Newest papers

A Unified Framework for Image-to-3D Part Generation via Variable Granularity

Qiwen Gu, Xin Zhou, Xiwu Chen, Zhongyang Zhu, Zhuo Zhang, Junqiao Zhao · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31671 · cited by 0 works (OpenAlex) · open-access PDF

Spanish

La generación de piezas 3D controlable es crucial para la creación de activos digitales, pero los métodos existentes carecen de la flexibilidad para acomodar los diversos flujos de trabajo de los usuarios que requieren diferentes niveles de control. En este artículo, presentamos FlexPart, un marco de generación de piezas unificado de imagen a 3D que soporta las condiciones geométricas de granularidad variable dentro de un único modelo. Al integrar sin problemas las entradas heterogéneas, desde puntos escasos hasta cajas y máscaras densas, a través de la modulación adaptativa por puertas, FlexPart desbloquea una poderosa sinergia transmodal. También proponemos un mecanismo de orientación geométrica asimétrica que aprovecha los fuertes prioridades espaciales para alinear las granularidades representativas, mejorando

Machine translation (Meta M2M100); the English original follows.

Abstract

Controllable 3D part generation is pivotal for digital asset creation, yet existing methods lack the flexibility to accommodate diverse user workflows requiring varying levels of control. In this paper, we introduce FlexPart, a unified image-to-3D part generation framework that supports variable-granularity geometric conditions within a single model. By seamlessly integrating heterogeneous inputs, ranging from sparse points to dense bounding boxes and masks, via gated adaptive modulation, FlexPart unlocks a powerful cross-modal synergy. We also propose an asymmetric geometric guidance mechanism that leverages strong spatial priors to align representations across granularities, improving geometric consistency and facilitating inter-task synergy. Under mask conditioning on PartVerseXL, FlexPart reduces Part CD by 16.6% relative to TripoSG + SAM3D and by 10.5% relative to OmniPart, while maintaining competitive whole-mesh quality. The code can be found at https://github.com/Gucci233/FlexPart

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Granularity, Consistency (knowledge bases), Ranging, Workflow, Bounding overwatch

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange