World-Shaper: A Unified Framework for 360° Panoramic Editing

Being able to edit panoramic images is crucial for creating realistic 360° visual experiences. However, existing perspective-based image editing methods fail to model the spatial structure of panoramas. Conventional cube-map decompositions attempt to overcome this problem but inevitably break global consistency due to their mismatch with spherical geometry. Motivated by this insight, we reformulate panoramic editing directly in the equirectangular projection (ERP) domain and present World Shaper, a unified geometry-aware framework that supports five distinct editing operations within a single ERP-native representation. To address the latitude-dependent geometric distortion inherent in ERP, we introduce a geometry-aware learning strategy comprising distortion-aware attention modulation (DAAM), which steers cross-attention with latitude-dependent strength at the feature level; layered shape loss (LSL), which enforces per-object geometric supervision at the output level; and progressive curriculum training to internalize panoramic priors. To overcome the scarcity of paired panoramic editing data, we train a dedicated ERP-native controllable generator that synthesizes objects directly in the equirectangular domain under user-defined conditions, enabling scalable paired data construction for learning diverse editing behaviors. Extensive experiments on our new benchmark, PEBench, demonstrate that World Shaper achieves superior geometric consistency, editing fidelity, and text controllability compared to state-of-the-art methods, enabling coherent and flexible 360° visual world creation with unified editing control. Code, models, and data are available at the project page: https://world-shaper-project.github.io/.

Panoramic images are widely used to create immersive 360° experiences, such as virtual tours, games, and VR environments. However, editing these images is difficult because they represent a full sphere on a flat image, which makes objects look increasingly distorted near the top and bottom. Most existing image editing tools are designed for ordinary perspective photos, so they often produce unrealistic shapes or inconsistent edits when applied to panoramas. Some methods split a panorama into cube faces before editing, but this can break the global structure of the scene. In this work, we develop World Shaper, a new framework for editing panoramas directly in their native equirectangular format. Our method learns how object shapes should change at different latitudes of a panorama and supports several common editing tasks, such as adding, removing, replacing, moving, and resizing objects. We also build a way to generate training data for these tasks at scale. Experiments show that World Shaper produces more geometrically consistent and controllable edits, making it easier to create coherent 360° visual worlds.