Text-Driven 3D Indoor Scene Synthesis in Non-Manhattan Environments

2026-07-02

AI Summary

Existing methods struggle to model non-orthogonal spatial relationships in non-Manhattan indoor environments, often resulting in layout geometries with high geometric conflicts and insufficient physical plausibility. To address this limitation, this work proposes SPG-Layout, a novel framework that integrates statistical priors of object distributions with a hierarchical layout strategy. Guided by a large language model, the approach prioritizes the placement of large-scale objects first, effectively emulating human-like design reasoning. Evaluated on a newly introduced benchmark comprising 500 non-Manhattan scenes, SPG-Layout significantly outperforms current state-of-the-art methods, achieving superior semantic fidelity and physical plausibility in both Manhattan and non-Manhattan settings.

Abstract

Large Language Models (LLMs) have demonstrated remarkable capabilities in 3D indoor synthesis for Manhattan environments. However, existing methods often fail to capture plausible object layout patterns in non-Manhattan settings, primarily because they struggle to model non-orthogonal spatial relationships, leading to high geometric violations and low physical fidelity. To address this challenge, we propose SPG-Layout, a novel text-driven framework designed to generate physically plausible indoor scenes within complex non-Manhattan environments. Specifically, we first utilize statistical priors of object distributions to guide the training process, enhancing environmental understanding and fidelity. Furthermore, mirroring human design workflows, we adopt a hierarchical layout strategy that prioritizes the placement of large objects, thereby substantially minimizing layout violations. By synergizing these components, SPG-Layout achieves a balanced optimization of semantic realism and physical plausibility. To evaluate performance in these complex settings, we constructed a new benchmark comprising 500 diverse non-Manhattan environments. Extensive experiments demonstrate that SPG-Layout consistently and significantly outperforms existing methods across both Manhattan and non-Manhattan environments. The code will be publicly released.

Problem

• non-Manhattan environments
• 3D indoor scene synthesis
• object layout
• spatial relationships
• physical plausibility

Innovation

• non-Manhattan environments
• text-driven 3D synthesis
• hierarchical layout
• statistical priors
• physical plausibility