Cartoon-Style renderings

Cartoons simplify shapes and give a wide audience the ability to immerse into a simple, yet powerful form of art. Various papers exist on the topic, mostly dealing with the application of 2D computer graphics to cel animation [#!durand91toon!#,#!lit91animation!#,#!feke95toon!#]. This includes work on the so-called ``Multiperspective Panoramas'' [#!EVL-1997-93!#], which are flat images that give the illusion of a camera traveling through 3D scene (see Figure ).

Figure: A multiperspective panorama from the Film Pinochio

Figure: The Gooch Lighting Model gives a more descriptive look to objects by using gradients for shading

4cm

Other approaches deal with the use of computer graphics for the improvement of cartoon drawings. Examples include a semi-automatic generation of shadows for cartoon scenes [#!petrovic2000!#] and the application of complex textures to hand-drawn objects [#!EVL-1998-431!#]. One of the first approaches with which cartoon rendering could be brought into the third dimension was the shading model by Gooch et al. [#!gooch98lighting!#]. Although not solely meant for use as cartoon shader, most approaches that try to render colored objects in a more non-photorealistic style use this color model. The key to this approach is the introduction of gradients into the shading model. Furthermore, edges are outlined so as to increase the illustrative look (see Figure ).

A complete framework for generating cartoon-style renderings in real-time was presented by Lake et al. [#!lake00stylized!#]. The rendering process is divided into two parts: The painter determines the shading information to fill the polygons, and the inker highlights visible silhouette edges. Heart of the implementation is a new cartoon shading model that uses solid colors which do not vary over the materials they represent (see Figure ). Shading is done with a color that is a darker version of the main material color. Technically, the approach does not smoothly interpolate shading across a model as in Gouraud shading, but finds a transition boundary and shades each side of the boundary with a solid color.

Figure: A mesh rendered in cartoon style
Figure: The ears of the mesh deform instead of just rotating with the head

5cm

The cartoon framework also embodies a pencil-style renderer, as well as the priorly mentioned inker, which detects and paints the silhouette edges of the mesh. Furthermore, motion-lines (lines giving hint of motion) are used to further emphasize animation. Another aspect of cartoon animation are the view-specific distortions of the models [#!EVL-1999-277!#] (see Figure ). There is a deformation on the 3d mesh which cannot be captured with conventional 3d models. Hence, the model has to change its shape autonomously with respect to the eye point.

Another similar idea presented by Rubin et al. [#!SYMP_INTACT_3D-2001-2!#] is to let objects be aware of their importance: Less important items in the scene fade into the background, and are visually grouped together with other objects in order to reduce their prominence (see Figure ). On the other hand, important objects stand out of the background and are thus emphasized (Figure ). This process of selecting areas of high interest in a piece of art is widely known as composition.

Figure:

[Distant objects are abstracted to form a single shape. The color is averaged over all objects.] [An example of counterchange - the tree adjusts its color to contrast with the background]