# 3D-CoS: A New 3D Reconstruction Paradigm Based on VLM Code Synthesis

arXiv: https://arxiv.org/abs/2606.10478 · HTML: https://mlsanity.com/paper/2606.10478/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 28 of its 42 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Builds on (most cited first, 28 of 28)

- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [DreamFusion: Text-to-3D using 2D Diffusion](https://mlsanity.com/paper/2209.14988/) (2022)
- [ShapeNet: An Information-Rich 3D Model Repository](https://mlsanity.com/paper/1512.03012/) (2015)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [Code as Policies: Language Model Programs for Embodied Control](https://mlsanity.com/paper/2209.07753/) (2022)
- [LION: Latent Point Diffusion Models for 3D Shape Generation](https://mlsanity.com/paper/2210.06978/) (2022)
- [3DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Models](https://mlsanity.com/paper/2301.11445/) (2023)
- [InstantMesh: Efficient 3D Mesh Generation from a Single Image with Sparse-view Large Reconstruction Models](https://mlsanity.com/paper/2404.07191/) (2024)
- [CLIP-NeRF: Text-and-Image Driven Manipulation of Neural Radiance Fields](https://mlsanity.com/paper/2112.05139/) (2021)
- [Rodin: A Generative Model for Sculpting 3D Digital Avatars Using Diffusion](https://mlsanity.com/paper/2212.06135/) (2022)
- [Text-to-3D using Gaussian Splatting](https://mlsanity.com/paper/2309.16585/) (2023)
- [DeepCAD: A Deep Generative Network for Computer-Aided Design Models](https://mlsanity.com/paper/2105.09492/) (2021)
- [GaussianDreamer: Fast Generation from Text to 3D Gaussians by Bridging 2D and 3D Diffusion Models](https://mlsanity.com/paper/2310.08529/) (2023)
- [Unique3D: High-Quality and Efficient 3D Mesh Generation from a Single Image](https://mlsanity.com/paper/2405.20343/) (2024)
- [SceneCraft: An LLM Agent for Synthesizing 3D Scene as Blender Code](https://mlsanity.com/paper/2403.01248/) (2024)
- [Chain of Code: Reasoning with a Language Model-Augmented Code Emulator](https://mlsanity.com/paper/2312.04474/) (2023)
- [NeRF-VAE: A Geometry Aware 3D Scene Generative Model](https://mlsanity.com/paper/2104.00587/) (2021)
- [LN3DIFF++: Scalable Latent Neural Fields Diffusion for Speedy 3D Generation](https://mlsanity.com/paper/2403.12019/) (2024)
- [Learning to Infer and Execute 3D Shape Programs](https://mlsanity.com/paper/1901.02875/) (2019)
- [3D Shape Generation with Grid-based Implicit Functions](https://mlsanity.com/paper/2107.10607/) (2021)
- [DECOR-GAN: 3D Shape Detailization by Conditional Refinement](https://mlsanity.com/paper/2012.09159/) (2020)
- [BlenderGym: Benchmarking Foundational Model Systems for Graphics Editing](https://mlsanity.com/paper/2504.01786/) (2025)
- [BlenderLLM: Training Large Language Models for Computer-Aided Design with Self-improvement](https://mlsanity.com/paper/2412.14203/) (2024)
- [BlenderAlchemy: Editing 3D Graphics with Vision-Language Models](https://mlsanity.com/paper/2404.17672/) (2024)
- [LL3M: Large Language 3D Modelers](https://mlsanity.com/paper/2508.08228/) (2025)
- [MeshCoder: LLM-Powered Structured Mesh Code Generation from Point Clouds](https://mlsanity.com/paper/2508.14879/) (2025)
- [TextSplat: Text-Guided Semantic Fusion for Generalizable Gaussian Splatting](https://mlsanity.com/paper/2504.09588/) (2025)

## Built on by (most cited first, 1 of 1)

- [3DHarnessBench: Probing Agentic 3D-to-Code Capabilities of Frontier Vision-Language Models](https://mlsanity.com/paper/2609.06535/) (2026)

## Abstract (arXiv)

Most recent 3D reconstruction and editing systems operate on implicit and explicit representations such as NeRF, point clouds, or meshes. While these representations enable high-fidelity rendering, they are fundamentally low-level and hard to control programmatically. In contrast, we propose and systematically evaluate a new 3D reconstruction paradigm, 3D Code Synthesis (3D-CoS), where 3D assets are constructed as executable Blender code, a programmatic and interpretable medium. To assess how well current VLMs can use code to represent 3D objects, we evaluate representative open-source and closed-source VLMs in code-based reconstruction under a unified protocol. We further introduce a suite of structured code-synthesis workflows, including blueprint-based planning, Retrieval-Augmented Generation (RAG) over Blender API documentation, few-shot geometric demonstrations, and a component-level Agent workflow for part-wise code generation. To demonstrate the unique advantages of this representation, we further evaluate localized text-driven modifications and compare our code-based edits with a point-cloud-based 3D editing baseline. Our study shows that code as a 3D representation offers strong controllability and locality, yielding stronger edit fidelity and better preservation of unedited regions in our targeted editing evaluation. Our work also analyzes the potential of this paradigm, delineates the current capability frontier of VLMs for programmatic 3D modeling, and highlights code synthesis as a promising direction for editable 3D reconstruction.