Digital Grotesque II (2017)
Digital Grotesque II - a full-scale 3D printed grotto - premiered at Centre Pompidou’s Imprimer le monde exhibition.
Digital Grotesque II explores a new relationship between designer and computer, in which the computer turns from a passive instrument to an active partner who can expand the imagination of the designer. The computer learns to generate architectural structures which evoke interest, curiosity, and an emotional response. A single volume spawns millions of branches, growing and folding again and again. Hundreds of square meters of surface are compressed into a 3.5 meter high block that is brought to life as a seven ton, 3D printed sandstone structure.
The resulting architecture is at once disorientating, intriguing, and evocative without being prescriptive. It inhabits a space between the man-made and the natural, between order and chaos, offering unexpected moments of surprise.
New Fabrication Methods Require New Design Tools
Today, we can fabricate anything. Digital fabrication now functions at both the micro and macro scales, combining multiple materials, and using different materialization processes. Complexity and customization are no longer impediments in design.
While we can fabricate anything, design arguably appears confined our instruments of design: we can only design what we can directly represent.
If one looks at 3D printed artefacts, there is oftentimes discrepancy between the magic of digital fabrication and the conventionalism of the printed objects.
What is needed is a new type of design instrument. We need tools for search and exploration, rather than simply control and execution. We require tools that go beyond the fulfillment or optimization of simple functional requirements, and that allow us investigate and advance more ambiguous factors of the design: soft criteria.
Digital Grotesque II was designed for a specific fabrication technology: 3D binderjet printing. This technology has the potential to produce geometry that cannot be created with traditional subtractive processes such as milling. It can produce forms with multiple layers of skins, nested branching, and recursive levels of porosity. In short: 3D binderjet printing can print not only articulated surfaces or topographies, but also forms with complex topologies.
As architects, we arguably lack the tools to design - or perhaps even to even imagine - such topological forms. Neither with pen and paper, nor by using common CAD programs, can we easily design forms with complex topologies. Most software packages use a boundary representation of form, in which a surface representing a single border between inside and outside is manipulated.
Digital Grotesque II, also uses a boundary representation of form, yet it overcomes some of its limitations by using an expanded topological subdivision process. When using standard subdivision processes to articulate form (as in the precursor project, Digital Grotesque I), the form always stays within its genus. A cube can be subdivided to form a sphere, but not to form a donut. For this project, the subdivision algorithms were expanded to allow a transition from one genus to another. Using these processes, a cube can be subdivided into a form with thousands of holes at many different scales. A topological design becomes entirely feasible.
Soft Criteria: Predicting Experience
Soft design criteria such as aesthetic values are highly relevant for the interplay between humans and the built environment, as they influence the emotions that people feel through the architecture. One can distinguish between aesthetic emotions (e.g. being moved, fascination, and awe) and epistemic emotions with deal with curiosity, interest and insight. Furthermore, one can qualify emotions of amusement, humor and joy and other that deal with energy or relaxation.
These criteria are nearly impossible to quantify and to formulate as a set of rules or program. We have therefore sought to identify geometric properties of architectural form that can be measured and correlated with human perception and desirable emotions. Three measures are identified as the principal indicators:
- Experience-ability: changes in the form’s appearance or visibility through a shift of viewer’s position or perspective.
- Compressibility: the degree of geometric information that cannot be simplified through symmetry groups.
- Depth-complexity: measuring the topological alternation between void and solid (derived from ray-tracing)
Learning to Imagine
In Digital Grotesque, instead of explicitly programming a set of predefined rule to reach these design criteria, the computer learns to calibrate the design with the goal of evoking stimulation and interest the beholder.
We train the computer using design variations that have been evaluated online by hundreds of volunteers. The abstract geometric properties identified above are paired with statistical data measuring viewer preference, as quantified through number of views, viewing duration, sharing, etc. Through this learning process the computers constructs a model of the relation between abstract geometric properties and the human perception.
The process functions without any architectural precedents and without direct control of the form. The computer program is able to actively design variations which are undrawable and unimaginable; which a human designer could not conceive of in their detail and differentiation. The role of the designer becomes one of a curator, steering the process and defining the appropriate design goals.
In the design process of Digital Grotesque, the computer proposed thousands of design solutions optimized to evoke an emotional response in the beholder. We learned about new design possibilities at each step in an iterative process. We selected specific solutions and amalgamated them into a singular form.
This new human-machine interaction frees the designer from thinking in terms of archetypes and categories, and ultimately serves to expand the designers’ imagination and creativity.
An Imaginary Space
The resulting architecture is at once disorientating, intriguing, and evocative without being prescriptive. It inhabits a space between the natural and artificial, between order and chaos, offering unexpected moments of surprise.
Standing in front of Digital Grotesque, one is struck by a hitherto unseen richness of detail that is at times overwhelming. Although designed entirely autonomously by the computer program, the result evokes analogies in multiple domains, from existing architectural styles to natural formations. It raises the question about the ultimate origin of the form.
Digital Grotesque is a testament to and a celebration of a new kind of architecture that leaves behind traditional paradigms of rationalization and standardization and instead emphasizes the viewer’s perception - evoking marvel, curiosity and bewilderment.
Digital Grotesque II in Figures
Virtual:
- 16,386 design variations
- 260 million individual surfaces
- 42 billion voxels
- 156 GB production data
- Calculated on ETH EULER high-performance cluster (2000 GB Ram)
Physical:
- Sand-printed elements (silicate+binder)
- 3.45 meters height
- 537 m2 total surface area
- 7 metric tons weight
- 280 µm layer resolution
- 4.0 x 2.0 x 1.0 meter maximum print space
Design Development: Printing: Assembly: |
2 years 1 month 2 days |
Team
| Fabrication and assembly: |
Michael Thomas Philippe Steiner Allegra Stucki Florentin Duelli Jan Francisco Anduaga Katharina Wepler Lorenz Brunnner Nicolas Harter Dominik Keller Max Spett Alexander Canario |
|
| Digital support: |
Matthias Leschok Alvaro Lopez |
|
| Photos: | Fabrice Dall'Anese Michael Lyrenmann Demetris Shammas Jann Erhard Hyunchul Kwon |
|
| Video: | Kaufmann & Gehring | |
Partners and Sponsors
| Chair for Digital Building Technologies, ETH Zurich |  |
| Department of Architecture, ETH Zurich |  |
| Centre Pompidou |  |
| Christenguss AG |  |
| Bosshard + Co. AG |  |
| Suter Elektro AG |  |
The geometry was calculated on the Euler High-Performance Computing Cluster at ETH Zurich. Components were printed on ExOne printers by Christenguss AG.