Welcome to the GitHub repository for the Metaball Variational Auto-encoder (MetaballVAE) algorithm.
MetaballVAE is an particle generation algorithm, which can learn particle morpholgoies from seveal parental particles and generate inifinte number of unexistent, new particles with coherent shape feature distributions. The generated particles are in Metaball-form and can be loaded into Metaball-Imaging Discrete Element method MI-DELBM for direct simulations of fluid-particle systems.
This repository contains the associated manuscript, code, data, and supplementary materials. We hope this work will be helpful to the research community and encourage further collaboration and exploration in the field. If you find my work of interest, please feel free to cite, implement, or reuse it as needed. Your engagement and collaboration are highly encouraged.
The paper of this algorithm can be found through this doi link. An pdf "MI_and_MetaballVAE" is also attached in this repository.
The Code and Data will be released soon.
The Intro video will be released soon.
@article{zhao2023reconstruction,
title={Reconstruction and generation of 3D realistic soil particles with metaball descriptor},
author={Zhao, Yifeng and Gao, Xiangbo and Zhang, Pei and Lei, Liang and Li, Stan Z and Galindo-Torres, SA},
journal={Computers and Geotechnics},
volume={161},
pages={105564},
year={2023},
publisher={Elsevier}
}