A generalized framework for procedural generation of three-dimensional static and dynamic plant model geometries

TL;DR

This paper introduces a flexible, efficient framework for generating realistic 3D plant models that can be dynamically animated, encoded for machine learning, and include collision physics, implemented in open-source tools.

Contribution

It presents a generalized, botanically-consistent model framework for procedural 3D plant generation that reduces creation time and supports dynamic, machine learning-compatible representations.

Findings

Framework supports dynamic plant geometries with realistic physics

Implemented in open-source Helios systems for easy use

Enables encoding plant structures for machine learning applications

Abstract

This work presents a new framework for procedural generation of dynamic 3D plant model geometries, which has been implemented in the Helios modeling system. Key goals of this work were to develop a model that 1) has a generalized set of parameters that are conserved across species, which are botanically-consistent and readily measurable; 2) significantly reduces the time and effort needed to create photorealistic, dynamically evolving plant models; 3) allows for encoding of the entire plant structure into a character-based representation that can integrated with machine learning models, and 4) includes realistic and computationally efficient collision physics. A model framework that satisfies these specifications is presented in this report. The model was implemented in the Helios C++ and PyHelios Python frameworks, which are open-source libraries that can be used to generate 3D plant geometries based on this model.