Towards Scalable and Consistent 3D Editing

3D editing—the task of locally modifying the geometry or appearance of a 3D asset—has wide applications in immersive content creation, digital entertainment, and AR/VR. However, unlike 2D editing, it remains challenging due to the need for cross-view consistency, structural fidelity, and fine-grained controllability. Existing approaches are often slow, prone to geometric distortions, or dependent on manual and accurate 3D masks that are error-prone and impractical. To address these challenges, we advance both the data and model fronts. On the data side, we introduce 3DEditVerse, the largest paired 3D editing benchmark to date, comprising 116,309 high-quality training pairs and 1,500 curated test pairs. Built through complementary pipelines of pose-driven geometric edits and foundation model-guided appearance edits, 3DEditVerse ensures edit locality, multi-view consistency, and semantic alignment. On the model side, we propose 3DEditFormer, a 3D-structure-preserving transformer. By enhancing image-to-3D generation with dual-guidance attention and time-adaptive gating, 3DEditFormer disentangles editable regions from preserved structure, enabling precise and consistent edits without requiring auxiliary 3D masks. Extensive experiments demonstrate that our framework outperforms state-of-the-art baselines both quantitatively and qualitatively, establishing a new standard for practical and scalable 3D editing. Dataset and code are available at https://www.lv-lab.org/3DEditFormer/

Modifying 3D digital objects—like changing an animated character's pose or adding an accessory to a model—is incredibly challenging compared to editing flat 2D images. Current tools are often slow, distort shapes, or force users to manually draw complex 3D masks, making the process tedious and impractical. Furthermore, researchers have historically lacked large-scale datasets to train AI models to perform these tasks effectively. To solve this, we created 3DEditVerse, the largest dataset of before-and-after 3D objects to date, featuring over 116,000 high-quality pairs to teach AI how to edit properly. We also developed a new AI model called 3DEditFormer that can accurately modify a 3D asset using only a single edited 2D image as a guide. Our model intelligently separates the specific regions that need to change from the structure that must remain untouched, completely eliminating the need for manual masks. This approach sets a new standard, significantly outperforming existing systems in both quality and consistency. By removing these technical bottlenecks, our work makes 3D asset modification faster and more accessible, bringing us closer to a world where creators can edit virtual reality, gaming, and digital art assets as easily as editing a photo on a smartphone.