The aim of the project was to explore Mesh Grammars: a rule-based mesh transformation system proposed by Michael Hansmeyer and Benjamin Dillenburger.

In the project the mesh vertices are evaluated based on few developed rules. The results are then remaped to an unified domain. Next step is to extrude the vertices along normals by the evaluated results. Every few iterations the mesh is subdivided to add a level of detail.

During design process the three evaluation rules have been developed:

Length of incoming mesh edges

Distance form centroid

Curvature

After applying all the rules simonteslusly, the resulting mesh becomes complex, self-intersecting geometry, which would be impossible to design by a human being.

As a proposal of implementation in architecture, the column has been designed using described method.