This contribution proposes a methodology for the tactile representation of the Ideal City painting housed in the Galleria Nazionale delle Marche in Urbino, with the aim of developing a three-dimensional model that is perceptually legible and suitable for integration into a museum setting.
The workflow combines perspective analysis, digital modelling in a BIM environment as a geometric and informational framework, the use of artificial intelligence algorithms to generate depth maps, and rapid additive prototyping. Convolutional neural networks are employed as tools for preliminary inference rather than as autonomous interpretative systems: their outputs are critically assessed by comparing them with the perspective reconstruction and the historical-critical reading of the artwork. The process results in a perspective bas-relief designed for display in an exhibition context, fostering greater museum accessibility and inclusion for visitors.
