# Modeling 3D Shapes by Reinforcement Learning

arXiv: https://arxiv.org/abs/2003.12397 · HTML: https://mlsanity.com/paper/2003.12397/

## Key facts

- Cited by 21 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Mar 2020.

## Built on by (most cited first, 21 of 21)

- [SolidGen: An Autoregressive Model for Direct B-rep Synthesis](https://mlsanity.com/paper/2203.13944/) (2022)
- [Part123: Part-aware 3D Reconstruction from a Single-view Image](https://mlsanity.com/paper/2405.16888/) (2024)
- [3D-Aware Object Goal Navigation via Simultaneous Exploration and Identification](https://mlsanity.com/paper/2212.00338/) (2022)
- [CADDreamer: CAD Object Generation from Single-view Images](https://mlsanity.com/paper/2502.20732/) (2025)
- [CPFN: Cascaded Primitive Fitting Networks for High-Resolution Point Clouds](https://mlsanity.com/paper/2109.00113/) (2021)
- [Integrating Reinforcement Learning with Visual Generative Models: Foundations and Advances](https://mlsanity.com/paper/2508.10316/) (2025)
- [Reinforcement Learning for Generative AI: A Survey](https://mlsanity.com/paper/2308.14328/) (2023)
- [Coverage Axis++: Efficient Inner Point Selection for 3D Shape Skeletonization](https://mlsanity.com/paper/2401.12946/) (2024)
- [MeshMosaic: Scaling Artist Mesh Generation via Local-to-Global Assembly](https://mlsanity.com/paper/2509.19995/) (2025)
- [Structured Outdoor Architecture Reconstruction by Exploration and Classification](https://mlsanity.com/paper/2108.07990/) (2021)
- [A Review of Deep Learning-Powered Mesh Reconstruction Methods](https://mlsanity.com/paper/2303.02879/) (2023)
- [Generative Thermal Design Through Boundary Representation and Multi-Agent Cooperative Environment](https://mlsanity.com/paper/2208.07952/) (2022)
- [Cut-and-Approximate: 3D Shape Reconstruction from Planar Cross-sections with Deep Reinforcement Learning](https://mlsanity.com/paper/2210.12509/) (2022)
- [Reinforcement Learning for Block Decomposition of CAD Models](https://mlsanity.com/paper/2302.11066/) (2023)
- [Split, Merge, and Refine: Fitting Tight Bounding Boxes via Over-Segmentation and Iterative Search](https://mlsanity.com/paper/2304.04336/) (2023)
- [Intelligent Generation of Graphical Game Assets: A Conceptual Framework and Systematic Review of the State of the Art](https://mlsanity.com/paper/2311.10129/) (2023)
- [AutoPoly: Predicting a Polygonal Mesh Construction Sequence from a Silhouette Image](https://mlsanity.com/paper/2203.15233/) (2022)
- [Fast-Image2Point: Towards Real-Time Point Cloud Reconstruction of a Single Image using 3D Supervision](https://mlsanity.com/paper/2209.10029/) (2022)
- [From Two-Dimensional to Three-Dimensional Environment with Q-Learning: Modeling Autonomous Navigation with Reinforcement Learning and no Libraries](https://mlsanity.com/paper/2403.18219/) (2024)
- [DiffeoMorph: Learning to Morph 3D Shapes Using Differentiable Agent-Based Simulations](https://mlsanity.com/paper/2512.17129/) (2025)
- [Q-MAGIC: Quad-only Mesh Autoregressive Generation via Singularity-Aware Circular Relative Offset Tokenization](https://mlsanity.com/paper/2610.19777/) (2026)

## Abstract (arXiv)

We explore how to enable machines to model 3D shapes like human modelers using deep reinforcement learning (RL). In 3D modeling software like Maya, a modeler usually creates a mesh model in two steps: (1) approximating the shape using a set of primitives; (2) editing the meshes of the primitives to create detailed geometry. Inspired by such artist-based modeling, we propose a two-step neural framework based on RL to learn 3D modeling policies. By taking actions and collecting rewards in an interactive environment, the agents first learn to parse a target shape into primitives and then to edit the geometry. To effectively train the modeling agents, we introduce a novel training algorithm that combines heuristic policy, imitation learning and reinforcement learning. Our experiments show that the agents can learn good policies to produce regular and structure-aware mesh models, which demonstrates the feasibility and effectiveness of the proposed RL framework.