Scholarlink

<p>The growing demand for scalable digital heritage content requires generative methods capable of preserving architectural typology rather than mere visual plausibility. Vernacular architecture poses a particular challenge due to its reliance on tacit material logic rather than explicit proportional systems. This paper presents a parametric Shape Grammar for the Morava house (moravska kuća), a nineteenth-century Serbian vernacular typology defined by bondruk construction, deep porches (trem), and optional corbeled projections (doksat). Based on the analysis of 54 documented examples, the study formalizes the geometric and tectonic rules underlying the typology and encodes them as a rule-based generative system. The grammar is evaluated through typological completeness tests and an expert-based visual evaluation. Results demonstrate that the full range of historical variants can be reproduced and that generated models achieve high typological fidelity. The findings confirm that vernacular architectural identity is computable through deterministic grammatical rules, while material authenticity remains a stochastic phenomenon.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Procedural Generation of Morava-Style Vernacular Houses for Video Game and Architectural Visualization

  • 期刊论文

  • 发表日期：

    2026年4月7日

查看全文
  • Milos Nikolic,
  • Nastasija Kocic,
  • Isidora Mitrovic,
  • Sonja Krasic,
  • Petar Pejic

摘要

The growing demand for scalable digital heritage content requires generative methods capable of preserving architectural typology rather than mere visual plausibility. Vernacular architecture poses a particular challenge due to its reliance on tacit material logic rather than explicit proportional systems. This paper presents a parametric Shape Grammar for the Morava house (moravska kuća), a nineteenth-century Serbian vernacular typology defined by bondruk construction, deep porches (trem), and optional corbeled projections (doksat). Based on the analysis of 54 documented examples, the study formalizes the geometric and tectonic rules underlying the typology and encodes them as a rule-based generative system. The grammar is evaluated through typological completeness tests and an expert-based visual evaluation. Results demonstrate that the full range of historical variants can be reproduced and that generated models achieve high typological fidelity. The findings confirm that vernacular architectural identity is computable through deterministic grammatical rules, while material authenticity remains a stochastic phenomenon.