Doring Stalls
AI Filmmaking; Transductive Authorship; Practice-Based Research
AI Filmmaking; Transductive Authorship; Practice-Based Research
Doring Stalls is a short narrative film created through multimodal generative AI. Set around a fictional street stall at the edge of the city, where the human and nonhuman intersect, the film explores encounters shaped by memory, technology, and ethical responsibility.
Developed as a four-creator omnibus, the film was produced through transformations across text, image, video, and editing. Through the production process, we explored how generating, selecting, rejecting, and reworking visual material shaped our creative decisions and our roles as authors.
The film was developed through an iterative process connecting narrative reconstruction, visual direction, AI image generation, image-to-video transformation, and editing. Rather than following a fixed sequence, the process involved repeated adjustments between stages as visual and narrative decisions evolved.
The transformation from initial visual direction to generated imagery involved different forms of creative judgment.
In Scene 2, repeated generations produced an unintended nonhuman appearance. These results were rejected and regenerated to achieve the intended character expression. In Scene 4, the original direction of warm, nostalgic imagery was set aside in favor of an elderly man's cybernetic close-up, shifting the scene toward questions of memory and technological ethics.
In Scene 3, an initial direction centered on mechanical activity was reconsidered during video production. A figure-centered laboratory scene was set aside in favor of an empty corridor illuminated by contrasting blue and orange light.
This shift transformed the scene's emotional atmosphere, allowing absence, space, and color to construct narrative tension.
This research was supported by the ANCHOR through the Seoul ANCHOR Center, funded by the Ministry of Education (MOE) and the Seoul Metropolitan Government (2025-ANCHOR-01-024-06).