ControlVP: Interactive Geometric Refinement of AI-Generated Images with Consistent Vanishing Points

TL;DR

ControlVP is a user-guided framework that corrects vanishing point inconsistencies in AI-generated images, improving geometric realism especially in architectural scenes, while preserving visual quality.

Contribution

It introduces a structural guidance method with geometric constraints to enhance spatial consistency in diffusion-based image generation.

Findings

Significantly improves geometric consistency in generated images.

Maintains visual fidelity comparable to baseline models.

Applicable to architectural and spatially complex scenes.

Abstract

Recent text-to-image models, such as Stable Diffusion, have achieved impressive visual quality, yet they often suffer from geometric inconsistencies that undermine the structural realism of generated scenes. One prominent issue is vanishing point inconsistency, where projections of parallel lines fail to converge correctly in 2D space. This leads to structurally implausible geometry that degrades spatial realism, especially in architectural scenes. We propose ControlVP, a user-guided framework for correcting vanishing point inconsistencies in generated images. Our approach extends a pre-trained diffusion model by incorporating structural guidance derived from building contours. We also introduce geometric constraints that explicitly encourage alignment between image edges and perspective cues. Our method enhances global geometric consistency while maintaining visual fidelity comparable to the baselines. This capability is particularly valuable for applications that require accurate spatial structure, such as image-to-3D reconstruction. The dataset and source code are available at https://github.com/RyotaOkumura/ControlVP .