From visual consistency to creative deviation: A study of image completion with AI in Zaha Hadid architecture

Pınar Çalışır Adem

doi:10.46372/arts.1744268

TR EN

Görsel tutarlılıktan yaratıcı sapmaya: Zaha Hadid mimarlığında yapay zekâ ile görsel tamamlama çalışması

Öz

Yapay zekâ araçları, mimari tasarımda biçimsel spekülasyon ve hayal gücünü destekleyen araçlar olarak öne çıkmaktadır. Fakat, mevcut algoritmalar, mimari formdaki anlamsal ilişkileri tam olarak kavrayamadıkları için üretimlerinde biçimsel tutarsızlıklar ortaya çıkabilmektedir. Bu çalışma, görsel tamamlama (inpainting) araçlarının farklı prompt ve maskeleme stratejileri aracılığıyla eksik mimari imgeleri ne ölçüde aslına uygun biçimde tamamlayabildiğini ve hangi noktada yaratıcı sapmalar ürettiğini sorgulamaktadır. Örneklem, mimari literatürde sıkça yer alan ve biçimsel dili açısından bütüncül, güçlü bir kimlik sergileyen on Zaha Hadid Mimarlık yapısından oluşmaktadır. Cephe görselleri farklı oranlarda eksiltilmiş ve FLUX-1 adlı Stable Diffusion tabanlı görsel tamamlama aracıyla tamamlanmıştır. Sonuçlar, biçimsel süreklilik ve mimari dil açısından niteliksel olarak ve Yapısal Benzerlik İndeksi (SSIM) ile niceliksel olarak değerlendirilmiştir. Bulgular, eksiltme oranı arttıkça yapay zekâ aracının yorum gücünün belirginleştiğini göstermektedir.

Anahtar Kelimeler

From visual consistency to creative deviation: A study of image completion with AI in Zaha Hadid architecture

Abstract

AI tools have emerged as instruments that support formal speculation and imagination in architectural design. However, current algorithms may fail to fully comprehend the semantic relationships within architectural form which leads to inconsistencies in their outcomes. This study investigates to what extent inpainting tools, through different prompt and masking strategies, can complete masked architectural images in a manner consistent to the original and when they begin to produce creative deviations. The sample consists of ten buildings by Zaha Hadid Architects, widely recognized in architectural literature for their coherent and distinctive formal language. Facade images were partially masked at different ratios and completed using FLUX-1, an inpainting tool based on Stable Diffusion. The results were evaluated qualitatively in terms of formal continuity and architectural language and quantitatively using the Structural Similarity Index (SSIM). Findings reveal that as the masking ratio increases, the interpretive capacity of the AI tool becomes more pronounced.

Keywords

Ethical Statement

kapsam dışıdır

References

Arielli, E. (2024a). Even an AI could do that. In E. Arielli & L. Manovich (Eds.), Artificial aesthetics: Generative AI, art and visual media (pp. 8–24). Retrieved from https://manovich.net/index.php/projects/artificial-aesthetics
Arielli, E. (2024b). Human perception and the artificial gaze. In E. Arielli & L. Manovich (Eds.), Artificial aesthetics: Generative AI, art and visual media (pp. 95–117). Retrieved from https://manovich.net/index.php/projects/artificial-aesthetics
As, I., Pal, S., & Basu, P. (2018). Artificial intelligence in architecture: Generating conceptual design via deep learning. International journal of architectural computing, 16(4), 306–327. https://doi.org/10.1177/1478077118800982
Boden, M. A. (2010). Creativity and art: Three roads to surprise. Oxford University Press.
Bohm, D. (1996). On creativity. Routledge.
Carpo, M. (2017). The second digital turn: Design beyond intelligence. MIT Press.
Chen, J., Shao, Z., & Hu, B. (2023). Generating interior design from text: A new diffusion model-based method for efficient creative design. Buildings, 13, 1861. https://doi.org/10.3390/buildings13071861
Dilaveroglu, B. (2024). The architecture of visual narrative: Can text-to-image algorithms enhance the power of stylistic narrative for architecture. International journal of architectural computing, 22(3), 432–457. https://doi.org/10.1177/14780771241234449

Enjellina, E., Beyan, E. V. P., & Rossy, A. G. C. (2023). Review of AI image generator: Influences, challenges, and future prospects for architectural field. Journal of artificial intelligence in architecture, 2, 53–65. https://doi.org/10.24002/jarina.v2i1.6662
Foote, K. D. (2022). The history of machine learning and its convergent trajectory towards AI. In S. Carta (Ed.), Machine learning and the city: Applications in architecture and urban design, (pp. 129–142). Wiley. https://doi.org/10.1002/9781119815075.ch9
Fogel, D. B. (2022). Defining artificial intelligence. In S. Carta (Ed.), Machine learning and the city: Applications in architecture and urban design, (pp. 91–120). Wiley. https://doi.org/10.1002/9781119815075.ch7
Hanafy, N. O. (2023). Retracted: Artificial intelligence’s effects on design process creativity: A study on used A.I. text-to-image in architecture. Journal of building engineering, 80. https://doi.org/10.1016/j.jobe.2023.107999
Hugging Face. (n.d.). Using ControlNet with diffusers [Documentation]. Hugging Face. Retrieved August 5, 2025, from https://huggingface.co/docs/diffusers/using-diffusers/controlnet
Kim, J. Y., & Park, S. J. (2025). AI-driven biophilic façade design for senior multi-family housing using LoRA and stable diffusion. Buildings, 15(9), 1546. https://doi.org/10.3390/buildings15091546
Koestler, A. (1964). The act of creation. Hutchinson.
Kuang, Z., Zhang, J., Li, Y., et al. (2025). Preserving architectural heritage in urban renewal: A stable diffusion model framework for automated historical facade generation. npj heritage science, 13, 256. https://doi.org/10.1038/s40494-025-01826-4
Leach, N. (2022). In the mirror of AI: What is creativity? ARIN, 1, 15. https://doi.org/10.1007/s44223-022-00012-x
Li, P., Li, B., & Li, Z. (2024). Sketch-to-architecture: Generative AI-aided architectural design. arXiv. Retrieved from https://arxiv.org/abs/2403.20186
Lugmayr, A., Danelljan, M., Romero, A., Yu, F., Timofte, R., & Van Gool, L. (2022). RePaint: Inpainting using denoising diffusion probabilistic models. arXiv. Retrieved from https://arxiv.org/abs/2201.09865
Manovich, L. (2019). Defining AI arts: Three proposals. In AI and dialog of cultures, Exhibition Catalog, Hermitage Museum, Saint-Petersburg, Russia. Retrieved from https://manovich.net/index.php/projects/defining-ai-arts-three-proposals
Ma, H., & Zheng, H. (2024). Text semantics to image generation: A method of building facades design based on stable diffusion model. In C. Yan, H. Chai, T. Sun, & P. F. Yuan (Eds.), Phygital intelligence. CDRF 2023. Computational design and robotic fabrication (pp. 24–35). Springer. https://doi.org/10.1007/978-981-99-8405-3_3
Nilsson, J., & Akenine-Möller, T. (2020). Understanding SSIM. arXiv. Retrieved from https://arxiv.org/abs/2006.13846
Picon, A. (2025). Artificial intelligence and architectural intention. Technology|architecture + design, 9(1), 6–9. https://doi.org/10.1080/24751448.2025.2465063
Ploennings, J., & Berger, M. (2023). AI art in architecture. AI in civil engineering, 2(8). https://doi.org/10.1007/s43503-023-00018-y
Rashid, M. (2025). Architect, AI and the maximiser scenario. AI & society, 40, 241–243. https://doi.org/10.1007/s00146-023-01848-1
Shi, Y., & Wang, C. (2025). Optimizing Ionic style facade creation by integrating shape grammars into stable diffusion. In H. Chai, D. W. N. Bao, Z. Guo, & P. F. Yuan (Eds.), Symbiotic intelligence. CDRF 2024. Computational design and robotic fabrication (pp. 297–306). Springer. https://doi.org/10.1007/978-981-96-3433-0_26
Sukkar, A. W., Fareed, M. W., Yahia, M. W., Mushtaha, E., & De Giosa, S. L. (2024a). Artificial Intelligence Islamic Architecture (AIIA): What is Islamic architecture in the age of artificial intelligence? Buildings, 14, 781. https://doi.org/10.3390/buildings14030781
Sukkar, A. W., Fareed, M. W., Yahia, M. W., Abdalla, S. B., Ibrahim, I., & Senjab, K. A. K. (2024b). Analytical evaluation of Midjourney architectural virtual lab: Defining major current limits in AI-generated representations of Islamic architectural heritage. Buildings, 14, 786. https://doi.org/10.3390/buildings14030786
Wang, Z., & Bovik, A. C. (2002). A universal image quality index. IEEE Signal Processing Letters, 9, 81–84. https://doi.org/10.1109/97.995823
Vela, L., Fuentes-Hurtado, F., & Colomer, A. (2023). Improving the quality of image generation in art with top-k training and cyclic generative methods. Scientific Reports, 13, 17764. https://doi.org/10.1038/s41598-023-44289-y
Zhang, J., Huang, Y., Li, Z., Li, Y., Yu, Z., & Li, M. (2024). Development of a method for commercial style transfer of historical architectural facades based on stable diffusion models. Journal of Imaging, 10, 165. https://doi.org/10.3390/jimaging10070165
Zhang, Z., Fort, J. M., & Giménez Mateu, L. (2023). Exploring the potential of artificial intelligence as a tool for architectural design: A perception study using Gaudí’s works. Buildings, 13, 1863. https://doi.org/10.3390/buildings13071863
Zylinska, J. (2020). AI art: Machine visions and warped dreams. Open Humanities Press. alimama-creative. (2023). FLUX-Controlnet-Inpainting [Computer software]. GitHub. Retrieved October 5, 2025, from https://github.com/alimama-creative/FLUX-Controlnet-Inpainting

Details

Primary Language

English

Subjects

Arts and Cultural Policy

Journal Section

Research Article

Authors

Pınar Çalışır Adem ^*
0000-0001-6139-6289
Türkiye

Publication Date

October 22, 2025

Submission Date

July 16, 2025

Acceptance Date

October 7, 2025

Published in Issue

Year 2025 Number: Yapay zekâ ve sanat özel sayısı

DOI

https://doi.org/10.46372/arts.1744268

IZ

https://izlik.org/JA32BP36LG

Cite

RIS / Bibtex

APA

Çalışır Adem, P. (2025). From visual consistency to creative deviation: A study of image completion with AI in Zaha Hadid architecture. ARTS: Artuklu Sanat Ve Beşeri Bilimler Dergisi, Yapay zekâ ve sanat özel sayısı, 29-59. https://doi.org/10.46372/arts.1744268