Recursive code world models build complex 3d scenes from images
Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs
Computer Vision and Pattern Recognition
Summary
Making detailed 3D worlds from a single picture is hard because scenes have many parts connected in complex ways. The authors introduce Recursive Code World Models, a method that breaks down a scene into smaller parts using a program that calls itself to build and refine the scene step by step. This approach helps create more accurate 3D worlds by repeatedly checking and fixing details while keeping the whole scene consistent. Their method improves over previous ones that turn images into 3D scenes using code.
What this means in practice
- •For 3d environment designers: generate complex editable 3D scenes automatically from one reference image for use in simulations or games
- •For augmented reality developers: create accurate 3D reconstructions of real-world environments to improve AR scene integration and interaction
Authors
Zhiqi Li, Yuxuan Liao, Bo Zhu
Abstract
Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recursive Scene Program (RSP) representation with a construction solver that recursively calls itself. An RSP represents the executable world as compositional scene code, while each solver call follows the same complete process: establish the whole, recursively reconstruct unresolved parts, and revisit the whole to refine their composition. This global-local-global recursion gives fine-scale structures their own perception-and-editing loops while preserving scene-wide geometry and relationships. Reference-aligned views propagate a shared camera projection across levels, while parent revisitation addresses boundaries, spatial relations, and shared errors that emerge after local refinement. A vision-language coding agent directly compares reference images with scene renders to guide refinement, recursive descent, and return. Across complex scenes, RCWM outperforms prior code-based image-to-scene reconstruction methods. Ablation studies further support the benefits of recursive construction and suggest that deeper calls can improve finer-scale reconstruction. RCWM provides a recursive construction principle for building complex executable worlds from visual evidence.