Antecedent Prototypes

Artificial Nature is an ongoing research project, evolving through various iterations to gradually approach the conceptual goals with integrity. These iterations are outlined on this page.

PrecursorS

Evosystem, Maya MEL, 2007 Spring A rendered animation utilizing Maya and Mel scripting to evolve a population of organisms as spatial forms. An initial exploration of evolutionary algorithms and a guiding aesthetic prototype for visual and spatial forms. Video link.

Multi-Agent Landscape, Max/MSP/Jitter, 2007 Fall A preliminary multi-agent system in a generative 2.5D landscape, with primitive population dynamics.

Iteration 1: Agents & Dissipative Field

Evo-field, Max/MSP/Jitter, 2007 Fall
In order to model dissipative structures in far-from-equilibrium conditions, independent temporal processing of the morphogen field was introduced. Processing included accretion (simulating the effect of free energy, such as the role of the sun in our own ecosystem), and spatial displacement with regional variation (vital to model dissipative structures, but also suggestive of weather or surface topographies for example). The interactions between such simple processes and the simple cells produces complex evolving images of aesthetic interest. Video link.

Alien-field, Max/MSP/Jitter, 2008 Spring The images below show more complex organism structures (model from the Evosystem) within a dynamic landscape field directly evolved from the Evo-field. Video link.

Iteration 2: Gamogenesis and Propagation

Multicellularity (fish school), Max/MSP/Jitter and Lua AV, 2007 Fall Properties of cells (size, speed, energy) directly impact quantitative aspects of their behavior. Video link.

Gamogenesis & Propagation, Max/MSP/Jitter and Lua AV, 2007 Winter Taking the organisms in abstraction, a series of prototypes produced a minimal model of 'evo-devo' interaction. This incorporates a developmental model of cell specialization according to gene structure, independent cell growth, metabolism and collapse, and an evolutionary model of genetic crossover and mutation between individuals of a population. Qualitative properties of cells derive from cell types simplified to basic functions: sensing, moving, energy exchange and reproduction. Recognizable varieties or races diverge, however speciation is not yet possible. Video link.

Iteration 3: Spatial Communication & Entrainment

Sync (waves of locally excited communication), Max/MSP/Jitter, 2008 Spring
The phenomenon of entrainment, or sync, is used to model interaction between multiple organisms. The classic example of entrainment is the synchronized phosphorescent pulsing of fireflies. The supposed mechanism is relatively simple: each firefly has a natural pulsing period (usually quite closely matched between individuals), but indeterminate phase. Entrainment, or phase alignment, occurs by either suppressing or encouraging the likelihood of a pulse according to the amount of light energy in the vicinity of the organism. We modeled this behavior using parallel collaborative routines (coroutines) per organism, reading from and writing to a global field of luminance. Early experiments are limited to frame-rate, however current exploration extends to audio-rate parallelism in the sync periods. Video link.

Iteration 4: Parametric Geometry

Parametric Geometry (the organism), LuaAV, 2008 Spring
The images below show a parametric geometry modified from boy surface functions. As regional field quantities change in three-dimensional space, the mobile behaviors and phyisical forms of the organisms change according to input parameters.

Iteration 5: From 2D to 3D world

Our ecosystem began with a 2D spatial field of pseudo-chemical elements, concentrations or morphogens (represented as RGB colors). The desire had always been to transition to a fully 3D space, for an immersive experience (looking from within rather than looking upon from outside). However, visualization of 3D fields is non-trivial!

Alien-field, Max/MSP/Jitter, 2007 Fall
A refinement of the original evo-field above, with a height-map. Rate of energetic replenishment is greatest on the peaks, while diffusion tends towards the valleys, leading to populations battling each other for control of higher ground.

Toroidal environment, C++ Application, 2007 Fall
An alternative mapping of the diffusive landscape onto a continuous surface in a 3D environment; combined with multi-cellular agents from earlier prototypes. This solution was quickly abandoned: the added third dimension had no significance to or variance in the world.

Tree-like network prototypes as immersive structures showing chemical flow, LuaAV, 2008 Spring
Representing the 3D field presented a design challenge: to render visible local variations in chemical concentrations without obscuring distant views, and provide structural variation across various regions of the world. Our first solution appeared in the form of tree- or vine-like network structures These constructions would be visually porous enough to reveal local and distant detail, could vary significantly in different regions, and furthermore offered not only a means to visualize chemical concentrations but also a topology through which matter-energy could flow. A natural extension, coherent with our general strategy, was the treatment of these network structures as living, metabolic entities in the world.

The tree/network structures persist in Artificial Nature, as plant-like organic matter; however the visualization of chemical fields and their diffustion was achieved through discretization into particles flowing in space. Particles can represent chemical content by color, dissipative flow through mass drift, and local variations through densities (given sufficient particles), without obscuring the view of distant space. Furthermore, the collision-space could be used to model attractions and reactions between particles as part of the physical underpinning of the ecosystem.