a long history in the study and description of landforms and their underlying geological structures. Imhof (2007, pp. 1-14) traced its development from Mesopotamian relief representations through Renaissance copper engravings and on to modern illustrations derived from topographic map construction. Flemer (1895) provided an exhaustive treatment of the theory and photographic technology supporting ground-based, perspective imaging of landforms. His work specifies methods for capturing panoramic images based on numerous individual photographs with known fields of view, and for constructing topographic maps from the panoramas. Raisz, himself one of the foremost practitioners of block and landform diagram construction, attributed the modern introduction and perfection of this form to the work of Grove Karl Gilbert and William Morris Davis (Raisz 1948, p. 301-308). Armin Lobeck, whose Block Diagrams has remained one of the most influential guides to their manual construction, argued that these pen-and-ink illustrations gain much of their effectiveness from their visual simplicity (Lobeck 1958, p. 1). This research explores the application of Lobeck's aesthetic to the automatic construction of perspective landscape illustrations from digital elevation models (DEMs) under the rubric of non-photorealistic rendering. It reviews earlier work on automated systems, introduces a new testing platform for stylistic experimentation, presents the relevant underlying algorithms for the production of the major visual elements, and discusses the contributions of these elements to good expression of landscape form.

Previous Related Work

Over the past decade, the rapid development cycle of fast, relatively inexpensive graphics processing chips, along with concomitant developments in 3D application programming interfaces (APIs), has fostered a period of intense growth in perspective rendering applications in cartography, GIS, and in non-geographic disciplines. Much of this work has been related to realistic renderings of landscapes, whether for gaming, site planning, or other uses that require a relatively "complete" description of a scene. Kennelly and Kimerling (2006) noted that another tendering paradigm, non-photorealistic rendering (NPR), provides a close analogy to the construction of traditional landscape line drawings by fostering the communication of specific, often scientific information through a language composed of symbols with well understood meanings. The goals of NPR are consistent with those of cartographic forms that use the rules and constraints of generalization to establish a message for an image and to focus a viewer's attention upon that message.

[FIGURE 1 OMITTED]

Saito and Takahashi (1990) laid out many of the ground rules for NPR in image space, using information from projected 2D images of 3D objects. Their successful demonstration of techniques for generating simplified line drawings from photographed 3D objects established a vocabulary for further NPR research on the extraction of edges (rendered as silhouette lines) and components of surface curvature, represented by hachures. Isenberg et al. (2003) provided a comprehensive classification of NPR modeling techniques for both image and object space approaches along with definitions of terminology. Kennelly and Kimerling (2006) provided an excellent overview of NPR approaches that are specific to renderings of landforms, placing them in the context of the manual methods defined and described by Raisz, Imhof, Lobeck, and other 20th century cartographers and illustrators.

Most current NPR research in landform representation has focused on object space approaches that derive perspective line-drawn images entirely from 3D spatial data represented in a digital elevation model (DEM). Lesage and Visvalingam (2002) used this approach to implement p-stroke sketching, which is itself closely related to the inclined contour method of Tanaka (1932). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.