A Sunset With A Sky Background

J. Makary

2025

Medium: 16mm film, transferred to digital video, AI

Dimensions: variable

Price: Upon Request

Description: As a scholar and filmmaker, I value experimentation and unorthodox use of tools, materials, and mediums. My films have explored unconventional narrative structures and deployed experimental techniques inspired by my dance and fine art training. After beginning my doctoral studies, however, I put my art practice to the side. A course with Shane Denson at the end of my second year gave me the opportunity to bring my creative work into dialogue with my emerging scholarly research. To reenergize my return to artmaking and balance the demands of doctoral work, I gave myself severe constraints: I would shoot nothing new, and no project should take more than six hours to create. A Sunset With a Sky Background is the result of one of these course experiments. Its core footage comes from eight reels I shot with a 16mm Bolex, capturing a year of environmental change in Malibu following the 2018 Woolsey Fire. The wildfire burned more than 100,000 acres, close enough to where I lived in Topanga Canyon that I was evacuated for eight days. On my return, I began documenting the fire's effects through frequent explorations around Malibu's canyons and shoreline, and this footage remained unedited at the time of my move to the Bay Area. At Stanford, I became interested in using methodologies foundational to cinema studies, such as psychoanalysis and semiotics, to interpret emergent visual forms of AI. For Sunset, I wanted to make an AI film without the AI aesthetic. The 16mm Malibu footage, with its analog color and ample grain, provided the perfect ground for this experiment. I selected portions of my footage with pronounced visual indeterminacy—marked by heavy fog, low light, discrepant grain and color, and close-ups lacking context—and isolated still frames from these sections. I uploaded these frames to several visual language models, which struggled to identify what is depicted in the images. The inaccurate captions generated by the AI were then remarried to the moving footage, creating a slippage effect where what the viewer sees and cognitively identifies diverges from what the AI suggests should be seen. The grain and texture of 16mm, and my shooting style, resist digital interpretation and algorithmic reading. However, the visual ambiguities of Sunset—and the subtle sound interventions that give the AI an “assist”—confound straightforward interpretations about the limitations of machine vision.