Answers are grounded in this paper and cite their sources.
Generative artificial intelligence (GAI) is rapidly transforming cinematic pre-production by enabling the efficient creation of high-quality concept art. Despite major advances in image generation models, limited research has examined how users cognitively and emotionally evaluate AI-generated cinematic imagery. This study investigated user acceptance of concept art generated by Midjourney v5.2 within East Asian science-fiction and fantasy film contexts through a multi-dimensional perceptual framework. A total of 384 participants, including both design professionals and non-design participants, evaluated AI-generated scenes across five cinematic categories. The proposed framework examined five perceptual dimensions: visual reasonability, narrative potential, aesthetic appeal, immersion, and overall acceptance. Reliability analysis and principal component analysis provided preliminary evidence for the internal consistency and coherent component structure of the measurement instrument. Multiple regression analysis revealed that all four perceptual constructs significantly predicted overall acceptance (R 2 = 0.275, p < 0.001). Among them, visual reasonability emerged as the strongest predictor (β = 0.217), followed by aesthetic appeal (β = 0.187), narrative potential (β = 0.175), and immersion (β = 0.151). Formal moderation analysis further identified expertise-related differences in the predictive roles of selected perceptual dimensions. Visual Reasonability and Immersion were more strongly associated with Overall Acceptance among non-design participants, whereas the effects of Narrative Potential and Aesthetic Appeal did not differ significantly between the two groups. Within the present Midjourney v5.2 and East Asian study context, these findings provide empirical evidence for the perceptual factors associated with user acceptance of AI-generated cinematic imagery and contribute a user-centered basis for further evaluation of human–AI collaborative workflows.
No takes yet. Share an insight, caveat, or question.
Luo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: