Polymorph II developed from an investigation into multisensory, distributed AI operating across heterogeneous material substrates. It treats these configurations as complex computational ecosystems in which physical and synthetic processes co-evolve without fixed targets or predefined relations. The system is structured to examine what occurs when generative models are not bound to a singular objective, but instead assemble and reorganise their own operational relationships. Frictions between domains, including misalignments between models, signals, and material conditions, as well as noise and interference, drive continuous transformations within the system.
It began as a project at the Artificial and Distributed Intelligence Lab (AiDLab Hong Kong Design Lab and Royal College of Art), developed through an art-led investigation into distributed synthetic systems operating across multiple sensory and computational domains. Early work drew on interspecies models of distributed coordination and morphogenetic pattern formation, including cephalopod sensing and avian swarming, where perception and response are not organised through centralised control. These principles informed the design of a system in which decision-making arises from interactions between components and the environment, leading to the construction of a continuously retrained complex adaptive system attuned to its own physical conditions.
The installation integrates continuously retrained models, including Stable Diffusion and RAVE, with real-time sensors, camera inputs, and environmental data streams. Its material components, including metal sheets, water tanks, conductive fibres, sound resonators, cameras, and speakers, are arranged so that each element functions as both input and output. Signals circulate through the system, are transformed, and return, generating feedback that conditions subsequent states.
Variations in light, sound, air movement, vibration, and electromagnetic activity influence both image and sound generation. These outputs are reintroduced into the system, allowing its behaviour to shift in response to its own activity. Across iterations, the system reorganises through changing configurations of materials and signals, forming distinct patterns as a consequence of its internal dynamics.
It began as a project at the Artificial and Distributed Intelligence Lab (AiDLab Hong Kong Design Lab and Royal College of Art), developed through an art-led investigation into distributed synthetic systems operating across multiple sensory and computational domains. Early work drew on interspecies models of distributed coordination and morphogenetic pattern formation, including cephalopod sensing and avian swarming, where perception and response are not organised through centralised control. These principles informed the design of a system in which decision-making arises from interactions between components and the environment, leading to the construction of a continuously retrained complex adaptive system attuned to its own physical conditions.
The installation integrates continuously retrained models, including Stable Diffusion and RAVE, with real-time sensors, camera inputs, and environmental data streams. Its material components, including metal sheets, water tanks, conductive fibres, sound resonators, cameras, and speakers, are arranged so that each element functions as both input and output. Signals circulate through the system, are transformed, and return, generating feedback that conditions subsequent states.
Variations in light, sound, air movement, vibration, and electromagnetic activity influence both image and sound generation. These outputs are reintroduced into the system, allowing its behaviour to shift in response to its own activity. Across iterations, the system reorganises through changing configurations of materials and signals, forming distinct patterns as a consequence of its internal dynamics.