AI-driven transformation of traditional architectural Ottoman style into modern mosque designs

The use of AI art-generating platforms has recently become a trend in different disciplines, including the field of architecture. The relation AI has with architecture led to the idea of this paper to explore the intersection between these platforms with traditional Islamic architecture, specifically the Ottoman style. To achieve a proper evaluation, this paper discusses the use of 6 different platforms to generate 18 images using the text-to-image method and by employing 3 distinct prompts from different complexities to guide the generation process. Moreover, a comparative analysis conducted to evaluate the generated images using 2 criteria, the first assesses the Ottoman architectural elements that appeared in the images such as the domes, minarets, and arches. Then the second investigates the overall excellence of the figures considering factors such as realism, quality and resolution, overall details, and lastly the consistency with the given prompt. The results demonstrate that across all the 6 platforms, the more detailed the prompt is, the better the outcome. However, Midjourney was the only platform that was able to produce all three images consistently from the first prompt with all the required elements, and as the prompts got more detailed, the other platforms provided more accurate results.