TreeGAN
Project Description
TreeGAN is a creative work comprising drawings, animations, and 3D-printed sculptures that examines the process of how machine learning creates 3-dimensional objects. It is also accompanied by a peer-reviewed journal article (Multimedia Tools and Applications, forthcoming see Appendix 16) and published dataset (the 3D Tree Dataset). When this project began in 2020, there were relatively few art projects that used machine learning to produce 3D objects and even fewer that were trained on 3D objects (as opposed to synthesising 3D forms from 2D images), partly due to the scarcity of conditional datasets of 3D objects. Previous studies for 3D machine learning tended to focus on geometrically simple objects such as IKEA furniture (Lim et al 2013) and industrial objects (Wu et al 2016). This project involved synthesising a dataset of 3D objects of geometrically complex, but highly recognisable objects - trees. Trees are often used in visual art as metaphors for the human experience, from the scholarly pines of Chinese ink painting (Clunas 2002) (McMahon 2003) to the martyred oaks of German Romanticism (Rosenblum 1975), and thus add an empathetic layer to this formal exploration. Three-dimensional trees are easy to produce on a large scale using Lindenmayer systems. I produced 76 unique tree templates, based on art historical references, and exported 350 random variations of these templates, giving us a dataset of just over 26,000 3D trees.
By exporting 3D files from various epochs of the training process, it was possible to observe the transition of beautiful abstractions as the system progressed from random 3D noise to recognizable trees, a process I likened to the analytical cubism of Picasso and Braque in the early 20th century, where one could observe a new technological system developing its form of figuration. TreeGAN has been widely exhibited across Hong Kong and internationally and is the subject of a forthcoming article in the journal Multimedia Tools and Applications - an A-ranked journal (H-Index 72), which will include the publication of the 3D Tree Dataset (n = 26,000). An early version of the work was accepted into the “Constructing Contexts” exhibition at City University Hong Kong as part of the Art Machines 2: International Symposium on Machine Learning and Art (2022). TreeGAN was accepted and exhibited at SiGGRAPH Asia 2023. TreeGAN was also a major feature of the acclaimed Machine Visions exhibition at the Osage Art Foundation (2022-2023).
Video
Project Investigator
Professor Peter NELSON (Academy of Visual Arts)
Publication
“The 3D Tree Dataset: An Artistic Experiment Using a Voxel-Based GAN”, Multimedia Tools and Applications, 30th September, 2024. https://link.springer.com/article/10.1007/s11042-024-20304-w
_1735890602.png)