# 3D Shape Synthesis for Conceptual Design and Optimization Using Variational Autoencoders

arXiv: https://arxiv.org/abs/1904.07964 · HTML: https://mlsanity.com/paper/1904.07964/

## Key facts

- Cited by 18 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Apr 2019.

## Built on by (most cited first, 18 of 18)

- [Deep Generative Models in Engineering Design: A Review](https://mlsanity.com/paper/2110.10863/) (2021)
- [Uncertainty Quantification in Machine Learning for Engineering Design and Health Prognostics: A Tutorial](https://mlsanity.com/paper/2305.04933/) (2023)
- [FakeNews: GAN-based generation of realistic 3D volumetric data -- A systematic review and taxonomy](https://mlsanity.com/paper/2207.01390/) (2022)
- [Conceptual Design Generation Using Large Language Models](https://mlsanity.com/paper/2306.01779/) (2023)
- [Constraining Generative Models for Engineering Design with Negative Data](https://mlsanity.com/paper/2306.15166/) (2023)
- [t-METASET: Tailoring Property Bias of Large-Scale Metamaterial Datasets through Active Learning](https://mlsanity.com/paper/2202.10565/) (2022)
- [Range-GAN: Range-Constrained Generative Adversarial Network for Conditioned Design Synthesis](https://mlsanity.com/paper/2103.06230/) (2021)
- [Learning to design without prior data: Discovering generalizable design strategies using deep learning and tree search](https://mlsanity.com/paper/2211.15068/) (2022)
- [Design Space Exploration and Explanation via Conditional Variational Autoencoders in Meta-model-based Conceptual Design of Pedestrian Bridges](https://mlsanity.com/paper/2211.16406/) (2022)
- [Design Strategy Network: A deep hierarchical framework to represent generative design strategies in complex action spaces](https://mlsanity.com/paper/2110.03760/) (2021)
- [Generative VS non-Generative Models in Engineering Shape Optimization](https://mlsanity.com/paper/2402.08540/) (2024)
- [Exploring the Potentials and Challenges of Deep Generative Models in Product Design Conception](https://mlsanity.com/paper/2407.11104/) (2024)
- [A Reward-Directed Diffusion Framework for Generative Design Optimization](https://mlsanity.com/paper/2508.01509/) (2025)
- [A Generative Neural Network Approach for 3D Multi-Criteria Design Generation and Optimization of an Engine Mount for an Unmanned Air Vehicle](https://mlsanity.com/paper/2311.03414/) (2023)
- [Data-driven intelligent computational design for products: Method, techniques, and applications](https://mlsanity.com/paper/2301.12382/) (2023)
- [Data-driven topology design based on principal component analysis for 3D structural design problems](https://mlsanity.com/paper/2409.01607/) (2024)
- [Enhanced Data-driven Topology Design Methodology with Multi-level Mesh and Correlation-based Mutation for Stress-related Multi-objective Optimization](https://mlsanity.com/paper/2504.14790/) (2025)
- [Masked Conditioning for Deep Generative Models](https://mlsanity.com/paper/2505.16725/) (2025)

## Abstract (arXiv)

We propose a data-driven 3D shape design method that can learn a generative model from a corpus of existing designs, and use this model to produce a wide range of new designs. The approach learns an encoding of the samples in the training corpus using an unsupervised variational autoencoder-decoder architecture, without the need for an explicit parametric representation of the original designs. To facilitate the generation of smooth final surfaces, we develop a 3D shape representation based on a distance transformation of the original 3D data, rather than using the commonly utilized binary voxel representation. Once established, the generator maps the latent space representations to the high-dimensional distance transformation fields, which are then automatically surfaced to produce 3D representations amenable to physics simulations or other objective function evaluation modules. We demonstrate our approach for the computational design of gliders that are optimized to attain prescribed performance scores. Our results show that when combined with genetic optimization, the proposed approach can generate a rich set of candidate concept designs that achieve prescribed functional goals, even when the original dataset has only a few or no solutions that achieve these goals.