Introduction

This is a record of the paper presented by Mahes Visvalingam at GIScience 2000 on October 29, 2000 at Savannah, USA.

A detailed paper based on the theme of this talk will appear in The Visual Computer.

The title of my talk identifies three challenges which my research students and I have been addressing.

Firstly, we regard line generalisation as an outstanding challenge. Some leaders in the field regard this as a solved problem. However, as far as we are concerned, we have not exhausted the research potential of even the line filtering algorithms.

Secondly, the evaluation of algorithms is another unsolved problem. Visual assessment is necessary but it is subjective. On the other hand, the current methods for mathematical evaluation become inappropriate as soon as we step beyond approximation. Since my line filtering algorithm was designed to address typification and caricature, the evaluation of algorithms is a continuing challenge for us.

My line filtering algorithm opened up a third challenge, namely the algorithmic sketching of DEMs and of surfaces in general.

This third challenge was inspired by the landscape drawings of Holmes, Lobeck, Raisz and others. dating from the early 20th, 19th and even earlier centuries.

Sketching involves the delineation and depiction, in the universal language of graphics, of the pattern of landforms. This sketch by Holmes illustrates that sketching involves the :

Abstraction of silhouettes to identify the main forms and their relationships in space;

Abstraction of other form-defining lines, such as the edges of the plateaus in the distance;

Completion of the sketch using a set of neat lines arranged almost in profile to bring out the shape of forms.

Lobeck, and others, regarded field sketching as a form of generalisation.

With the advent of photorealism and GIS, there has been little interest in landscape sketching.

I was keen to revive interest in the dying art of landscape drawing. Kurt Dowson experimented with some algorithms for sketching DEMs in his PhD project, completed in 1994.

An overview of our research on sketching is provided in the web-record of an invited exhibition, entitled Art in Scientific Visualisation of Terrain Data, mounted at the UK Royal Institution.

Now that the challenge of photorealism has been largely met, there is a surge of interest in non-photorealistic rendering (NPR) at several sites in America and Europe. Jim Foley (2000), the principal author of the most popular text on computer graphics, has listed abstraction and sketching among the top ten problems left for computer graphics in year 2000.

There have been periodic papers on sketching in computer graphics conferences but NPAR 2000 was entirely devoted to Non-photorealistic Animation and Rendering. This conference was co-sponsored by SIGGRAPH and the Eurographics Association.