Self-Evolving 3D Scene Generation from a Single Image

TL;DR

EvoScene is a self-evolving, training-free framework that progressively reconstructs detailed 3D scenes from a single image by combining geometric reasoning and visual knowledge through iterative 2D-3D domain alternation.

Contribution

It introduces a novel self-evolving, training-free method that iteratively improves 3D scene reconstruction from a single image by integrating existing models in a multi-stage process.

Findings

Achieves superior geometric stability and view-consistent textures.

Effectively completes unseen regions in 3D scenes.

Produces ready-to-use 3D meshes for practical applications.

Abstract

Generating high-quality, textured 3D scenes from a single image remains a fundamental challenge in vision and graphics. Recent image-to-3D generators recover reasonable geometry from single views, but their object-centric training limits generalization to complex, large-scale scenes with faithful structure and texture. We present EvoScene, a self-evolving, training-free framework that progressively reconstructs complete 3D scenes from single images. The key idea is combining the complementary strengths of existing models: geometric reasoning from 3D generation models and visual knowledge from video generation models. Through three iterative stages--Spatial Prior Initialization, Visual-guided 3D Scene Mesh Generation, and Spatial-guided Novel View Generation--EvoScene alternates between 2D and 3D domains, gradually improving both structure and appearance. Experiments on diverse scenes demonstrate that EvoScene achieves superior geometric stability, view-consistent textures, and unseen-region completion compared to strong baselines, producing ready-to-use 3D meshes for practical applications.