/

/

FRAME_02

FRAME_02 continues the “FRAME_XX” series in which I explore visuals using AI. For this experiment, it does not differ from the work FRAME_01.

FRAME_02 continues the “FRAME_XX” series, which serves as my personal laboratory for visual exploration.

I don’t come from a traditional film background, and I’ve never actually handled a professional cinema camera. So, the reality is, AI has become my way of reverse-engineering the craft. My process is a form of Visual Deconstruction:

  1. Analysis: I select a still from a film that moves me and ask AI to break it down—analyzing the lighting, focal length, and film stock.
  2. Calibration: I use that text description to generate a new image. If the AI output matches the “vibe” of the original, I know those specific keywords trigger the right visual response. It’s how I build a reliable vocabulary that I can mix and match later to create something new.
  3. Direction: While the texture is a study of the masters, the motion is personal. I direct the camera movement, motion and pacing based on my own instinct, not the source material.

Realistically, it might be years before I’m shooting on an ARRI or RED—or maybe that day never comes 😅. But by working backwards like this, I can hopefully understand the mechanics of cinematography.

Prompt List (Gist Github)

This sequence explores the visual language of the following films:

  • Wrong Move (1975) — Wim Wenders
  • War of the Worlds (2005) — Steven Spielberg
  • Sidewalls (2011) — Gustavo Taretto
  • The Invitation (2015) — Karyn Kusama

Release Date

January 25, 2026

Runtime

00:02:03

Directed by

Taruma Sakti

Technique

txt2img, img2vid

AI Platform

OpenArt, Google Flow

Text/LLM Model

Gemini 3 Flash

Image Model

Nano Banana Pro

Music

The Gordian Knot by Jakob Ahlbom

posted by

Recent Works