Controllable 3D Outdoor Scene Generation via Scene Graphs

[ICCV 2025] Yuheng Liu^1,2, Xinke Li³, Yuning Zhang⁴, Lu Qi⁵, Xin Li¹, Wenping Wang¹, Chongshou Li⁴, Xueting Li^6*, Ming-Hsuan Yang^2*

¹Texas A&M University, ²The University of Cailfornia, Merced, ³City University of HongKong, ⁴Southwest Jiaotong University, ⁵Insta360 Research, ⁶NVIDIA

Three-dimensional scene generation is crucial in computer vision, with applications spanning autonomous driving, gaming and the metaverse. Current methods either lack user control or rely on imprecise, non-intuitive conditions. In this work, we propose a method that uses scene graphs—an accessible, user-friendly control format—to generate outdoor 3D scenes. We develop an interactive system that transforms a sparse scene graph into a dense BEV (Bird's Eye View) Embedding Layout, which guides a conditional diffusion model to generate 3D scenes that match the scene graph description. During inference, users can easily create or modify scene graphs to generate large-scale outdoor scenes. We create a large-scale dataset with paired scene graphs and 3D semantic scenes to train the BEV embedding and diffusion models. Experimental results show that our approach consistently produces high-quality 3D urban scenes closely aligned with the input scene graphs.

NEWS

• [2025/06/25] Our work is accepted by ICCV 2025.
• [2025/03/10] Our work is now on arXiv.
• [2024/11/15] Official repo is created, code will be released soon.