An early experiment in video-to-video stylization — feeding source footage through a generative pipeline so the underlying motion is preserved while the visual language is rewritten frame by frame.

Built around Stable Diffusion's emerging video tooling at the time, this was part of the practical learning that calibrated my sense of where AI sits in a real production pipeline: where it adds value, where it costs control, and how to keep a director's hand on a tool that wants to drift.

Brief

Personal · video-to-AI stylization study

Role

Direction · GenAI workflow · Edit

Stack

Stable Diffusion · ComfyUI · post tooling

Output

Process clip · Instagram post