The perspective restitution of photographic images and historical iconographic sources represents a complex operation, grounded in the interpretation of implicit spatial structures and the geometric reconstruction of representations. The emergence of generative artificial intelligence tools and computer vision systems introduces new possibilities for automatic and semi-automatic analysis of perspective images, while at the same time exposing the process to forms of interpretive opacity and geometric approximation. This contribution investigates the potential of AI in the analysis and identification of perspective construction structures, verifying whether and to what extent such systems are capable of recognizing certain implicit geometric elements – the principal point, the horizon line, and the circle of distance – and assessing the risk of interpretive errors or the production of formally persuasive but geometrically incoherent and metrically unsound results. In relation to the theme of Disegno Furioso, artificial intelligence is interpreted as a technology with high generative potential but intrinsically unstable, capable of amplifying both the understanding and the distortion of the represented space. The article therefore proposes the definition of geometrically controlled work-flows in which automation is integrated within a critical process of graphic and geometric verification, restoring to drawing its role as an instrument for controlling and validating perspective hypotheses.
La restituzione prospettica di immagini fotografiche e fonti iconografiche storiche rappresenta un’operazione complessa, fondata sull’interpretazione delle strutture spaziali implicite e sulla ricostruzione geometrica delle rappresentazioni. L’emergere di strumenti di intelligenza artificiale generativa e di sistemi di computer vision introduce nuove possibilità di analisi automatica e semiautomatica delle immagini prospettiche, ma al contempo espone il processo a forme di opacità interpretativa e di approssimazione geometrica. Il contributo indaga le potenzialità dell’AI nell’analisi e nell’identificazione delle strutture della costruzione prospettica, verificando se e in che misura tali sistemi siano in grado di riconoscere alcuni elementi geometrici impliciti - punto principale, linea d’orizzonte e cerchio di distanza - e valutando il rischio di errori interpretativi o la produzione di esiti formalmente persuasivi ma privi di coerenza geometrica e rigore metrico. In relazione al tema del Disegno Furioso, l’intelligenza artificiale viene interpretata come tecnologia ad alto potenziale generativo ma intrinsecamente instabile, capace di amplificare tanto la comprensione quanto la distorsione dello spazio rappresentato. L’articolo propone quindi la definizione di workflow geometricamente controllati nei quali l’automatismo venga integrato all’interno di un processo critico di verifica grafica e geometrica, restituendo al disegno il ruolo di strumento di controllo e validazione delle ipotesi prospettiche.
Restituzione prospettica e intelligenza artificiale: tra automatismo e controllo geometrico | Perspective Restitution and Artificial Intelligence: Between Automation and Geometric Control
Stefano Chiarenza;Flavia Camagni;
2026-01-01
Abstract
The perspective restitution of photographic images and historical iconographic sources represents a complex operation, grounded in the interpretation of implicit spatial structures and the geometric reconstruction of representations. The emergence of generative artificial intelligence tools and computer vision systems introduces new possibilities for automatic and semi-automatic analysis of perspective images, while at the same time exposing the process to forms of interpretive opacity and geometric approximation. This contribution investigates the potential of AI in the analysis and identification of perspective construction structures, verifying whether and to what extent such systems are capable of recognizing certain implicit geometric elements – the principal point, the horizon line, and the circle of distance – and assessing the risk of interpretive errors or the production of formally persuasive but geometrically incoherent and metrically unsound results. In relation to the theme of Disegno Furioso, artificial intelligence is interpreted as a technology with high generative potential but intrinsically unstable, capable of amplifying both the understanding and the distortion of the represented space. The article therefore proposes the definition of geometrically controlled work-flows in which automation is integrated within a critical process of graphic and geometric verification, restoring to drawing its role as an instrument for controlling and validating perspective hypotheses.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


