SCULPT: Subtractive Composition for 3D Part Generation
2026-08-13 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionGraphics
AI summaryⓘ
The authors present SCULPT, a method for generating 3D objects that are made of parts, allowing easier editing and use. Unlike past methods that either split finished shapes or build parts separately and then combine them, SCULPT extracts parts by jointly removing pieces from a whole object in a way that keeps their boundaries clean and consistent. It uses a special predictor that generates an extracted part and the remaining object at the same time, avoiding gaps or overlaps between parts. The method adapts the number of parts per object automatically and shows strong performance in creating detailed, textured 3D parts from various inputs.
3D generationpart decompositionlatent spacedenoising predictorsubtractive compositionvoxel supportobject segmentation3D reconstructiontextured 3D parts3D shape editing
Authors
Sikuang Li, Chen Yang, Jiemin Fang, Jiazhong Cen, Yuhe Wei, Jichen Pang, Wei Shen, Qi Tian
Abstract
Part-aware 3D generation aims to create digital assets that are coherent as complete objects while exposing structural parts for editing, material assignment, animation, and reuse. Existing methods impose this structure outside the native generation loop: segmentation-based methods partition an already generated shape, while additive methods synthesize parts from predefined layouts, boxes, or tokens and then reconcile them into a whole. The former preserves the generated geometry but fixes the object before part boundaries are determined; the latter exposes part cardinality but often leaves shared boundaries vulnerable to gaps, interpenetrations, and material discontinuities. In this paper, we propose SCULPT, a framework that addresses these challenges through subtractive composition. Given a complete object represented in a structured 3D latent space, SCULPT iteratively applies a joint split predictor to generate one extracted part together with the remaining object. The predictor performs a coupled denoising process conditioned on both the image and the current 3D state, so the extracted part and updated remainder are generated together rather than reconciled after generation. The joint split predictor processes both outputs on the union of their native sparse 3D supports, allowing neighboring supports to overlap rather than imposing a disjoint voxel partition. The rollout ends when the remainder support becomes empty or reaches a fixed safety cap, allowing the number of generated parts to adapt to each object within that bound. Extensive experiments demonstrate state-of-the-art geometry on PartObjaverse while preserving strong complete-object reconstruction after part assembly. Results on four dataset images, one text-to-image-generated input, and one real-world photograph further show fine-grained textured part decomposition beyond the benchmark.