CONTRIBUTOR	invited speaker speaker	refereed paper artist/tech. talk	panel chair panellist	artwork music	co-authors collaborators
Grant Dunlop & Andrew Maher					Mark Burry, Jane Burry

Spatial Information Architecture Laboratory School of Architecture + Design RMIT University Melbourne, Australia	{} {}
Grant.Dunlop@rmit.edu.au

TITLE: Paramorphs: An Alternative Generative System

ABSTRACT: The ongoing series of Paramorph projects provides an interesting alternative to the application of a 'pure' approach to generative processes within the context of architectural design and practice. These projects take advantage of the potential of parametric modelling or associative geometry software to construct variable geometric computer models, from which a wide range of formal possibilities may be generated. The models that are produced differ from conventional models, possessing an underlying structure and logic that may then be iteratively explored. They are not so much forms themselves, but the possibility of form. Such an approach to generative design, where the code takes the form of a geometric construction rather than the abstraction of a computer program, is one that is, arguably, more suited to the nature and demands of architecture and its design. This technique provides the opportunity for direct visual control and engagement with the model, something which current design thinking and practice within architecture favours.

paramorphs [fig1] (click to enlarge)

As a method for developing geometric computer models, parametric modeling differs greatly from traditional methods. It may also be viewed as occupying an intermediate position between conventional modeling and the generation of architectural form through the use of genetic algorithms and the like. The independence of 'code' and 'product' present in many examples of generative processes is avoided in parametric design.

With a generative approach to creative production it is possible for there to be a clear distinction between what is generated and what generates; between the code and the resultant objects. With a parametric or associative geometry approach, the underlying structure or system (or code) is manually constructed and forms an integral component to what may then be produced.

Alternatively, the structure of a parametric model, in the form of the various geometric entities and their associated relationships, may be seen as performing the same function as the programmed code in a 'pure' generative system. The difference is obviously in the language in which the structure is produced; its level of abstraction (e.g. computer programming code, C++, versus geometric modelling actions and elements), and the manner in which the structure is produced (e.g. computer programming, compiling and execution, versus direct or manual computer modelling).

In this particular case, the 'generative code' is the result of the ordered recording of a number of 'design acts' – a construction history – and which is integral to the resultant model. Rather than explicitly defining a single geometric model (or description, as required by traditional means), parametric design offers the possibility of form – the object is defined by what it could be rather than what it specifically is.

Latent within the base model is a family of forms – related but different – which may then be re-visited and explored further. Parametric design is an approach foreign to a purely accidental attitude to creative production. The freedom and potential of the model is dependent upon the schema adopted at the outset and the nature of the geometric entities and their relationships chosen to construct the model. It is predicated upon having a developed understanding of the design to the extent that its constituent parameters can be recorded historically along with relationships between geometrical entities. The underlying parametric structure is carefully constructed so as to be able accommodate as wide a range of possible variations. The direction of exploration and the variations generated may be either directed or free. A directed approach may be adopted when an existing design has been developed to the extent where the primary consideration is the development or resolution of that concept, carried out through a highly resolved and specialized model. A free exploration would typically generate a wide range of forms from the original model, testing the formal potential of that schema.

In the illustrated example, a circle 'C' of radius 'r' can be linked by a line 'L' of length 'l' between the circle's centre down to the top right hand corner of a rectangle 'R' measuring 'x' by 'y' lying below and to the left of the circle. 'C' is an 'object', and 'c' is a parameter [fig. 1]. In a carefully considered design, a so-called explicit design, parameters have declared values. In almost all CAD packages the design is explicit, regardless of how the software user might have produced it. 'Grip points' give the impression of parametric variability – which to an extent they are. The user 'grips' the centre of 'C' at 'c', and moves it with the mouse to a new position. But the line 'L' remains in place in its original position. If the geometry is associated the situation changes quite distinctly. If entity 'C' for example moves through the redefinition of the position of 'c', the end of line 'L' whose end is linked to 'c' moves with it while remaining attached to the rectangle [fig. 2].

paramorphs [fig2] (click to enlarge)

The precursor to Paramorphs I and II, was a variable surface controlled by a series of spline curves, which were themselves defined by a grid of endpoints from a number of variable lines. This simple exercise was a semi-ironic response to the willful exploitation and abuse of complex modeling and animation software by a number of architects at that time. Paramorph I was a natural continuation of that earlier work; moving from surface to volume, with an accompanying increase in complexity. The intention of the 'dumb box' was to begin to explore the potential of such an approach for the generation of complex architectural form, and what opportunities existed for employing such a technique as a design tool at an early stage.

Paramorph II, or the Sutton Place Gateway to the South Bank, was a collaborative project carried out with dECOi architect(e)s in Paris, France. The primary requirement for this project was to provide accurate and meaningful information regarding the geometric nature of the design materials. This involved being able to capture and recreate the design intent and formal or gestural nature of the early design work, but within a highly controlled environment. The underlying geometry of the original models is recorded and transposed to the "parametric tube" constructed for the project. From this a number of surfacing regimes (such as free form NURBS, facetted or ruled surfaces) may be extracted.

This method of working accommodates some of the common difficulties of such a dispersed and intensive collaboration. With the base "parametric tube" constructed, it is then possible to 'drive' the model using text based command scripts which alter the values of the appropriate parametric values and relationships and allow the model to be reconfigured in the desired manner.

Through the Paramorph experiments the parametric approach has proved its metal both as a framework for a very free generative tool and as a highly controlled means of design refinement and precise geometrical description for construction. Thus it holds out promise for future exploration into different ways of setting up the underlying parametric structure, and for beginning to exploit the latent potential of these forms through their integration or connection to other systems of control or generation. Perhaps these Paramorphs may be driven by other independent but connected generative tools and systems, responding to their various inputs or stimuli, including those of a 'real' architectural nature.

ABOUT: The Spatial Information Architecture Laboratory draws on students, academic staff, and industry led researchers towards design enquiry, speculation, and realisation in three core areas, each servicing the other:

• design philosophy;
• total design culture; and
• interdisciplinary design practice.

Activities that can only take place within an academic environment are encouraged within the laboratory in the context of, and drawing intellectual comfort from, practical relationships with industry and practice.

aegis sequence (at CEBIT)

main	contributor list