This exploratory study examined the color characteristics of AI-generated Korean and Chinese residential space images and the visual attention responses of Chinese international students to these images. Eight residential space images (four Korean, KSI 1–4; four Chinese, CSI 1–4) were generated using Midjourney based on keywords derived from Korean and Chinese residential trend analyses. Color characteristics were quantified using the Korea Standard Color Analysis (KSCA) program, and eye-tracking data from seven valid participants were analyzed for fixation and saccade frequencies. Both image groups showed a high distribution in the Red–Yellow color region, but differed in lightness composition. Korean images were characterized by high lightness and a restrained, clean tonal organization, whereas Chinese images showed broader lightness ranges and stronger contrast through medium- and low-lightness colors. The highest fixation frequency was observed for KSI 2, suggesting concentrated visual attention, while the highest saccade frequency was observed for CSI 3, suggesting active spatial exploration. Image-level Spearman correlation analysis further suggested a positive association between the proportion of low-lightness colors and saccade frequency (ρ = +0.61). As a preliminary single-group study, the findings suggest that lightness distribution and contrast configuration may be associated with different patterns of visual attention and spatial exploration. Generative AI shows potential as a preliminary tool for visualizing culture-related spatial attributes; however, the findings should be interpreted as exploratory evidence based on Chinese international students’ responses rather than as a generalized comparison between Korean and Chinese cultures. The results provide a basis for future cross-cultural research involving Korean participants, expert validation of AI-generated images, and diverse spatial typologies.
Choi et al. (Sun,) studied this question.