Project Description
Cinematic Spaces in the AI Lens: Mapping Arousal & Perception is a design analytics research project that examines how spatial elements in cinematic environments influence human emotion and perception. Starting from selected movie scenes across different genres and atmospheres, the project uses AI image and video generation tools to systematically modify key spatial variables, including setting, lighting, and material texture, creating a matrix of controlled cinematic video clips for user testing.
Through a mixed-method study combining self-reported emotional ratings, eye-tracking data, and physiological EDA signals, the project maps how viewers respond to different spatial transformations. The findings suggest that spatial variables do affect emotional perception, but their impact is strongly shaped by the original cinematic context. Among the tested variables, lighting emerged as the most disruptive factor, often reducing perceived valence, atmosphere, and safety, while texture modifications tended to enhance more positive emotional responses. Eye-tracking results further indicate that visual anchoring plays a key mediating role: spaces with clear semantic or structural anchors help stabilize gaze and emotional response, while minimal or anchor-less AI-modified spaces encourage scanning, uncertainty, and perceptual instability.
Rather than treating AI as only a generative tool, this project uses AI as an experimental lens for isolating and testing spatial qualities that are usually embedded within filmic composition. By connecting cinematic space, affective response, physiological sensing, and visual attention, the project explores how design analytics can reveal the emotional agency of spatial environments.