Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

Procedura: AI Agent That Writes 3D Models as Editable Code

AI By Crimson AI Hugging Face Papers 28 August 2026 · 00:00 33 views
Share: X Telegram

Hugging Face researchers introduce Procedura, an agentic 3D modeling framework that generates editable, part-structured procedural assemblies from text prompts, outperforming native generators and prior code-based agents on judged quality.

Procedura: AI Agent That Writes 3D Models as Editable Code

Key points

Hugging Face researchers have unveiled Procedura, a novel 3D modeling agent that transforms text prompts into editable, parametric 3D programs. Unlike traditional native 3D generators that produce dense meshes, Procedura writes objects as procedural assemblies—code with named parts joined by typed, machine-checkable mates.

The framework leverages the coding ability of large language models (LLMs) to treat 3D shape as code. From a text prompt, the agent plans an assembly graph and writes the program part by part, solving each placement from mated frames rather than guessing. Parts are admitted only after passing compile, mate, and connectivity checks, ensuring geometric validity.

A decoupled vision critic refines the assembly iteratively, diagnosing and fixing issues one at a time. The resulting graph includes per-part materials and simulator-validated articulation, making the output fully editable and part-structured—a key advantage over dense mesh outputs.

In evaluations on P3D-Bench and the new MechBench-36 hard-surface benchmark, Procedura outperformed state-of-the-art native 3D generators and all prior 3D-code agents on judged quality. It also produced the sharpest edges among all methods tested, and was the only one to deliver an editable, part-structured program.

Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 15 Oct 2026

New Survey Maps How Multimodal AI Agents Remember and Evolve

A Hugging Face survey proposes a two-axis framework for multimodal agent memory, separating how information is formed and represen...

9
Research paper
Hugging Face 15 Oct 2026

CriticOPD Fixes On-Policy Distillation by Letting Teachers Critique Student Errors

Hugging Face researchers introduce CriticOPD, a method that restores the teacher's ability to recover from student mistakes during...

9
Research paper
Hugging Face 15 Oct 2026

ActionOPD: Distilling Agent Decisions, Not Tokens

A new Hugging Face paper proposes ActionOPD, an action-level on-policy distillation method that aligns a student agent's decisions...

10