Image Processing Reference
In-Depth Information
success of a visualization technique, this is not normally the primary motivation for
its creation. Media artists, on the other hand, generally place aesthetic considerations
at the forefront of their concerns. Similarly, physical simulations are concerned with
accuracy and realism rather than extensibility and interaction, which are central to
media arts. However, media arts projects, due to time constraints or limitations in
technical knowledge, often incorporate readily-available techniques not originally
intended for artistic production, and therefore not as easily adaptable to artistic sit-
uations. Fluid Automata provides a flexible system that can be used in many situa-
tions, including on less powerful mobile devices, while retaining enough realism to
be useful for simulations. Figure 13.1 shows examples of different outputs created
with the Fluid Automata system.
13.3
Simulation and Visualization in the
Fluid Automata
Project
The Fluid Automata system is composed of three integrated components: a fluid
simulation engine; a flow visualization technique; and an interface that encourages
interaction with the visualized fluid simulation [25]. Here I look at the technical
details of the fluid simulation and flow visualization components that are shared
across multiple deployments of the project.
13.3.1
Fluid Simulation
Fluid Automata introduces a novel fluid simulation technique, inspired by CA sys-
tems. While its title was inspired by Tibor Gánti's discussions of “chemotons”
[31, 32], the initial kernel of insight for the Fluid Automata system arose from think-
ing about how to create a simple rule-based system to produce emergent behavior.
CA systems demonstrate complex behavior emerging through an iterative system
that updates the state of each element (positioned in a uniform grid) based on a
set of basic rules. These rules determine the next state of each element by querying
each of its neighbors. In Conway's original automata system, each pixel has a binary
state, and either “lives” (is set to 1) or “dies” (is set to 0) based upon the number
of surrounding pixels that are on or off. Other CA systems, including many that
are examined in Stephen Wolfram's “A New Kind of Science” [64], utilize differ-
ent rule sets and involve multiple states. In particular, a range of CA approaches to
fluid simulation have been introduced. For instance, an early paper by Gerard Vich-
niac describes discretized models for fluid simulation based on CA, but indicates
that precise values for specific parameters are necessary in order to avoid “diver-
gences” [60]. Another paper surveys different CA paradigms for fluid simulation,
comparing lattice gas, digital fluid, and lattice-Boltzmann strategies [16]. A 1998
paper describes 3D fluids that can be represented with hexagonal and rhombodo-
decahedron grids instead of uniform squares or cubes [11]. Stephen Wolfram also
explored general non-linear approximations of fluids via CA [63].
Search WWH ::
Custom Search