Work

Knowledge-constrained image generation and early-stage design screening based on Yaozhou Kiln ceramic cultural genes

Year: 2026

Type: article

Abstract: The rapid development of artificial intelligence has brought new opportunities for the ceramic industry. As an intangible cultural heritage in the field of ceramics, Yaozhou kiln ceramics, although having certain development potential, lack innovative vitality to a certain extent. To address this problem, this study proposes a digital design process for traditional patterns of the Yaozhou Kiln ceramics using artificial intelligence. The design process is based on Grounded theory and extracts and summarizes the ceramic patterns of Yaozhou kiln, forming a multidimensional aesthetic feature library. Guided by the feature library, the core elements of Yaozhou kiln ceramic patterns were hierarchically processed and quantified with specific proportional weights through the KANO model questionnaire and AHP expert evaluation. To optimize the output results, in the generation stage, the diffusion model is style-oriented within the parameter-efficient fine-tuning (LoRA) framework, and the color gamut, lines, and composition syntax are controlled using structured prompts and negative constraints. Cross-model comparisons are also conducted to test the transferability of styles. The innovative approach lies in the digital generation method, which replaces the physical sampling of high consumables with high concurrency iterations. This not only reduces the cost of trial and error but also provides a basis for the implementation of sustainable strategies such as low-temperature glaze application, structural simplification, and recyclable packaging. The focus of this study is to propose a verifiable transfer path from cultural knowledge to model distribution, providing a replicable methodology and evaluation framework for the digital activation of ceramic intangible cultural heritage patterns.

Source: Scientific Reports

DOI: 10.1038/s41598-026-66245-2

Authors: Feng Xin, Zhekai Cai, Xiao’an Xu, Jingwen Zhang, Ran Hou, Rui Yan

Institutions: Shaanxi University of Science and Technology, Guangxi Science & Technology Normal University

Language: English

Topic: Generative Adversarial Networks and Image Synthesis

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,690