A body of work on perception, measurement and the images produced between them. The research question is simple: when an instrument makes something visible that no human eye can directly see, who is the observer of that image?
Three existing bodies of work establish the line of enquiry behind this proposal: an operational thermal-paper archive, a moving-image study of correction and standardisation, and a realised field-sensing system.
The sensing system can be followed beyond this page through its ongoing online archive and field-instrument view.
A 24 × 32 measured thermal field is translated into contour and density on continuous thermal paper. The system does not identify faces, infer health or emotion, or synthetically repair failed measurements.
Digital records may remain stable while the thermal-paper record physically fades. The asymmetry between persistent measurement and fading matter is part of the work.
A hand-drawn graphite image passes through a sequence of correction operations: grain reduction, edge stabilisation, interval equalisation, symmetry, perspective and surface smoothing. The more “correct” the space becomes, the more the authored structure disappears.
The main arc comes from the Ferris wheel at Suseongmot, one stable visual coordinate during a childhood of repeated moves around Daegu.
An ESP32 field instrument listens to anonymous aggregate 2.4 GHz Wi-Fi conditions. The measured field reorganises structural traits drawn from 101 original works — lines, density, voids and fractures — while site-specific states accumulate online rather than returning to a neutral state.
No register contains the whole field. Human drawing and machine sensing remain incomplete records of the same site.
External working archive. The latest field state may change as the instrument connects, while accumulated records remain part of the project.
These six works establish continuity across structural perception, unstable evidence, site sensing, no-reset accumulation and long-duration data translation.
Historical memory, generated imagery and live signals disturb the reliability of a single image. Relevant here because representation is treated as a negotiated state rather than evidence that arrives intact.
Watch work ↗
Ink-like diffusion, density and boundary memory become the event before hue does. A concise proof of the artist’s structural visual grammar.
Watch work ↗
Field drawings, labour time, visitor signals and place-specific research accumulate through a system that does not return to neutral. Previous traces remain active for the next visitor.
Environmental records pass through a fragile body as pressure and fracture. The work shows a longer practice of treating data as a force that changes structure, not as explanatory graphics.
Watch work ↗
Ilwolobongdo becomes a data landscape under seismic and temporal pressure. Cultural form survives not as a fixed image but as structure, vibration and record.
Watch work ↗
The artist’s own hand precedes the model. 101 drawings establish an authored boundary for AI and a repeatable grammar of interval, density, fracture, surface and pressure.
The portfolio’s works are connected by a method rather than by one look: self-authored sources, structural perception, incomplete registers and no-reset traces.
101 hand-drawn works → structural traits → controlled AI extension. The source remains artist-owned and visible.
Density · interval · contour · fracture · texture · luminance · pressure. Colour is not removed; it is not allowed to carry the structure alone.
Red-green colour vision deficiency does not make one vision more truthful than another. It does make the limits of a supposedly neutral visual reference tangible. Thermal imaging sharpens that problem because an invisible measurement must be translated before any human can see it.
That experience is not the conclusion of the project. It is the position from which I ask how measurement becomes representation and how one visual convention becomes legible to one observer but not another.
Dalí: “to look is a form of invention.” Four years later, the CIE formalised a standard colorimetric observer for normal colour vision. The project places these two histories beside each other without claiming a causal link.
An infrared signal is not a visible-colour image. Detection, calibration, thresholding and rendering mediate what a human observer ultimately reads.
My perception does not match the normal-colour-vision population that the CIE standard observer is intended to represent.
Hold the measurement constant. Re-map visual translation toward density, interval, contour, luminance and pressure. Then compare what becomes more or less legible — and for whom.
Research visualisation. Structural translation. Fading material trace. None is presented as the correct image.
The research begins where invisible radiation becomes a measurable signal, before that signal is rendered as an image for a human observer.
The residency question is therefore operational: keep the measurement constant, change the visual translation, and compare what becomes legible, what disappears, and for whom.
The representation used to make a measurement interpretable within its scientific context.
Density, interval, contour, luminance, texture and pressure reorganise the same measurement into a different information hierarchy.
The measurement becomes thermal paper: a physical record whose legibility changes over time.
Research reference: CIE 1931 Standard Colorimetric Observer · ICFO advanced imaging/sensing and recent MWIR/LWIR bolometric detection research. No ICFO affiliation, access or unpublished data is implied.
Korean media artist working across moving image, generative systems, spatial installation, public media and sensor-based environments. His practice asks how perception, memory and invisible signals are translated by technological systems — and what changes when those translations become images, archives or data.