Codex : folio 307

Research output: Creative and Literary Works in Non-textual Form (RGC: 42, 43, 44, 46, 47)43_Film, video

View graph of relations



Original languageEnglish
Publication statusPublished - 20 Sep 2019


TitleLeonardo da Vinci. Art & Science Then & Now
LocationCity University of Hong Kong
PlaceHong Kong
Period20 September - 15 December 2019


Codex: folio 307 consists of two prints inspired by folio 307 in Leonardo's Codex Atlanticus, in the exhibition. As in Leonardo's drawing, the pattern in the first print is created by following a specific set of rules. It is then modified by adding elements that smooth it out and so remove an excess of order. Two rows of drawings show patterns derived from a recursive procedure. In addition to the presence of color, the lower row differs from the upper one in that the underlying rule allows for randomness. In the second print, the overall pattern in Leonardo's folio 307 is reproduced but with a different motif. This motif is less symmetric than Leonardo's and is further subject to variations derived from a machine learning algorithm known as Variational Autoencoder (VAE) trained on some of the images used in the first print. Both prints are inscribed with numbers representing the Minkowski dimension of some of the visual patterns. This dimension increases, in our figures, with their complexity. Each print represents a different programming paradigm. The first print follows a classical approach. The programmers wrote an explicit set of rules that models a desired visual pattern. In the second print, the programmers chose a machine learning procedure, which automatically generates a probabilistic model of the visual pattern. The contrast between the two prints foregrounds recent changes in the ongoing history of generative art. The print also contains quotations from Leonardo, Sigmund Freud, Paolo Pasquinelli, and Wikipedia, as well as technical information about the procedures employed.

Research Area(s)

  • Computational print

Bibliographic Note

Record validation is based on the information provided by the researcher(s) concerned. Suggestions for further amendments or supplementary information can be sent to