Presentation

· Contributors · Organizations · Search Program

Latent L-systems: Transformer-based Tree Generator

DescriptionThe study introduces a Transformer-based deep learning framework for generating 3D tree models using L-systems. Trained on 155k geometries, this neural model replaces manual rule creation, accurately learning and generating L-strings. It shows high fidelity in branching patterns and geometric features, enhancing realistic tree model generation.

Authors

Jae Joong Lee

Purdue University

Bosheng Li

Purdue University