SpaceFlow: Locally Controllable 3D Generation

2026-10-08

AI Summary

This study addresses the lack of local control in existing 3D generation methods, where geometry and appearance cannot be independently specified across regions. We propose a training-free, locally controllable 3D generation framework that introduces a novel part-proxy mechanism based on geometric primitives. By assigning differentiated control levels, this approach achieves a flexible balance between strict geometric constraints and generative completion. Furthermore, region-wise appearance routing is accomplished via segmentation matching, effectively suppressing cross-part feature leakage. Experiments demonstrate that highly controlled regions preserve geometric fidelity while low-control regions permit shape variation. Under fixed geometry, both prompt faithfulness and material accuracy achieve state-of-the-art performance. User studies further confirm that the overall generation quality is highly competitive.

Abstract

Current 3D generation methods lack explicit local control: geometric adherence is often defined by a global control strength, and appearance cannot be specified locally. We present SpaceFlow, a training-free pipeline for locally controllable 3D generation from text descriptions and a collection of geometric primitives. Each primitive serves as a proxy for an object part and is assigned a local control level, enabling users to specify whether regions should strictly follow the input shape or allow generative completion. During structure generation, we enforce these spatial constraints within the generative flow process. For appearance synthesis, the generated structure is segmented and matched to the primitives. Each generated part is conditioned only on its assigned text or image cue, thereby limiting cross-part leakage. Regional geometry metrics demonstrate that SpaceFlow preserves the specified geometry in high-control regions and enables plausible shape variation in low-control areas. A user study further indicates that the resulting balance between geometric fidelity and generative freedom remains competitive in overall quality. When evaluating appearance on fixed geometry, text-conditioned routing achieves state-of-the-art prompt faithfulness and color/material accuracy. Qualitative results additionally show localized routing of image cues. The project page is available at SpaceFlow3D.github.io.

Problem

• 3D generation
• local controllability
• geometric adherence
• appearance synthesis
• cross-part leakage

Innovation

• Locally Controllable 3D Generation
• Training-free Pipeline
• Geometric Primitives
• Generative Flow
• Cross-part Leakage Prevention