SpatialEdit: Benchmarking Fine-Grained Image Spatial Editing

2026-04-06

AI Summary

Existing image editing methods struggle to achieve fine-grained spatial control and lack a unified evaluation framework that balances geometric fidelity with perceptual plausibility. To address this gap, this work proposes SpatialEdit-Bench—the first benchmark for jointly assessing geometric accuracy and perceptual quality in spatial image editing—and introduces SpatialEdit-500k, a large-scale synthetic dataset comprising precise geometric annotations generated through a controllable Blender-based rendering pipeline. Building upon this foundation, the authors develop SpatialEdit-16B, a dedicated baseline model capable of accurately manipulating object poses and camera viewpoints. Experimental results demonstrate that the model excels in general editing tasks and significantly outperforms existing approaches in fine-grained spatial manipulation.

Abstract

Image spatial editing performs geometry-driven transformations, allowing precise control over object layout and camera viewpoints. Current models are insufficient for fine-grained spatial manipulations, motivating a dedicated assessment suite. Our contributions are listed: (i) We introduce SpatialEdit-Bench, a complete benchmark that evaluates spatial editing by jointly measuring perceptual plausibility and geometric fidelity via viewpoint reconstruction and framing analysis. (ii) To address the data bottleneck for scalable training, we construct SpatialEdit-500k, a synthetic dataset generated with a controllable Blender pipeline that renders objects across diverse backgrounds and systematic camera trajectories, providing precise ground-truth transformations for both object- and camera-centric operations. (iii) Building on this data, we develop SpatialEdit-16B, a baseline model for fine-grained spatial editing. Our method achieves competitive performance on general editing while substantially outperforming prior methods on spatial manipulation tasks. All resources will be made public at https://github.com/EasonXiao-888/SpatialEdit.

Problem

• spatial editing
• fine-grained manipulation
• geometric fidelity
• image editing
• benchmarking

Innovation

• spatial editing
• fine-grained manipulation
• synthetic dataset
• geometric fidelity
• benchmark