Newest papers

InfiniDream: Open-Ended Photorealistic 3D Worlds Through Procedural Generation

Yiyao Huang, Hao-Jen Chien, Zheng-Hui Huang, Yu-Lun Liu, Kaipeng Zhang, Zhixiang Wang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26853 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

La generación de código basada en VLM desempeña un papel clave en la creación de marcos de datos sintéticos y el desarrollo de juegos, ofreciendo una mayor controlabilidad, edición y diversidad que las representaciones de mesas estáticas. Sin embargo, la generación de código basada en VLM a menudo produce activos 3D relativamente simples, mientras que los sofisticados sistemas procedimentales fotorreales requieren un amplio desarrollo manual. Los métodos procedimentales guiados por VLM existentes siguen siendo en gran medida restringidos por las fábricas predefinidas y las APIs, limitando su capacidad para crear nuevos tipos de activos. Introducimos InfiniDream, un marco basado en VLM para la generación mundial de procedimientos 3D abiertos. En primer lugar, proponemos la generación de activos simbólicamente restringidos, que

Machine translation (Meta M2M100); the English original follows.

Abstract

Procedural generation plays a key role in synthetic data creation and game development, offering greater controllability, editability, and diversity than static mesh representations. However, VLM-based code generation often produces relatively simple 3D assets, while sophisticated photorealistic procedural systems require extensive manual development. Existing VLM-guided procedural methods remain largely constrained by predefined factories and APIs, limiting their ability to create novel asset types. We introduce InfiniDream, a VLM-driven framework for open-ended procedural 3D world generation. First, we propose Symbolically Constrained Asset Generation, which leverages existing procedural algorithms to create novel, photorealistic assets while preserving editability and seed-based diversity. Second, we develop an end-to-end text-to-world pipeline that combines semantic scene planning with Deterministic Frustum-Guided Grid Placement to construct large-scale 3D worlds with precise object placement, procedural population, and animation. Extensive evaluations on 100 object and 20 scene generation tasks demonstrate improvements across four perceptual metrics. Diversity analysis, ablation studies, and human preference evaluations further validate the effectiveness of our framework. Project page: https://yichuanh.github.io/infinidream-page/

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Procedural modeling, Grid, Object (grammar), Human–computer interaction, Key (lock)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange