生成·記載·延續—在AI與量產設計的時代，中國建築紋飾如何重新展開科技與傳統的對話

在大量設計與數位標準化盛行的時代，建築正逐漸與場域、記憶與文化脫節。曾是建築表現核心的「裝飾」，在現代主義中多被視為多餘或過時的元素。本計畫提出另一種觀點：裝飾不僅重要，更是不可或缺的。它可以是記憶的機器、構築的語言，以及文化抵抗的載體，並透過人工智慧獲得新的賦能。

本設計靈感來自中國南方嶺南傳統建築，元素包括斗拱、門窗、磚瓦、脊飾與雕刻等。我運用機器學習中的擴散模型，生成模組化且可適應的裝飾系統。這些元素不只是表面的裝飾，而是構成結構性的建築語彙，被嵌入於瓦片、柱體、托座與立面之中，傳遞象徵意涵、材料邏輯與文化記憶。
–
At a time of mass design and digital standardization, architecture is becoming increasingly divorced from site, memory, and culture. Architectural ornament, which was central to architectural expression, was largely considered superfluous or anachronistic in modernism. This project proposes an alternative perspective that ornament is not merely significant, but essential. Ornament can be a machine of memory, a tectonic language, and a carrier of cultural resistance, empowered by artificial intelligence.

Inspired by traditional Lingnan architecture in southern China, which features elements such as Dougong, doors and windows, brick and tile, ridge tiles, and carvings, I utilize machine learning diffusion models to generate modular and adaptive ornamentation systems. These elements are not mere decoration; rather, they form an architectural language of structure, incorporated into tiles, columns, corbels, and building facades, conveying symbolic value, material rationality, and cultural memory.