ProcFunc: Function-Oriented Abstractions for Procedural 3D Generation in Python

2026-04-29

AI Summary

Traditional procedural 3D generation methods suffer from limitations in modularity, composability, and usability, hindering the efficient creation of large-scale, diverse training data. This work proposes a function-oriented procedural generation paradigm, implemented as a Python library built on Blender, which unifies procedural geometry and material generation through composable semantic abstractions. The approach substantially reduces coding errors, enables vision-language models (VLMs) to assist in generating high-quality procedural code, and facilitates a high-detail, high-efficiency indoor scene synthesizer. Experimental results demonstrate that the system excels in generation diversity, runtime efficiency, and scalability, effectively supporting the construction of large-scale synthetic 3D datasets.

Abstract

We introduce ProcFunc, a library for Blender-based procedural 3D generation in Python. ProcFunc provides a library of easy-to-use Python functions, which streamline creating, combining, analyzing, and executing procedural generation code. ProcFunc makes it easy to create large-scale diverse training data, by combinatorial compositions of semantic components. VLMs can use ProcFunc to edit procedural material and geometry code and can create new procedural code with significantly fewer coding errors. Finally, as an example use case, we use ProcFunc to develop a new procedural generator of indoor rooms, which includes a collection of new compositional procedural materials. We demonstrate the detail, runtime efficiency, and diversity of this room generator, as well as its use for 3D synthetic data generation. Please visit https://github.com/princeton-vl/procfunc for source code.

Problem

• procedural generation
• 3D content creation
• function abstraction
• synthetic data generation
• compositional modeling

Innovation

• procedural generation
• function-oriented abstraction
• 3D synthetic data
• compositional materials
• visual language models