August 2010

I am very pleased to announce that Robert Cervellione has nearly finished porting the Kangaroo Physics engine to work with Generative Components.
To read more and see some videos, visit

I’m excited to see how it gets used in this different software environment.
Thank you Robert!
Development for both the GC and GH versions will continue in parallel.

Also coming soon – a new version of Kangaroo for GH, and a first draft of the (long awaited) manual.


Much of my recent work has been about form-finding in terms of starting with some geometry and making small changes to lengths and angles to achieve a shape which is in some way optimal (e.g. all members of an inverted catenary vault are in compression, or mean curvature is minimized in a fabric canopy).

But another way of using these simulation techniques is to form-find not only the geometry, but also the topology of a structure – the overall arrangement of elements and which connects to which.

Here I think much can be learned by looking to studies of the microscopic world.
When writing Kangaroo I drew on some computational techniques which have found popularity in the game and animation industries and tried to bring them to the architectural design world. But many of these techniques were actually first developed in the field of Molecular Dynamics. So it seems a natural step to try exploring some ideas related to molecular modelling in Kangaroo.

Self-organization is the process by which basic elements can arrange themselves into more complex structures through simple local interactions.
(See also the closely related but more specific term self-assembly)
It occurs at many different scales – from the formation of molecules to the clustering of galaxies.

All of the videos in this post are purely procedural – I am not controlling them at all beyond setting the initial positions and some simple distance dependent forces of attraction and repulsion.
These are pretty basic first experiments, and nothing too rich in terms of emergent order is apparent yet, but I am excited about the potential.

Self-organization can be seen as the mechanism by which ‘higher’ domains emerge from lower ones – chemistry from physics, biology from chemistry.

Learning about how materials work at a molecular level might inform how we build with them, but I think the processes could also be abstracted and have relevance at quite different scales.

For example – electrostatic repulsion can be used to find an even distribution of cladding tiles on a doubly curved surface…

Or water surface tension pulling threads together could maybe inform road layouts…

(This last experiment is one that seems to have become quite popular recently – see also the work of Marek Kolodziejczyk, Peter von Buelow, Yiannis Chatzikonstantinou, Danny Holten and Jarke van Wijk, Corneel Cannaerts, StudioMode, and David Reeves)

…What other ways could self-organization be used in design ?
I think there is still much unexplored territory here!

The idea of taking forms from organic and inorganic nature and adapting them for our own designs is certainly not new, but I think what is is the facility and speed with which it is now becoming possible to digitally experiment with the processes of formation in nature.

Changing the way we design to something more process based can require quite a shift of thinking.

Optimization is a powerful tool with lots of exciting potential for finding the ‘best’ solution, but it requires that the designer think carefully about how to define what ‘best’ is in a way that can be communicated to the computer.

It will take a bit of getting used to to not directly design the final destination, but to define an invisible energy landscape and then let the computer climb the hills, sometimes ending up on a peak we had never guessed was there.

Thanks to loop.pH – working with them and discussing their fullerene type structures inspired some of these lines of exploration.
Also thanks to Robert Hodgin for inspiring me with this video,Martin Tamke and Jacob Riiber for this ‘Lamella flock’ video,and Daniel Davis for his swarming & dynamic relaxation on a surface post.

For further reading I recommend the writings of Peter Pearce and Stephen Hyde.
Also – this very nice Self-Organization FAQ from Chris Lucas.