[Imager Theses and Major Essays] [Imager] [UBC Computer Science]


Imager

Marcelo Walter

Integration of Complex Shapes and Natural Patterns


Degree:  Ph.D.
Type:  thesis
Year:  1998
Supervisor: Alain Fournier
Electronic:  [PDF], 2102561 bytes
Hardcopy: 186 pages

Abstract

The process of generating an image for a computer graphics object is traditionally broken down into three steps: modelling of the shape or geometric attributes (such as height, width, and length), modelling of the visual attributes (how the object is going to look), and an integration step that connects the first two (a visual attribute is defined for every point on the surface of the object). The separation of modelling the shape from modelling the visual attributes makes the whole process highly flexible and powerful; from a conceptual point of view, the process is easier to handle.

While generally good for many classes of objects, this separation is prone to problems when the geometry of the object is complex. For example, the mapping of visual characteristics to every point of such complex surfaces is non-trivial. Furthermore, this separation assumes that these two steps are independent of each other, but for some objects, there is an interaction between the shape modelling and visual modelling that plays a significant role on the final image. Typical examples are patterned animals such as giraffes and leopards, where the pattern visible on the fur of an adult animal is the result of a process that took place while the animal was an embryo in the womb. In this case, modelling the interplay between the embryo growth process and the pattern formation process is as important as modelling the individual processes themselves.

In this thesis we introduce a novel solution for integrating shape and visual modelling. This solution defines the visual attributes directly on the surface of the object as the object changes shape, for example, due to growth. We present results of applying this solution to a giraffe model.

This thesis makes three contributions: (1) a new model of mammalian pattern formation called Clonal Mosaic, suitable for computer graphics purposes and with strong biological plausibility. The model is based on cell division and cell-to-cell interactions, and it can generate repeating spotted and striped patterns occurring in several species of mammals, especially the big cats and giraffes; (2) a technique to modify the shape of an object based, for example, on a small set of input measurements. The technique consists of defining local coordinate systems (cylinders) around the growing parts of the body, each one being transformed according to the relevant growth data while maintaining their relationship with the adjoining parts and the continuity of the surface. The local coordinates also permit ordinary animation mainly as relative rotation such as in articulated objects; and, (3) the integration of the modelling of Clonal Mosaic patterns with the shape modification technique.

Finally, this thesis advances the notion of integration of independent tools as an important development in the field of computer graphics. Individual tools have been reaching exceptional levels of performance and therefore we need efficient ways to integrate them smoothly.


@PhdThesis{Walter1998,
	author = {Marcelo Walter, Ph.D},
	title = {Integration of Complex Shapes and Natural Patterns},
	school = {UBC},
	year = {1998},
	supervisor = {Alain Fournier},
}