What the work is
A series of site-specific, real-time, data-driven video installations that use light as material and data as pigment. Each transforms architectural surfaces, live performance, or archival knowledge into immersive visual fields.
How it was made
Anadol built installations using projection mapping, custom software, neural networks (including StyleGAN and recurrent neural networks), quantum computing tools, and sensor-integrated systems. He tracked conductors’ heartbeats and movements with biometric sensors and 3D cameras. He trained AI on large-scale cultural archives — including 1.7 million items from 40,000 publications — and on neuroscience data such as EEG scans from an anonymous Alzheimer’s dataset.
What works
The integration of biometric input (e.g., conductor’s heartbeat) with generative visuals creates temporal alignment between human physiology and machine output. Projection mapping across all room surfaces — without mirrors — achieves perceptual infinity. Using neuroscience datasets to visualise memory decay grounds abstraction in clinical evidence.
What does not
The work does not produce standardised tools, reusable code libraries, or documented design processes. It does not engage material constraints like fabrication tolerances, user interaction fidelity, or long-term maintenance. It does not resolve tensions between real-time data responsiveness and narrative coherence.
What it influenced
It influenced how institutions frame AI as a spatial and sensory medium — not just a predictive tool. It established data painting as a category for real-time, site-specific visual translation of intangible sources. It shifted expectations for orchestral performance environments and museum-based data interpretation.
Is it worth your time
Yes — if you need to understand how algorithmic perception reshapes architectural space and collective memory. No — if you expect durable objects, reproducible interfaces, or design systems transferable beyond singular commissions.