Rapid development in manufacturing technologies has given rise to a new architectural form: media architecture, which demonstrates significant potential in shaping urban spatial experiences. In people’s interaction with media content, visual attention plays a vital role. Existing studies have explored visual attention in diverse urban contexts, but few have researched the dynamics within media architecture. Therefore, this paper aims to bridge the gap by modelling visual attention and analysing the impact of digital content and spatial layout by both qualitative and quantitative methods. This study conducted 17 field experiments to collect video data by head-mounted cameras and then analysed the participants' visual attention. Various attributes of digital elements were quantitatively analysed. The results have indicated that the relative animation speed and levels of interactivity impact most visual attention in media architecture. The findings revealed the complex dynamics of how viewers engaged with digital content in media architecture, providing evidence and insight for creating urban media environments.