4DCodeBench: Benchmarking Agents on Inverse Graphics of Dynamic Scenes

• 类型：arxiv
• 标识：2610.03715
• 链接：https://arxiv.org/abs/2610.03715
• 主分类：evaluation
• 形态：benchmark
• TLDR：We introduce 4DCodeBench, a benchmark for 4D inverse graphics through code generation, in which agents reconstruct dynamic scenes from video as executable graphics programs. To accomplish this, agents must translate visual observations into compact representations of scene structure and dynamics, by implementing abstractions such as physical simulations to reproduce complex behavior. To evaluate this capability, we curate a set of real-world videos and construct synthetic scenes spanning diverse physical phenomena, including deformation, fluid flow, and fracture. We perform extensive benchmark
• 副分类：agent
• 待LLM分类：否
• 标题中文：4DCodeBench：动态场景逆向图形中的 Agent 基准测试
• TLDR中文：本文提出 4DCodeBench，一个通过代码生成进行 4D 逆向图形的基准测试，其中 Agent 将动态场景从视频重建为可执行的图形程序。为此，Agent 必须将视觉观察转化为场景结构与动态的紧凑表示，通过实现物理仿真等抽象以复现复杂行为。为评估这一能力，我们策划了一组真实视频并构建涵盖多种物理现象（包括变形、流体流动与断裂）的合成场景，并对多种方法进行了广泛的基准测试
• 来源文件：
• /inbox/tom/_candidates/2026-10-06-agent-rag-longcontext-candidates.json