Procedural modeling for cultural heritage

Abstract

Accurate 3D reconstruction and realistic visualization of cultural heritage allow experts to fine-tune their theories on the lost links in the history of civilization. Although the 3D reconstruction is a significant challenge, precisely because of the state of degradation over the years, it constitutes a crucial task for experts to study and interact with long disappeared settlements and structures. Furthermore, the public, in general, will be provided with the conditions to explore them in virtual environments, thus fostering cultural, social, and scientific participation. Highly accurate reconstruction is, nevertheless, a very complex task, where all stages of image synthesis must be carefully executed from highly detailed 3D models to obtain a faithful depiction of the object of interest. Meanwhile, the textual descriptions and geospatial data collected by archaeologists on-site may be used to overcome the absence of visual information. Still, this data will not suffice, in which case procedural modeling turns out to be essential to avoid a great deal of time and labor-consuming modeling processes. Procedural modeling tools automatically generate three-dimensional models through computational processes that extend the base information according to a specific algorithm. In order to avoid reprograming the procedural modeling systems, we use mathematical methods that operate on parametrical symbolic descriptions that, flexibly, can model different types of objects. The most used mathematical methods are fractal geometry and formal grammars, particularly L-systems and shape grammars. In this chapter, we will approach the current advances in the area of procedural modeling and how these tools can be used to generate 3D models of cultural heritage. We also explore the relevant dimension of time, extending the modeling tasks to 4D. These applications do not focus on very specific landmarks, like cathedrals or palaces, which require manual effort or image-based techniques to capture the model with a high level of visual fidelity. Instead, we focus on modeling cities and their evolutions or the surroundings of these landmarks, that allow for an increased automation of the modeling process.