Procedura: Agentic 3D Modeling with Procedural Control
Procedura: Agentic 3D Modeling with Procedural Control
Procedura:基于程序化控制的智能体 3D 建模
Abstract: Native 3D generators now recover impressive mesh geometry from a single image. However, a dense mesh stays soft where a machined object should be sharp, it carries no part decomposition, and it exposes no parameter a user could edit. To address this, we explore the paradigm of 3D shape as code, leveraging and scaling the coding ability of an LLM for 3D modeling.
摘要: 原生 3D 生成器目前已经能够从单张图像中恢复出令人印象深刻的网格几何结构。然而,对于机械加工物体而言,这些致密的网格在原本应锐利的地方显得过于圆润,且不具备部件分解功能,也不提供任何可供用户编辑的参数。为了解决这一问题,我们探索了“将 3D 形状视为代码”的范式,利用并扩展大语言模型(LLM)的编码能力来进行 3D 建模。
We introduce Procedura, a novel 3D modeling agent framework that writes an object as a procedural assembly, a parametric program whose named parts are joined by typed, machine-checkable mates. From a text prompt, the agent plans the object as an assembly graph and writes the program part by part, solving each placement from the mated frames rather than guessing it, and admitting a part only once compile, mate, and connectivity checks pass.
我们引入了 Procedura,这是一个新颖的 3D 建模智能体框架。它将物体编写为程序化装配体——即一种参数化程序,其命名的部件通过类型化、可由机器校验的配合关系(mates)连接。根据文本提示,智能体将物体规划为装配图,并逐个部件编写程序;它通过配合框架(mated frames)来求解每个部件的位置,而非进行猜测,并且只有在通过编译、配合和连通性检查后,才会确认该部件。
A decoupled vision critic then refines the assembly one diagnosed fix at a time. Moreover, the same graph carries per-part materials and a simulator-validated articulation. We evaluate on P3D-Bench under its assembly judge, and with the same judge on MechBench-36, our hard-surface benchmark.
随后,一个解耦的视觉评估器(vision critic)会针对诊断出的问题对装配体进行逐一修正。此外,同一图结构还包含了各部件的材质信息以及经模拟器验证的关节运动。我们在 P3D-Bench 的装配评估标准下进行了测试,并在我们自己的硬表面基准测试集 MechBench-36 上使用了相同的评估标准。
On both, Procedura outperforms state-of-the-art native 3D generators and every prior 3D-code agent on judged quality, produces the sharpest edges of any method we evaluate, and is the only one whose output is an editable, part-structured program.
在两项测试中,Procedura 在评估质量上均优于最先进的原生 3D 生成器及所有先前的 3D 代码智能体;它生成的边缘比我们评估过的任何方法都要锐利,并且是唯一能够输出可编辑、具有部件结构程序的模型。