New 3D scene generation framework GuidedSceneGen maintains absolute scale

Researchers have introduced GuidedSceneGen, a novel text-to-3D scene generation framework designed to overcome scale ambiguity and geometric drift common in previous methods. This system maintains an absolute world coordinate frame throughout the generation process, starting with a textual description to predict a global 3D layout. A diffusion model then synthesizes 360° imagery aligned with this layout, and a video diffusion model aids in exploring unobserved regions efficiently. The generated views are fused using 3D Gaussian Splatting to create a navigable 3D scene at an accurate scale, with results demonstrating improved spatial coherence and layout plausibility. AI

IMPACT This framework could enable more accurate and interpretable 3D scene creation from text, impacting fields like virtual reality and architectural design.

RANK_REASON The cluster contains a research paper detailing a new method for 3D scene generation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• GuidedSceneGen
• Hugging Face
• ScienceCast
• Stefan Ainetter

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →