Invisible City

Concept

From Walter Benjamin’s unfinished Arcades Project (1927–1940) to Bruno Latour’s web-based essay Paris: Invisible City (1998), philosophers, architects, and artists have explored how modern cities take shape through ways of moving, seeing, and remembering. Benjamin’s flâneur wanders the arcades of Paris with detached curiosity, attuned to fleeting impressions and ambient rhythms. Latour, by contrast, assembles Paris as a digital montage of viewpoints, suggesting that urban space emerges through the accumulation of partial, situated perspectives.

In recent decades, advances in 3D scanning, virtual reality, and machine learning have expanded these inquiries into immersive, multisensory media. Palimpsest (Torisu et al., 2016), developed at the Bartlett School of Architecture, layers LiDAR scans with residents’ oral histories in a VR environment, reconstructing a London neighborhood as overlapping strata of remembrance. Zeynep Abes’s Memory Place (2020) captures a family apartment through 3D scanning and presents it as an immersive video, where the fragmented visual field and shifting perspectives evoke the fragility of memory. Refik Anadol Studio scales this inquiry to the urban level: works such as Machine Hallucination: NYC (2019) and Latent Being (2019) transform millions of photographs into large-scale audiovisual installations, rendering the city as a fluid archive of collective memory.

While these projects powerfully visualize memory, they often privilege surface affect or technological spectacle over embodied spatial experience. This proposal joins that lineage with a more critical lens, positioning contemporary computational methods within a broader intellectual tradition that understands urban form as inseparable from the acts of moving, seeing, and remembering that animate it. Rather than rendering memory as a static archive, this project asks: How does spatial memory form through embodied movement and perception in a city—and how might we visualize this dynamic process through computational design?

“Invisible City: Simulating Movement, Perception, and Memory” explores this question by adopting three distinct modes of movement—wandering, walking, and dwelling—each grounded in a philosophical framework. The flâneur, drawn from Charles Baudelaire (1863) and Benjamin (1927–1940), engages the city through drifting attention and contemplative encounter. Michel de Certeau’s (1980) pedestrian practices frame walking as a spatial tactic, where everyday movement inscribes meaning. Martin Heidegger’s (1951) notion of dwelling emphasizes rootedness, presence, and the deep temporality of inhabiting space. Each mode proposes a different relationship between body, city, and memory.

The project takes a sparse point-cloud form—an architectural model constructed not through solid mass but through fragments, gaps, and ephemeral outlines. Long used in digital reconstruction, the point cloud also holds expressive potential: its ghostly visual character has been embraced by artists to evoke the dreamy and unstable texture of memory. Here, it becomes both a technical format and a conceptual metaphor.

To foreground bodily experience, Invisible City draws from phenomenology, making the perceptual position of the observer central to the simulation. A dual-screen video installation presents two perspectives: one screen follows a first-person traversal through a spectral city, where only visible areas are rendered fully while unseen spaces dissolve into shadow; the other displays a top-down memory map that evolves over time, with locations brightening, fading, or distorting based on their mnemonic imprint.

The simulation is informed by Aldo Rossi’s (1966) idea of the city as a repository of collective memory, built from recurrent architectural forms that endure across time. While not set in a specific historical or cultural context, Invisible City borrows from pedestrian-oriented typologies around the world: alleyways, porticos, arcades, and colonnades. These forms, with their rhythms of enclosure and exposure, foster intimate, ground-level encounters that resonate with the project’s focus on embodied perception.

The mechanism for spatial memory draws on Kevin Lynch’s (1960) theory of imageability, in which paths, edges, districts, nodes, and landmarks anchor the mental image of a city. These elements appear in the simulation as architectural motifs that shape how space is remembered and forgotten. To generate the cityscape itself, the project uses a computational design process grounded in shape grammar (Stiny, 1982), which encodes architectural logic into rule-based spatial sequences. Unlike procedural content generation in games or film—which often prioritizes scale or randomness—this system attends closely to architectural logic, generating varied spatial sequences with attention to typological clarity. The modeling process is implemented in Grasshopper, a parametric design environment commonly used in architectural workflows. The simulation is then rendered in Unity 3D, where the visual and mnemonic layers are dynamically linked.

Gallery