Project DisCo: Choreographing Discrete Building Blocks
in Virtual Reality
Jan Philipp Drude1, Andrea Rossi2
and Mirco Becker1
1 Leibniz University Hannover, Welfengarten 1, 30167 Hannover, Germany
2 TU Darmstadt, Karolinenplatz 5, 64289 Darmstadt, Germany
Abstract. Current excursions within architectural research are exploring the po-
tential of discrete design strategies at different scales. Starting with the introduc-
tion of the Great Invention Kit (GIK) and the subsequent development of reversi-
ble 3D printing processes based on “digital materials” at the MIT Center for Bits
and Atoms  similar concepts of additive manufacturing have recently entered
the field of architecture. This development hints at the potential for new reversi-
ble fabrication methods , as well as new ways to define architectural shapes as
bottom-up syntactical aggregations of modular building blocks.
Within this emerging field of “Discrete Architecture”, Gilles Retsin show-
cases prototypical architectural designs with his Diamond House among other
projects , also focusing on the possibilities for robotic assembly, while José
Sanchez explores techniques borrowed from game-design to define loose assem-
blies based on their specific “topological diagrams” .
This paper introduces Project DisCo (Discrete Choreography), an application
to integrate bottom-up aggregation of modular building blocks and intuitive spa-
tial design into Virtual Reality (VR). The work presented here builds on
Sanchez’s approach to discrete interactive design within gaming environments,
though it is neither based on a sequential placement of individual parts, nor does
it utilize static vector fields. In contrast, it allows the designer to choreograph
large amounts of building blocks interactively through physics simulations as a
means of form generation.
Keywords: Virtual Reality, discrete assemblies, digital material.
1.1 Virtual Reality
With the advent of commercial VR headsets (Oculus Rift 2012), VR has opened up to
a mass audience. While mainly focused on the gaming market, some tools for geomet-
ric modeling have emerged as well. Concept modeling tools like gravity sketch® are
advanced applications and deliver quick results to the skilled designer, whereas draw-
ing tools like Google's Tilt Brush® have enthused people, especially in the consumer
The ability to paint in three dimensions, exploiting full body movement, quickly
leads to spatial results, although the inherent lack of precision makes it unusable for
anything but concept models. Additionally none of these existing tools is tailored for
architectural design and, while some architectural research is being done, both in the
context of real-time simulation , and in their didactic potential , the choice of tools
for the creation of architectural objects, building on the capabilities of VR and game-
engines, is relatively sparse.
Fig. 1. Project DisCo: Screenshot
To fill this gap, we propose a novel way of generating architectural aggregate structures
inside VR: A choreographing methodology, in which the designer/choreographer is al-
lowed to use intuitive physics-based interactions to assemble discrete modular struc-
This assembly mechanism is achieved by allowing interaction with a field of discon-
nected digital building blocks, initially floating in space freely. By using controller
movements to apply forces to them intuitively, the building blocks are aggregated ac-
cording to their inherent connection logic, allowing the choreographer to control the
overall process while leaving the exact detailing of the assembly to a combination of
pre-defined assembly rules and chance since the sequence of realized connections is
computed by the system.
This process of modeling with virtual physics formally harks back to Greg Lynn’s
thesis of “animate form” , though the use of discrete building blocks allows embed-
ding the discretization needed for assembly directly within the design process, avoiding
the need for complex post-rationalization of the generated form.
2 Project DisCo
2.1 Designing with Digital Material in (programmatic) CAD: Wasp
The initial definition of building blocks and assembly rules builds on Wasp, a Grass-
hopper plug-in for discrete modeling . Wasp enables the creation of descriptions of
each discrete building block, combining geometric and topological information, as well
as the definition of the assembly rules, used to aggregate structures out of basic units.
Building blocks are defined by their geometry as 3D meshes, and by their connections
to other blocks as oriented planes on the faces of the part geometry. Assembly rules are
defined as syntactic statements, specifying which building blocks and which connec-
tions are allowed to connect during aggregation (see Fig. 2). Users are allowed to ag-
gregate structures by applying different algorithms. The main limitation of the approach
proposed by Wasp lies in the fact that design is either fully bottom-up and dependent
on the aggregation rules (when using stochastic aggregation procedures), or entirely
pre-determined in a top-down fashion by a guiding scalar field, based on input geome-
tries or generated density fields. These methods pose limitations to the application of
Wasp in a flexible design process, where intuitive modeling is often at odds with rule-
based processes .
Intending to overcome such limitations, we propose a combination of Wasp’s dis-
crete modeling approach with intuitive design in VR, strengthening the possibilities of
both systems: this improves the accessibility of modular design approaches, by offering
a user-friendly interaction through VR. Concurrently this approach allows for a more
direct link between design in VR and fabrication, by embedding the assembly logic
directly into the design workflow. The choreographer is enabled to create the discrete
system’s connection logic in the familiar design-environment of Rhino and Grasshop-
per® and then export the required information, including both parts geometry and as-
sembly rules, to our VR modeling tool Project DisCo. The export format follows the
same conventions of Wasp, using a .json file for storing aggregation information. This
allows full bi-directional communication between Wasp and DisCo.
Fig. 2. Wasp Rules, Connections, and Geometry
2.2 Choreographing Digital Material in VR: DisCo
To tightly integrate discrete modeling with a more direct and intuitive design approach,
Project DisCo combines the connectivity information provided by Wasp with physics-
based interaction gameplay, allowing to generate discrete spatial structures in a VR
modeling environment quickly.
Fig. 3. Vector Field – Direct Placement – Recursive Placement – Choreography
Gameplay. Within DisCo, the choreographer is prompted with a number of building
blocks floating around in space freely. The default aggregational tool allows the chore-
ographer to position blocks at specific locations as starting points for an assembly.
The primary means of aggregation is the choreography mode, where a set number of
free building blocks become responsive to the movements of the controller. Its direction
and velocity apply forces to the responsive blocks, where they are affected proportion-
ally to the speed of the controller itself. This responsiveness will result in a flock-like
behavior that reacts to the players steering movements, resulting in a globally predict-
able system, though non-predictable local parts behavior (see Fig. 4).
A / a »» B / b
Fig. 4. Controller Direction – Controller Velocity – Drag
Responsive building blocks continuously scan their surroundings for open connec-
tions close-by, snapping to them if conditions of proximity and alignment are met (see
To preserve computer performance, the choreographer can control the number of
affected parts as well as disable specific assembled parts for connection, also resulting
in a mode for influencing the direction of growth.
Further aggregational tools for guided growth are several filters. The choreographer
can disable connection rules following their associated rules grammar, created in Wasp
or select types of building blocks to remain unaffected altogether. Aggregational tools
can be chosen from a tool menu attached to the controller. (see Fig. 5)
Based on Wasp class libraries, DisCo is programmed in Unity software and adds
functionality for real-time behavior and physics, while also defining an exchange pipe-
Fig. 5. Controller Menu for HTC Vive®
END>END SIDE>S IDE SIDE>
Chose Delete Disable
Save Load New
Fig. 6. Process of Connection Snapping
Connection Scanning. Before the start of a session, the choreographer is required to
set up the game area as well as a dimension for its division into a voxel grid, to represent
this space-partition in a three-dimensional array, where connections are stored accord-
ing to their voxel position as soon as a part is placed in the aggregation.
An affected part can thus scan for open connections by referencing its position to an
item in the array and only get the free connections stored there. (see Fig. 7) The stored
connections are then checked against the connections on the affected parts itself, look-
ing for the best fit according to proximity and alignment as well as active rules and
rules grammar. (see Fig. 8) If the best valid fit delivers a combined value of proximity
and alignment below a predefined threshold, the part attempts to realize the connection.
(see Fig. 6) For this, it needs to check against collisions in its new position, only real-
izing it, when there is no overlap with the adjacent parts. In the case of a collision event
taking place, the part is transformed back to its previous position, and the rule is deac-
tivated for future connections.
Fig. 7. Space-Partition Grid
Pipeline. The exchange pipeline with Wasp is fully bidirectional. After designing the
part geometry in Rhino and the connection syntax in Grasshopper Wasp, the choreog-
rapher exports a file from a Grasshopper node to be interpreted by DisCo. This exporter
stores all relevant information from Wasp. Contained in this file are both the parts with
their names, IDs, and connections, as well as valid assembly rules. Furthermore, it in-
cludes mesh geometry information for both representation and collision. DisCo builds
the meshes for these purposes during the initialization phase, while interpreting the ex-
Conversely, once aggregation is completed, DisCo stores a choreographed assembly
in a file that can either be reloaded at a later point or imported into Wasp. This workflow
allows the choreographer to work on multiple assemblies that are then loaded into one
project, as well as working with a simplified version of the part's geometry, that can be
exchanged for a higher resolution model after assembly. An exported assembly can also
serve as a starting point for further aggregation in Wasp. Assembled base geometry can
thus be thickened and grown with a stochastic algorithm in Wasp, growing a roughly
sketched spatial assembly in complexity.
Fig. 8. Connection Filters
Simulation. As a means of testing an assembly for real-world buildability, DisCo fea-
tures the ability to apply gravity to the aggregated geometries, exposing the complete
aggregation to gravitational pull and letting it fall to the ground, to test its stability. (see
Fig. 9) The result can be chosen as the new status quo upon which further to aggregate.
The ability to quickly simulate rigid-body physics during the aggregation process offers
an efficient tool to test the stability of a structure not only when completed, but also
during the assembly phase, providing relevant insights regarding the need for scaffold-
ing or counter-balancing blocks during real-world construction.
The physics simulation capabilities inherent to game engines will further be tested
in the future. The aim will be to implement structural analysis of assemblies to check
forces occurring at the joints.
END>END SIDE>SID ELONGSIDE>
Fig. 9. Simulation Process
Surroundings. In addition to the discrete building blocks and their associated connec-
tions and assembly rules, the pipeline also allows for the import of additional geometry.
The choreographer can thus create a design space in the otherwise disassociated sur-
roundings inside DisCo.
As well as being a scale reference or an inspirational space, this entails the possibility
of building upon modeled parts of the real surroundings.
In combination with the physics simulation, this makes it possible to include fabri-
cation constraints during aggregation, since it allows to model the physical surround-
ings of a real construction site and use the simulation to test stability and location of
It also can help the degree of immersion of the choreographer, since it adds the pos-
sibility of interaction with the physical space around them. This could also mean di-
rectly assembling at the construction site, not only seeing geometrical constraints but
also experiencing the space with other senses.
These sensations would, of course, be amplified by the use of mixed reality glasses.
Though the existing systems do not have the computational power to run the real-time
connection scanning in DisCo, this change of platform can be a goal for the future.
While unique in its application, Project DisCo bears similarities to other existing sys-
tems. Examples for this can be found in existing software solutions as well as in every-
day Objects that follow discrete connection logic.
The LEGO® reference.  One of the simplest and most popular systems for discrete
building blocks is offered by the LEGO Group. While following detailed instructions
towards a finished model is rather easy, depending on the complexity of the build, it is
much harder to design a meaningful LEGO assembly. While children do it hands-on,
the LEGO Group also offered a piece of software for that very purpose. The LEGO
LDD software was a design suite, where users could create LEGO assemblies in a
CAD-like manner, create the instructions and both upload your design and order the
necessary parts for your build online .
Though indeed a gift for enthusiasts, this approach has two huge disadvantages. One
is that it shares with physical LEGO blocks the work-intensive assembly process of
looking for the needed piece and then one after the other placing it in the right position.
The second is the reduction of geometric complexity to isometric views. This problem
is even more relevant for any building blocks leaving the guidance system of an orthog-
In the field of discrete architecture, designers have attempted to overcome such lim-
itations by using different computational ideas to aggregate parts. Methods of growth
range from purely stochastic algorithms to more elaborate parametric approaches using
vector fields or geometry as directional guides.
DisCo wants to add a further more playful and intuitive approach to this challenge.
Basically a system, built around connection syntax, it aims at creating assemblies very
much like the LEGO LDD software, without the rigidity of CAD, but as an engaging
game where one can immerse oneself in the act of making. The most critical disad-
vantage of traditional CAD certainly is the lack of three-dimensional representation that
VR offers to solve. Furthermore, only the introduction of a system where we can ag-
gregate parts much faster than in serially building with single units makes it a viable
option for designing larger structures.
The pace in which a design can be generated in DisCo ranks somewhere between
direct placement and generative solutions, though it offers a format where the choreog-
rapher can be closer and have increased control over the actual process of aggregation
than with conventional computational approaches.
Fig. 10. LEGO® block aggregation
The multibody CAD reference. Another reference to the workings of DisCo would
be 3D CAD with multibody systems, like the ones used in mechanical engineering.
Mainly an assembly is composed of several rigid bodies constrained to one another via
rigid joints. In mechanical engineering, a big part of the design process is utilizing off
the shelf parts to cut cost. Assembling a system out of given parts by constraining them
to one another at assigned connections bears quite some similarities to DisCo. If we
consider the LEGO® approach of placing one building block after another, any multi-
body CAD tool could be used to achieve an assembly like with DisCo more laboriously.
Assigning degrees of freedom to a connection could be a feature for a future release
of DisCo. This feature would make the modeling of kinematic systems possible, either
for the choreographer to be able to change an aggregation or as a means of creating
Towards an open platform. To be a DisCo choreographer, users do not necessarily
need to be proficient with Rhino and Grasshopper. Project DisCo will be provided with
multiple example files, ready to be used to choreograph an aggregation. In the future,
it could well be possible to exchange aggregational systems directly via an associated
online platform and crowdsource design ideas from an audience that is not concerned
with the system's creation, mirroring José Sanchez’s concept of “platform design” .
Fig. 11. Example Aggregations
The discreteness of the system makes it possible to have analog equivalents of the dig-
ital building blocks with the connection logic embedded into their geometry. Through
mass production, many of these building blocks for use in different architectural con-
texts, could be produced in automated manufacturing processes and be brought to the
construction site for assembly.
Since the accuracy of the system is embedded into the discreteness of the parts’ logic,
all manners of different scales are imaginable. From the assembly of small structures
by hand with only a digitally controlled guidance system, all the way to the manufacture
of large structures by swarms of autonomous robots, different means of construction is
imaginable . Since the discreteness of the aggregation can act as a guidance system
through its geometry, small robots could assemble it without expensive sensing mech-
This process not only has the possibility to achieve a fully automated construction
site but could also lead to a democratization of construction, since groups of nonpro-
fessionals could erect even complex structures out of mass-produced parts with embed-
ded assembly logic.
Project DisCo, in its current version, will be tested in depth for its design applications
in the following semester at our M.Sc. program. It is to be understood as a first alpha
version that will be further developed. Improvements in both performance and interac-
tion are already put into a roadmap, and a first public release is available online .
The roadmap also includes additional functionality such as the physics simulations
mentioned above, an exchange platform for discrete systems and their associated Rhino
and Grasshopper files and tools for guided construction. The DisCo export node is also
embedded directly in the current Wasp releases to enable users to migrate designs to
VR quickly. 
Future alterations will bring the application from the virtual into reality by incorpo-
rating structural analysis methods as well as fabrication-aware methodologies.
Fig. 12. Assembly Rendering
1. Popescu, G. A., Kunzler, P., Gershenfeld, N.: Digital Printing of Digital Materials. In: Dig-
ital Fabrication 2006 Final Program and Proceedings, pp. 55-57. Society for Imaging Sci-
ence and Technology, Denver, Colorado (2006).
2. Jenett, B., Cellucci, D., Gregg, C., Cheung, K.: Meso-Scale Digital Material: Modular, Re-
configurable, Lattice-Based Structures. In: Proceedings of the 2016 Manufacturing Science
and Engineering Conference MSEC2016, ASME, Blacksburg, Virginia (2016)
3. Retsin, G.: Discrete Assembly and Digital Materials in Architecture. In: Herneoja, A., Öster-
lund, T., Markkanen, P. (eds.) eCAADe 34 – COMPLEXITY & SIMPLICITY, vol. 1, pp.
143-151. eCAADe, Brussels (2016).
4. Sanchez, J.: Polyomino – “The Missing Topology Mechanic”. In: Kretzer, M., Hovestadt,
L. (eds.) ALIVE – ADVANCES IN ADAPTIVE ARCHITECTURE 2013, Applied Virtu-
ality Book Series, vol. 8, p 8. Birkhäuser Verlag GmbH, Basel (2014).
5. Quinn, G., Galeazzi, A., Schneider, F., Gengnagel, C.: StructVR Virtual Reality Structures.
In: Mueller, C., Adriaenssen, S. (eds.) IASS 2018 BOSTON SYMPOSIUM:
APPROACHES FOR CONCEPTUAL STRUCTURAL DESIGN, Proceedings of IASS An-
nual Symposia, vol. 2018, pp. 1-8. IASS, Boston (2018)
6. Jovanovic, D.: Platform Sandbox – A pedagogical design software. In: Lamere, J., Parreño
Alonso, C. (eds.) ACADIA 2017 – DISCIPLINES DISRUPTION, Projects Catalog of the
37th Annual Conference of the Association for Computer Aided Design in Architecture, pp.
Acadia Publishing Company, Cambridge, Massachusetts (2017).
7. Lynn, G.: Animate Form. 1st edn. Princeton Architectural Press, New York City (1999).
8. Rossi, A., Tessmann, O.: From Voxels to Parts: Hierarchical Discrete Modeling for Design
and Assembly. In: Cocchiarella, L. (eds.) ICGG 2018 – PROCEEDINGS OF THE 18th
INTERNATIONAL CONFERENCE ON GEOMETRY AND GRAPHICS, AISC vol. 809,
pp. 1001-1012. Springer, Milan (2018)
9. Johns, R.L.: Augmented materiality: Modelling with material indeterminacy. In: Gramazio,
F., Kohler, M., Langenberg, S. (eds.) FABRICATE 2014: NEGOTIATING DESIGN AND
MAKING, pp. 216-223. Gta Verlag, Zurich (2014).
10. These materials are not sponsored by or affiliated with Unity Technologies or its affiliates.
“Unity” is a trademark or registered trademark of Unity Technologies or its affiliates in the
U.S. and elsewhere.
11. LEGO® is a trademark of the LEGO Group of companies which does not sponsor, authorize,
or endorse this material.
12. LEGO Group: LEGO Digital Designer (2012), http://ldd.lego.com/, last accessed
13. Sanches, J.: Architecture for the commons. Architectural Design 89(2), 22-29 (2019).
14. Gershenfeld, N.: Programming a new reality. In: TEDxCERN (2015), min. 8:00.
https://www.youtube.com/watch?v=EA-wcFtUBE4, last accessed 2019/04/02.
15. Project DisCo is available at http://www.project-disco.com
16. Wasp is available at https://www.food4rhino.com/app/wasp