Manuscript received September 12, 2025; revised October 16, 2025; accepted November 19, 2025; published May 29, 2026.

Abstract—Visual storytelling in short films relies heavily on directing audience attention, yet traditional methods for analyzing visual saliency often lack cinematic context and temporal dynamics. We propose a dual-map framework merging Artificial Intelligence (AI)-generated attention heatmaps with focus maps to deliver data-driven insights for visual storytelling. A focus map is defined as a temporal representation of sustained viewer attention that captures the dynamic evolution of gaze across frames. The attention heatmap model, trained on cinematic datasets, identifies salient regions while the focus map model, which includes temporal attention gates, captures dynamic shifts in viewer focus. The framework employs spatial clustering to localize coherent areas of interest and quantifies the consistency between predicted attention and actual focus patterns. Furthermore, the system produces annotations that are compatible with standard editing software, which permits filmmakers to improve their work based on empirical evidence. The proposed method fills a crucial void in computational film analysis by merging static saliency with temporal attention modeling, thus creating a tool thatconnects artistic intent and audience perception. Experimental findings show that the framework successfully identifies gaps between intended and actual viewing behaviors, yielding practical insights for optimizing visual storytelling. This study advances both film studies and human-computer interaction by introducing a scalable, interpretable method for examining and improving cinematic narratives. Merging attention and focus maps creates novel opportunities for data-driven filmmaking, enabling artistic choices to be guided by empirical visual assessment. This research contributes a dual-map framework that bridges computational film analysis and practical filmmaking, providing a reproducible, interpretable tool for directing visual storytelling through data-driven attention modeling.

Keywords—short films, artificial intelligence, attention heatmaps, visual storytelling, focus maps

Cite: Abdunroni Samaeng and Athitaya Somlok, "Enhancing Short-Film Visual Storytelling with AI-Generated Attention and Focus Maps: A Dual-Map Framework for Data-Driven Analysis," Journal of Image and Graphics, Vol. 14, No. 3, pp. 425-433, 2026.

Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).