A physical model, resembling an arbitrary terrain was created. Qualities that are computationally expensive, yet physically effortless were prioritized. After playful experimentation, the complete work space was photographed from multiple angles. These photographs were used via photogrammetry in order to recreate the work space in animation software.
The physical model was cleaned up and isolated for digital experimentation
The next step was to generate random data to embed into the model. Similar in nature to a geographical information system.
Arbitrary values for wind speed and direction were randomly generated, emulating data collected from 7 observation points. The values were interpolated via an inverse distance weighting algorithm in order to create maps. These maps were kept as the basis for field data, directly affecting the motion of the swarm.
Morphing rules were developed in order to emulate the adaptation behavior of each unit. These rules were utilized as key frames in order to interpolate in-between states.
Morph targets, orientation and morph amount were driven by values, obtained through the wind maps embedded to the digital terrain.
While this approach provided a direct connection between data and parameters, results were static. Positions of the units could only be instantiated for a single moment.
Marcel Duchamp,
Nude Descending a Staircase, No. 2 (1912)
Philadelphia Museum of Art, Louise and Walter Arensberg Collection
Etiene Jules Marey, Man Walking,
1890-1891
Geoffrey Mann, Long Exposure, 2015
For further experiments, a form-making method, similar to long exposure photography was adopted. This approach is congruent with natural ways of shaping things. by directly outputting g-code, it is possible to choreograph digital fabrication tools. Since motion can be naturally changed and rearranged, the method provides an amount flexibility and fluidity that is uncommon in CAD. Rapid gestures can be translated to form, similar to analogue sketches, while maintaining an explicit hierarchy of nodes. This workflow made it possible to create a volume of intricate shapes, that can be redrawn with ease while the virtual matter reacts to virtual forces, in similar fashion to watercolor and gravity.
Inverse Kinematics was exploited to design units, that can react to the changing values of data. Any geometric shape can be inserted into the skeleton structure, and geometric attributes can be dynamically connected to locally sampled data values.
Through replication of the units, a swarm was assembled, working on the principles of the boids algorithm by Craig Reynolds. Three simple rules are defined: repulsion, orientation, and attraction. Since these rules are defined as attributes, their values can also change in time. The swarm may change characteristics depending on context.
Resulting swarms are self-organizing, and demonstrate emergence. By using each unit as a source for the long exposure method, forms resembling both characteristics of contextual data, and gestures of the user are obtained.
Results were video mapped onto the physical model for presentation. Demonstrating the dynamic relationship between the swarm and the surface that affected its formation.
Results of the study were presented at the 4th International Symposium, Formal Methods in Architecture in Porto / Portugal