跳到正文
arXiv cs.AI· Xiaokang Ye, Siddhant Hitesh Mantri, Zimeng Chen, Edward Zhang, Zhaoxu Zheng, Yuanheng Li, Yizhao Chen, Tianyang Huang, Lianhui Qin·· 2 小时前AI 评分45

Code4Scene:评测编码智能体构建与编辑 3D 场景的基准

Code4Scene: Benchmarking Coding Agents for Constructing and Editing 3D Scenes

AI 导读

研究者推出 Code4Scene 基准,用 170 个 Unreal Engine 真人组装场景评测编码智能体的 3D 场景构建与编辑能力,v2 版本更新至 170 个案例并修订评测结果。

正文 · 原文

View PDF HTML (experimental)

Abstract:Frontier coding agents can now write and execute code that authors 3D environments, but whether they reliably understand 3D structure and precisely control scene state remains unclear. The generated 3D scene is a persistent, executable artifact: a convincing render can hide incorrect spatial relations, intersecting objects, or unintended modifications. We introduce Code4Scene, a benchmark of 170 Unreal Engine cases built from human-assembled scenes that evaluates coding agents on two complementary settings under a shared execution interface. Construction tests scene-level spatial reasoning from open-ended language specifications, where many realizations are valid; editing tests precise control of scene state, where the agent must recover the target scene from reference images while preserving everything else. Rather than scoring code or rendered views, Code4Scene evaluates the generated engine-native scene for task fulfillment, artifact integrity, and static physical validity, with edits additionally compared against withheld ground truth. Across 14 coding-agent systems on the 83-case public set, construction and editing performance are strongly correlated but not interchangeable (Spearman $\rho = 0.67$): Claude Fable 5.1 leads construction, Gemini 3.8 Flash leads editing. Spatial Composition is the weakest construction category for every agent, while editing remains imprecise: the best Repair F1 is only 0.562, and 34.2% of edits that fully recover the target still introduce unintended changes elsewhere in the scene. These results expose a gap between plausible 3D generation and reliable spatial reasoning and state control. The project webpage is at this https URL.
Comments: 46 pages, including appendices. v2 updates the benchmark to 170 cases and revises the evaluation results
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2609.36777 [cs.AI]
  (or arXiv:2609.36777v2 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2609.36777

arXiv-issued DOI via DataCite

Submission history

From: Xiaokang Ye [view email]
[v1] Tue, 29 Sep 2026 05:49:37 UTC (44,753 KB)
[v2] Tue, 17 Nov 2026 08:28:07 UTC (59,336 KB)

来源:arXiv cs.AI · arxiv.org