EVALUATING THE EFFECTIVENESS AND PROMPT ALIGNMENT OF TEXT-TO-IMAGE AI MODELS IN OFFICE LIGHTING DESIGN: A COMPARATIVE ANALYSIS

Authors : Yaren Şekerci, Parla Özkul

Pages : 233-256

View : 52 | Download : 314

Publication Date : 2026-01-01

Article Type : Research Paper

Abstract :This study demonstrates the capabilities of artificial intelligence technology in interior design, particularly in office lighting design. Lighting is an essential environmental factor and instantly affects how individuals evaluate a room. Preference for lighting level and CCT varies. A little research has been found to investigate the efficiency of AI technologies in lighting. The study involved 28 scenes and lighting designs for an office-built environment generated by 7 distinct AI text-to-image models. 2 light CCTs (warm and cool) and 2 illuminance levels (high and low) were used to produce four prompts. This study compares and assesses various AI systems\\\' accuracy and efficacy in applying predefined lighting prompts. This comparative analysis provides valuable insight into the potential applications of current AI technologies in interior design, particularly in office lighting design. The research results indicate that the most efficient office lighting design models are DALL-E and Midjourney, whereas Craiyon regularly performs lower.
Keywords : Yapay Zeka, Metinden Görsele, Ofis Aydınlatma Tasarımı, Aydınlatma Düzeyi, Bağıl Renk Sıcaklığı (CCT)

ORIGINAL ARTICLE URL