The Aim of the Task

The goal of this task is to implement a software tool that converts a detailed LoD2.2 building model to a simplified LoD1.2 representation. This conversion reduces the complexity of the building geometry by removing detailed roof structures and flattening the roof surface, resulting in a block-like model useful for various urban analysis applications (e.g., energy estimation, noise simulation, and visualization).

To simplify the problem:

• Input and output are in the Wavefront .OBJ format.
• Each input file contains a single triangulated building.
• Semantics (e.g., building parts or functions) are not considered—only geometry is processed.

Definition of LoD1.2 and LoD2.2

According to Biljecki (2014):

• LoD1.2: A block-shaped building model with a single flat roof, including smaller building extensions (e.g., alcoves). The building is extruded to a single height.
• LoD2.2: A detailed building model with general roof shapes (as in LoD2.0 and LoD2.1), including roof superstructures larger than 2 meters in size or 2 m² in area (e.g., dormers, chimneys).

LoD1.2 models strike a balance between information richness and computational efficiency, making them suitable for simulations of shadows, energy demand, noise, flooding, and urban wind comfort.

Methodologies

Step0: Preprocessing

Before performing the LoD2.2 to LoD1.2 conversion, a minimal but essential preprocessing step was applied to ensure the robustness of the surface classification and height estimation steps.

The core challenges in this task revolve around:

• Identifying ground surfaces using only geometric properties, and
• Determining an appropriate roof height that represents the building's overall geometry without being biased by outliers.

However, the presence of invalid or unrealistic geometry can significantly impact both steps. For example, in the bk.obj model, there was a set of spike-like faces protruding downward from the building base—an artifact that does not correspond to any real-world structure.