Meandering River, an ambitious ongoing project by digital design collective 
Onformative, will be presented at the annual Google Development festival Google I/O from 7 to 9 May in California.
Meandering River is an audiovisual installation which explores rivers: their behaviour, their power, their transforming effect on the Earth’s surface, the way water flows and shifts. Many of the civilizations and transport networks of the world were built around rivers and could not have survived without them. Rivers are also capable of destroying cities, just as cities are capable of irreversibly changing rivers via redirection and via pollution.
Meandering River reflects on the globally-transforming properties of computation at the same time as it explores rivers. Global communication networks shape social interactions and city infrastructures; they have capacity to strengthen and restore or to shut down, to emulsify or to homogenize.
The starting point for Meandering River was the transforming effect of water on land. Academic and scientific research on erosion and deposition and their effects on soil colour formations informed the development of algorithms which simulated the behaviour of rivers and their rhythmic influence on their surroundings.
Displayed across six screens, the piece comprises visuals which are generated in real-time by a bespoke rendering pipeline and a soundtrack produced using Artificial Intelligence. Satellite images inspired the aesthetics of the work; algorithms and shaders, applied in real-time, simulate geological features such as meander scars and oxbow lakes, while vegetation and the texture of the Earth’s surface were added in post-production. The river simulation transforms in real time, changing iteratively and altering the simulated landscape around it.
The score was created in collaboration with kling klang klong, using the Google Magenta Performance RNN learning model and a custom-built procedural system and incorporating live work by pianists. The piano work, improvised in response to the transformations of the visuals, became the inventory for the Artificial Intelligence to learn from and respond to in turn. A complex temporal relationship evolved between score and visuals, and six visual modes with sonic accompaniments were selected as final project outcomes.
