The Beginnings

Figure: G-Buffers (left) in conjunction with unidirectionally hatched objects (middle) are used to produce hatched illustrations (right)

One of the first approaches that tried to render images interactively in line-art style was on "Comprehensible rendering of 3D-Shapes" [#!saito90comprehensible!#]: By applying 2D image processing operations (e.g. edge enhancement) to a set of projected images (the so-called geometric buffers or 'G-Buffers'), contour lines, edges and hatches can be visualized (see Figure ). A similar approach which uses half-toning as such an image processing operation was presented in [#!EVL-1998-467!#].

It is crucial to mention that this approach did not try to copy the way in which artists work, but concentrated on the operations necessary to produce a more illustrative look, for example in CAD programs. Similar approaches that change the shading model for that purpose were presented in [#!gooch99interactive!#,#!gooch98lighting!#] (see Figure ). Information flow in such a sketchy modeler is discussed in [#!strithotte94how!#] and [#!schumann96assessing!#], enhancement of ordinary renderings with line-art illustrations in [#!dooley90automatic!#].

"Computer-Generated Pen-and-Ink Illustration" [#!CONF_CG-5!#] was the first paper which investigates how pen-and-ink illustrations are made by artists.

Figure: Strokes indicate tone as well as texture. The more strokes are blended, the darker the tone is.

A Stroke is produced by moving a nib with varying pressure along a slightly jagged path. Multiple strokes form a stroke texture (see Figure ), and multiple stroke textures are painted over eachother to produce a desired tone. If the scale of the texture is increased, more strokes are necessary to produce the same tone.

Figure: A house rendered with outlines

Outlines are special types of strokes which give a hint of the size and shape of an object and emphasize areas of interest (see Figure ): Boundary outlines surround an object, while Interior outlines are used to express shadow direction and give view-dependent accents to the illustration.