Parametric modelling 4

Conclusions

We conclude this session with a fairly developed parametric design applying a wide range of core techniques introduced earlier in abstract manner. The goal was to show how general ideas about geometry and sequence processing become quite tangible when confronted with specific design challenges.

Hierarchical Design

Implicitly the exposition guided you to a process of building hierarchically ever more complex logical constructs: we started with a spine, we constructed a skeleton over it, then we fleshed it with dress-up geometry etc. This was neither accidental nor innate to parametric design.

It is a “top-down” approach to design, which progresses from more abstract, or coarse, to more specific, or fine topics. This allow us to retain control of the bigger picture, eg. the overall design gesture is embodies by the spine, and progressively compartmentalize complexity into finer and finer layers.

This is a general strategy independent of whether the aim is to design physical structures, computer software or mathematical models.

Incremental Generalization

Another aspect which may be equally useful as a strategy is the idea of incremental generalization. We started with a rigid rotating frame and progressively added more and more control in our model such that can achieve broader ranges of designs. With the ability to shape the progression of frame sizes we were able to span a larger range of designs, the constant size being only one specific instance thereof.

The lesson here is that starting directly from the most general version of a design may be overwhelming sometimes. Also consider that in most cases we are required to create a design, the best design, not every design. Augmenting design step-by-step is pragmatically easier to get things going. However, there is a cost involved in the backtracking and reestablishing relationships, for the sake of generalizing which sometimes is not as trivial.

Deferring Design Choice

The core value of parametric design is in the ability to not have to deal immediately with specific dimensions. Incidentally, “parametric” etymologically means deferred measurement. We thus do not have to commit to, but instead we can postpone making certain design choices, eg. how large or small, how few or many. We work in relative instead of absolute terms.

Traditional design media were often so laborious, imagine having to draw all those lines by hand for a moment, that strongly mandated making those choices upfront. Compounding complexity at each design phase often meant that those upstream choices were pretty much set in stone for avoidance of rework. The cost of change was too high. Now consider also that the most important or perhaps high-level choices in design are made early where the least information is available.

By building relationship among geometries, instead of geometries themselves, we retained the ability to modify key design features at will and at zero cost. Here we pushed all key parameters at the left-most side of the graph to highlight this. In addition, we were also able integrate performance measuring processes and evolve design options informed of their further downstream implications.

Exercising Design Schemes

Building complex models can become an obsession. At some point, or intermittently, we need to develop the habit of switching hats between modeling ie. generalizing or making the model more expressive, designing ie. finding unique interesting instances and contextualizing against human, cultural, urban and environmental parameters, and engineering ie. simplifying and improving quantitative performance metrics. For those who want to practise modelling another complex structure, you may try Modular Design.