The Digital Image in Architecture as an Autonomous Reality: A Comparative Reading of Dionisio González and Filip Dujardin

Abstract

Since antiquity, architectural images have mediated between the imagined and the materialized, acquiring value from their capacity to stand in for built form. At times, however, this subordinated relationship is suspended: the image acts not as a mere reflection but as an autonomous reality, generating architectural meaning irrespective of constructability. Historical precedents such as Giovanni Battista Piranesi’s Carceri and Étienne‑Louis Boullée’s Cenotaphs already staged credible yet unbuilt architectures that shaped disciplinary imagination; in the twentieth century, John Hejduk and Lebbeus Woods extended this lineage with textual visual constructions and speculative architectures that operated discursively without seeking material realization. This paper approaches the conditions under which certain digital images can operate as autonomous realities, that is, produce architectural discourse without depending on constructability. Through a comparative analysis of two corpora—Dionisio González’s Cartografias para a remoção and Filip Dujardin’s Fictions—the paper introduces operational criteria (verisimilitude, coherence, operativity, and critical narrative) to discuss image autonomy and its limits. The analysis shows that, although autonomy manifests differentially—in González, as a critical device addressing the urban condition; in Dujardin, as formal exploration that tests disciplinary canons—high verisimilitude supports autonomy while decoupling value from possibility. As a result, an epistemological shift in architectural evaluation emerges: validation moves from a regime anchored in constructability to one grounded in discursive coherence and operational efficacy of the image itself. This clarification suggests how digital images can inform pedagogy, competitions, and critical practice, while noting that autonomy here is conceptual operational rather than a denial of built work. Although grounded in two authors, the proposed criteria are portable and invite further testing across other digital practices.