Data-Driven Generative Art Installation for Wellbeing
This project sits at the intersection of data-driven art practice, empirical aesthetics, and arts in health, bringing together live environmental data, psychology research, and computational visual practice to produce a continuously evolving mixed-media coastal landscape for hospital environments within South Wales.
At the heart of the project is an idea about breaking down the boundary between institution and environment. Hospitals are sealed spaces — climate controlled, artificially lit, cut off from the rhythms of the outside world. Time can feel warped; minutes stretch, days blur, and normal markers of passing time can disappear when in an anxious or worried state.
This project reintroduces those temporal rhythms using live tidal, meteorological and lunar data from the Penarth coastal monitoring station under the National Network of Regional Coastal Monitoring Programme (NNRCMP) to bring the natural world inside hospital walls, translated into a slowly shifting visual landscape.
Practice based research
Psychology/User Experience (UX) based research
Computational art inspiration comes from artists who treat data as material for affective experience, such as:
I plan to implement a surrealist quality into the artwork drawing on the distorted temporality of Dalí's works and the dreamlike landscapes of Studio Six'n'Five, mapping surrealism's language of warped time onto the hospital experience — not to unsettle, but to afford reflection.
Flow (2025) — Zheyuan Zhang, West Middlesex University Hospital
Wind Map (2012) — Viegas & Wattenberg
The River is a Circle (2022) — Marina Zurkow
A 1 hour 15 minute focus group discussion was conducted online via Teams in April 2026 with 6 health workers across departments at local South Wales hospitals. All participants had an interest or experience in initiatives promoting art within hospitals for wellbeing.
The 1st focus group and empirical aesthetics research both feed into the data driven art practice →
Secondary research into the effects of water movement and textures on human emotions informed the visual and motion parameters of the digital sea overlay.
Amplitude: Toet et al (2011) found large wave amplitudes feel unpleasant to participants so wave amplitude is kept low across all weather conditions, with some differentiation for stormy vs calm conditions but wave amplitude will be capped even when stormy
Perlin Noise and Layering: Toet et al (2011) found complex but organic motion with water is relaxing — using Perlin noise across layers creates uniform, organic motion simultaneously
Sine Waves: Toet et al (2011) found regular rhythmic motion is the most pleasant — multiple sine waves stacked at slightly different speeds and offsets achieve this
The wave motion parameters in these sketches are informed by the findings of a single paper (Toet et al., 2011). Further research will include a more comprehensive literature review to establish whether these findings are consistent across the broader empirical aesthetics literature.
Direct observation and iterative computational experimentation informed each scene element in turn.
WebGL used to experiment with 3D pebble forms, then moved to 2D circular designs informed by focus group feedback on fractals, direct observational sketches, and beach photography.
These sketches uses WebGL to render a 3D sphere representing the moon, adjusting the lighting strength and direction onto the sphere based on the current lunar phase. A picture of the moon from NASA is then used as texture across the sphere. Positioned using SunCalc astronomical library.
Experimenting with using photographs taken of the South Wales coast and creating a digital overlay of the sea controlled by real-time data. Based on focus group feedback, I moved from purely rendered forms to a more realistic visual style combining digital and real coastal landscape. I would ideally like to render more realistic looking sea without needing to use GPU due to constraints on how the artwork would be powered.
Live installation sketch — driven by real-time tidal, lunar and weather data from Penarth. Please allow a few seconds to load. This prototype is experimenting with mixed media - a real photograph of a viewpoint called Lavernock point which would be familiar to most people who live in cardiff/nearby due to two famous islands steepholme and flatholme being in shot.
tideHeight Controls how far the 'rendered water' covers the pebble beach. Low tide exposes the rocks, high tide submerges them.
; windSpeed Scales wave movement speed and amplitude via windFactor. Stronger wind = faster, more active waves. Waves will never look too choppy to keep the relaxing nature of the scene.
phase_fractionSets the directional light angle on the 3D moon sphere so the lit/shadow side matches the real moon phase tonight.
, illumination_percent and moon_name drive the moon appearanceazimuth + altitude positions the moon as it would be seen from this viewpoint; getDaylight() drives the day/night sky gradient
Currently in progress as part of Computational Arts Masters study at Goldsmiths, University of London.