Journals Information
Civil Engineering and Architecture Vol. 13(3), pp. 1527 - 1537
DOI: 10.13189/cea.2025.130308
Reprint (PDF) (2451Kb)
Investigating Architectural Style: Preservation and Innovation Utilizing AI - Rifat Chadirji's Dataset from Iraq as a Case Study
Raghad N. Jasim 1,*, Anwar S. Al-Qaraghuli 1, Zuhair A. Nasar 2
1 Department of Architectural Engineering, University of Technology, Iraq
2 Department of Urban Planning, Faculty of Physical Planning, University of Kufa, Iraq
ABSTRACT
Architecture is constantly evolving with the emergence of diverse technologies, in particular, the subject of architecture and artificial intelligence (AI). As such, this research focuses on architectural style utilizing Generative Adversarial Networks (GANs) and AI tools, especially in image-to-image technology. Further, the study examines StyleGAN 2-ADA, which can generate images as a general summary of the data input, besides DALL-E 3 and Midjourney AI tools to create images. This study analyzes AI models using local architectural data from the works of Rifat Chadirji, a renowned Iraqi architect (1926鈥�2020) recognized for integrating modernist principles with traditional Iraqi architectural elements, resulting in a distinctive and influential style. The images of the facades were chosen for this research due to their extensive temporal range and the adaptability of their design approach. The research provides greater control over design outcomes using image-based inputs, addressing how architects can leverage AI tools to balance heritage preservation with innovation. Moreover, it offers practical insights into merging these tools into professional workflows and cultural applications. The methodology of this work has two main parts: generation and evaluation. In the generation process, the dataset was input into StyleGAN 2-ADA for image generation, whereas, in DALL-E 3 and Midjourney, the dataset was classified into subgroups before generating images. The evaluation process includes two methods. Firstly, the SSIM metric is applied to determine the structural similarity between the original and generated images. Further, the questionnaire was selected to investigate architects' opinions and assess attributes such as historical aspects, artistic elements, patterns, and materials, ensuring alignment with architectural standards. The results demonstrated that StyleGAN 2-ADA excels in tasks centered on retaining architectural heritage. On the other hand, DALL-E 3 proves to be a valuable tool for fostering innovation. Meanwhile, Midjourney provides a flexible method by balancing preservation and renewal.
KEYWORDS
Architectural Style, Artificial Intelligence, Rifat Chadirji, StyleGAN 2-ADA, DALL-E 3, Midjourney
Cite This Paper in IEEE or APA Citation Styles
(a). IEEE Format:
[1] Raghad N. Jasim , Anwar S. Al-Qaraghuli , Zuhair A. Nasar , "Investigating Architectural Style: Preservation and Innovation Utilizing AI - Rifat Chadirji's Dataset from Iraq as a Case Study," Civil Engineering and Architecture, Vol. 13, No. 3, pp. 1527 - 1537, 2025. DOI: 10.13189/cea.2025.130308.
(b). APA Format:
Raghad N. Jasim , Anwar S. Al-Qaraghuli , Zuhair A. Nasar (2025). Investigating Architectural Style: Preservation and Innovation Utilizing AI - Rifat Chadirji's Dataset from Iraq as a Case Study. Civil Engineering and Architecture, 13(3), 1527 - 1537. DOI: 10.13189/cea.2025.130308.