In recent years, breakthrough developments in AI-based image generation, panoramic imaging, and photogrammetry have rekindled interest in foundational research on the cognition of dynamic narrative images. Spanning the fields of physiology, psychology, narratology, and computer science, this line of research collectively influences how humans process and understand audiovisual media. Among these dimensions, spatiotemporal cognition is key to establishing visual continuity, substantially affecting narrative comprehension. Understanding dynamic narrative images requires constructing a situation model, which is composed of an event structure integrating multiple cognitive dimensions. Due to its complexity, past research has struggled to isolate different dimensional variables for measurement and to identify sample groups with broad applicability. This study proposes a new method for effectively controlling and separating variables by focusing on the visual-spatial dimension, examining the spatial presentation of crossing-the-line editing and the degree of visual content load as the primary analytical dimensions. This research study aims to investigate how shot combinations in the editing process—specifically, the use of dynamic or static shots in crossing-the-line editing—and the level of visual element load (the simplicity or complexity of foreground and background) affect viewers’ spatial cognition. An experiment was conducted by recruiting 84 participants through convenience sampling. The results indicate that: (1) The combination of dynamic and static crossing-the-line shots is the main factor producing significant differences, notably influencing the accurate depiction of characters’ relative positions, the continuity of space, positional consistency, and the smoothness of editing with regard to spatial cognition. (2) When accurately describing characters’ relative positions, the shared resource factors of visual recognition and tracking attention underscore the impact of the complexity of foreground and background content. These findings offer clear theoretical support for the practical production of dynamic narrative images. Applied to various narrative contexts, they enable creators to convey messages precisely while enhancing viewers’ immersion in the viewing process.