<![CDATA[T2: Mixing Procedural Techniques and Shape Grammars]]>Shape Grammars are an architectural tool which can be used to analyze or design architectural scenes based on a set of rules. While it is mainly seen as a theoretical tool, several case studies as shown it benefits to support architectural workflow such as it can been see to analyze the application of architectural treatise (FCT Digital Alberti project) or mass customization of housing (publication Duarte05) . However this representation requires computational expertise to be use effectively by architects. On
the other hand, Computer graphics propose procedural modeling techniques inspired by shape grammar such as the split grammars to automate the generation of complex 3D models or virtual cities. While it provides scripting solutions, existing procedural modeling lost several of the key advantages of Stiny shape grammars removing architects from the generation loop.

This task is subdivided into four subtasks and looks forward in order to overcome such limitations and make procedural techniques more adapted to interactive modeling and architects needs.

Subtask 2.1: Enriching Split Grammars for Curves and Surfaces
Current Split Grammars can only reproduce a subset of shapes and have been particularly successful to automate the generation of building facades. However their formalism should be enriched to support more complex shapes such as curves and surfaces. Conics are an essential representation and needs to be accessible to any geometric representation. While split grammars are adapted for subdivision and elevations, they should be enriched for more complex representation such as surface revolutions. This task will add support to existing split grammars to represent a more broad range of shape.

Subtask 2.2: Shape Grammars: From a visual paradigm to a scripting modeling language
This task will focus on the mapping from visual shape grammars to our extended split grammar formalism. The complete definition of a new scripting language more adapted to rule definition than coding will be defined bringing the gap between architectural formalism and computer graphics procedural modeling techniques. Completeness and robustness verification of the language should be done following programming language theory. Thanks to this new representation architects will be able
to create shape grammars without needing strong programming skills and focus on the design and logic of shape grammar rules.

Subtask 2.3: Procedural Modeling Engine for Shape Grammars
This subtask will implement a procedural modeling engine using the state of the art technique in this field. The resulting engine will enable to generate procedural model based on the shape grammar and adapted to be visualized in our visualization Framework. The engine will enable to generate several level of detail model and see the progress of the engine in an interactive way on demand. The result of this engine will be compared to existing procedural modeling software.

Subtask 2.4: 2D Sketch Based Shape Grammar Creation and Edition
Finally a demonstrator of our procedural engine will be produced reducing the need of scripting skills. We will devise a sketch based editing interface to allow creating and editing grammar based models using a visual representation closer to the initial concept of shape grammars.

]]>