SceneDreamer: unbounded 3D scene generation from 2D image collections

• المؤلفون الرئيسيون: Chen, Zhaoxi, Wang, Guangcong, Liu, Ziwei
• مؤلفون آخرون: School of Computer Science and Engineering
• التنسيق: مقال
• اللغة: English
• منشور في: 2024
• الموضوعات: Computer and Information Science; Neural Rendering; Unbounded Scene Generation
• الوصول للمادة أونلاين: https://hdl.handle.net/10356/173443
• المؤسسة: Nanyang Technological University

In this work, we present SceneDreamer, an unconditional generative model for unbounded 3D scenes, which synthesizes large-scale 3D landscapes from random noise. Our framework is learned from in-the-wild 2D image collections only, without any 3D annotations. At the core of SceneDreamer is a principle...