Personal Memories, Political Echoes
Core Statement: Blending the Personal and the Political
In the quiet moments of family gatherings, captured in old VHS tapes from the early 1990s, one might not expect to encounter political undertones. Yet, embedded within the everyday footage of birthdays, reunions, and celebrations, lie subtle reflections of the political and cultural shifts that shaped Istanbul during that period. The 1990s were marked by significant social upheavals, economic challenges, and a wave of political transformations that reverberated through the streets of Istanbul. This project seeks to reframe these ordinary memories by introducing a technological element: artificial intelligence.
Inspiration: AI, TensorFlow, and Data Manipulation
TensorFlow is an open-source machine learning library developed by Google, widely used for tasks like image recognition, natural language processing, and more. For this project, TensorFlow’s image recognition capabilities allow the AI to detect and analyse elements within my family’s home videos. Additionally, by introducing history of the objects, professions and status of family members, newspaper clippings, and images from the era’s media—the narration gains the AI aesthetic with a sense of the political atmosphere that informs the footage, adding depth to the seemingly simple moments.
The Political Context of 1990s Istanbul:
Unveiling Hidden Narratives
In the 1990s, Istanbul was a city in transition. The decade was marked by political instability, economic crises, and the tumultuous rise of new media. Television was a key player in shaping public opinion, and newspapers played a critical role in disseminating political discourse. However, the personal moments captured in my family’s home videos offer a stark contrast to the national headlines. By isolating key elements from the footage—television screens, newspaper headlines, and even products in the background—the narration reveals the often-overlooked political undercurrents that shaped everyday life.
The project challenges the traditional notion of private life as apolitical and invites viewers to reconsider the political implications of their own memories.
Some of the topics heard, or seen in the videos include:
https://en.wikipedia.org/wiki/1994_Italian_general_election
https://en.wikipedia.org/wiki/Uğur_Mumcu
https://en.wikipedia.org/wiki/Erdal_İnönü
https://en.wikipedia.org/wiki/Tansu_Çiller
Rough cut with subtitles 8:27 min. https://youtu.be/vMFlPFam_7s
Personal narration via computer vision – graphical style 4:00 min. https://youtu.be/D6_Z1eQJ1gM
Project Status
From over 8 hours of material, I have currently trained 55 minutes of footage using the COCO Dataset to help the AI recognize and generalize the objects and people on screen. Next, I will begin training the system with more personalized data, incorporating additional context such as the history of objects, the expressions of the actors, and, finally, the emotional and political context of the scenes. This process will help refine the overall narrative and statement of the project.
First rough training material (55 min, without sound): Watch here
Option 1
My goal is to condense the footage to a 8 -16 minute short film format. AI will lead the narration with overlayed information and text boxes. This could be presented as a multi-channel video installation, featuring two screens: one with the unformatted video and subtitles, and the other with an augmented computer vision version. Additionally, I will be rendering a more compact version of the work for submission to film festivals.
To optimize computational power and time, I am training the AI on low-resolution footage, and will then upscale all the scenes to 4K for the final render.
Option 2
In this option, I will engage AI as an information tool, using it to illuminate subtle nuances and details that may have initially eluded me during my first interactions with the material. The format of the work will evolve into an essay film, approximately 25 minutes in length, incorporating additional video footage, voice-over commentary, and sound design to articulate both my personal and political narratives. Drawing from AI Vision as a distinct graphic language, I will freeze-frame key moments, integrating “forensic analysis” through on-screen text or visual overlays that will interact with the original footage.
Both options aim to deepen the critical examination of the work, offering an enriched, multi-layered perspective that merges personal reflection with a broader cultural and political context.