Procedural Content Generation

Follow this topic Ask about this topic

Procedural content generation studies how to create digital content automatically, from playable levels and maps to 3D worlds, characters, materials, and narratives. Approaches include rule- and constraint-based systems, search and evolutionary methods, reinforcement learning, and generative models such as diffusion models and large language models, often combined in agent workflows. These techniques can make content more varied and easier to adapt, while supporting synthetic data and reproducible environments for games, robotics, and agent training. A central challenge is balancing diversity and visual or narrative quality with controllability, playability, structural consistency, and reliable evaluation.

Latest papers

67 papers8 papers in the last four weeks, level with the four weeks before. 0.0% of all new papers.

A Vision Language Model for Generating Procedural Plant Architecture Representations from Simulated Images
Mar 23, 2026Heesup Yun, Isaac Kazuo Uyehara, Ioannis Droutsas +4Procedural Content Generation Plant Phenotyping
Biological and Agricultural Engineering, One Shields Avenue, Davis, 95616, CA, USA · Department of Plant Sciences, One Shields Avenue, Davis, 95616, CA, USA · Wageningen University & Research, P.O. Box 9101, 6700 HB Wageningen, The Netherlands +1
DiscoGen: Procedural Generation of Algorithm Discovery Tasks in Machine Learning
Mar 18, 2026Alexander D. Goldie, Zilin Wang, Adrian Hayler +17Automated Algorithm Discovery Synthetic Benchmark Generation
University of Oxford · University of California, Santa Barbara · University College London +1
Generation of Maximal Snake Polyominoes Using a Deep Neural Network
Mar 12, 2026Benjamin Gauthier, Alain Goupil, Fadel ToureProcedural Content Generation
Université du Québec à Trois-Rivières · Université du Québec à Trois-Rivières, Trois-Rivières, Canada
FabricGen: Microstructure-Aware Woven Fabric Generation
Mar 7, 2026Yingjie Tang, Di Luo, Zixiong Wang +3Procedural Content Generation T2I Diffusion Models
College of Computer Science, Nankai University, Tianjin, China · School of Intelligence Science and Technology, Nanjing University, Suzhou, China
High Dimensional Procedural Content Generation
Feb 21, 2026Kaijie Xu, Clark VerbruggeProcedural Content Generation
McGill University Montreal, Quebec, Canada
LETGAMES: An LLM-Powered Gamified Approach to Cognitive Training for Patients with Cognitive Impairment
Feb 18, 2026Jingwei Shi, Shengyu Tao, Xinxiang Yin +3Procedural Content Generation
Shanghai University of Finance and Economics, China · Northwest Polytechnical University Xi’an, China · Institute of Data Science, National University of Singapore, Singapore +1
DAV-GSWT: Diffusion-Active-View Sampling for Data-Efficient Gaussian Splatting Wang Tiles
Feb 17, 2026Rong Fu, Jiekai Wu, Yang Li +33D Scene Generation 3D Gaussian Splatting
Independent Researcher

Mar 23, 2026J/K move · Enter open · S save

A Vision Language Model for Generating Procedural Plant Architecture Representations from Simulated Images

Heesup Yun, Isaac Kazuo Uyehara, Ioannis Droutsas, Earl Ranario+3

Biological and Agricultural Engineering, One Shields Avenue, Davis, 95616, CA, USA · Department of Plant Sciences, One Shields Avenue, Davis, 95616, CA, USA · Wageningen University & Research, P.O. Box 9101, 6700 HB Wageningen, The Netherlands+1

Procedural Content Generation Plant Phenotyping 3D Reconstruction Vision-Language Models

Open paper AskPDF arXiv

Three-dimensional (3D) procedural plant architecture models have emerged as an important tool for simulation-based studies of plant structure and function, extracting plant architectural parameters from field measurements, and for generating realistic plants in computer graphics. However, measuring the architectural parameters and nested structures for these models at the field scales remains prohibitively labor-intensive. We present a novel algorithm that generates a 3D plant architecture from an image, creating a functional structural plant model that reflects organ-level geometric and topological parameters and provides a more comprehensive representation of the plant's architecture. Instead of using 3D sensors or processing multi-view images with computer vision to obtain the 3D structure of plants, we proposed a method that generates token sequences that encode a procedural definition of plant architecture. This work used only synthetic images for training and testing, with exact architectural parameters known, allowing testing of the hypothesis that organ-level architectural parameters could be extracted from image data using a vision-language model (VLM). A synthetic dataset of cowpea plant images was generated using the Helios 3D plant simulator, with the detailed plant architecture encoded in XML files. We developed a plant architecture tokenizer for the XML file defining plant architecture, converting it into a token sequence that a language model can predict. The model achieved a token F1 score of 0.73 during teacher-forced training. Evaluation of the model was performed through autoregressive generation, achieving a BLEU-4 score of 94.00% and a ROUGE-L score of 0.5182. This led to the conclusion that such plant architecture model generation and parameter extraction were possible from synthetic images; thus, future work will extend the approach to real imagery data.