Research: Teaching Water to Machines

This ongoing research explores whether the physical behavior of Ebru can be studied and simulated through artificial intelligence.

Ebru patterns emerge from a complex interaction of pigment, viscosity, surface tension, and motion. Unlike many visual systems, the artist does not fully control the outcome; the image forms through a physical process that unfolds on the water’s surface.

This project investigates whether machine learning models can observe and learn these dynamics. Using video documentation of live marbling sessions, the research asks whether AI can begin to recognize, predict, or simulate the formation of patterns produced by fluid interactions.

The goal is not simply to generate digital marbling, but to study how a physical artistic process might be translated into computational understanding.

This work is currently in the early research and planning phase.