View-Dependent Surrogate Terminals for Procedural Geometry Generation

Information

Abstract

Procedural geometry generation plays an ever-increasing role in the movie- and video gaming industry. Shape grammars have established themselves as the preferred solution for procedural architecture generation. Research in past decades drastically improved the speed of geometry derivation through shape grammars, making it possible to generate 3D buildings on-demand and in real-time. However, the constantly rising demand for high-quality visualizations requires new measures to reduce complexity in 3D models generated by shape grammars without sacrificing visual quality. This thesis explores the feasibility and benefits of inserting view-dependent surrogate terminals into a shape grammar. Surrogate terminals end grammar derivation early and approximate finer details with pre-rendered images. We find a possible solution for implementing view-dependent surrogate terminals and describe a scheme to automatically insert them into a shape grammar. Results show that contrary to previous approaches, our method avoids the generation of visibly incomplete geometry. However, even though the modified shape grammars evaluate faster than the original in large scenes, previous methods provide a more significant performance gain. We conclude that view-dependent surrogate terminals provide promising results, but further optimization is necessary to match the performance of prior techniques.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{roth_2021_vdst,
  title =      "View-Dependent Surrogate Terminals for Procedural Geometry
               Generation",
  author =     "Moritz Roth",
  year =       "2021",
  abstract =   "Procedural geometry generation plays an ever-increasing role
               in the movie- and video gaming industry. Shape grammars have
               established themselves as the preferred solution for
               procedural architecture generation. Research in past decades
               drastically improved the speed of geometry derivation
               through shape grammars, making it possible to generate 3D
               buildings on-demand and in real-time. However, the
               constantly rising demand for high-quality visualizations
               requires new measures to reduce complexity in 3D models
               generated by shape grammars without sacrificing visual
               quality. This thesis explores the feasibility and benefits
               of inserting view-dependent surrogate terminals into a shape
               grammar. Surrogate terminals end grammar derivation early
               and approximate finer details with pre-rendered images. We
               find a possible solution for implementing view-dependent
               surrogate terminals and describe a scheme to automatically
               insert them into a shape grammar. Results show that contrary
               to previous approaches, our method avoids the generation of
               visibly incomplete geometry. However, even though the
               modified shape grammars evaluate faster than the original in
               large scenes, previous methods provide a more significant
               performance gain. We conclude that view-dependent surrogate
               terminals provide promising results, but further
               optimization is necessary to match the performance of prior
               techniques.",
  month =      jul,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Research Unit of Computer Graphics, Institute of Visual
               Computing and Human-Centered Technology, Faculty of
               Informatics, TU Wien ",
  keywords =   "procedural geometry, real-time, rendering",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2021/roth_2021_vdst/",
}