Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2507.13861 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909993794535424 |
|---|---|
| author | Hu, Junjie Han, Tianyang Ma, Kai Gao, Jialin Yang, Song He, Xianhua Luo, Junfeng Wei, Xiaoming Zhang, Wenqiang |
| author_facet | Hu, Junjie Han, Tianyang Ma, Kai Gao, Jialin Yang, Song He, Xianhua Luo, Junfeng Wei, Xiaoming Zhang, Wenqiang |
| contents | Recent subject-driven image customization excels in fidelity, yet fine-grained instance-level spatial control remains an elusive challenge, hindering real-world applications. This limitation stems from two factors: a scarcity of scalable, position-annotated datasets, and the entanglement of identity and layout by global attention mechanisms. To this end, we introduce PositionIC, a unified framework for high-fidelity, spatially controllable multi-subject customization. First, we present BMPDS, the first automatic data-synthesis pipeline for position-annotated multi-subject datasets, effectively providing crucial spatial supervision. Second, we design a lightweight, layout-aware diffusion framework that integrates a novel visibility-aware attention mechanism. This mechanism explicitly models spatial relationships via an NeRF-inspired volumetric weight regulation to effectively decouple instance-level spatial embeddings from semantic identity features, enabling precise, occlusion-aware placement of multiple subjects. Extensive experiments demonstrate PositionIC achieves state-of-the-art performance on public benchmarks, setting new records for spatial precision and identity consistency. Our work represents a significant step towards truly controllable, high-fidelity image customization in multi-entity scenarios. Code and data: https://github.com/MeiGen-AI/PositionIC. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_13861 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | PositionIC: Unified Position and Identity Consistency for Image Customization Hu, Junjie Han, Tianyang Ma, Kai Gao, Jialin Yang, Song He, Xianhua Luo, Junfeng Wei, Xiaoming Zhang, Wenqiang Computer Vision and Pattern Recognition Recent subject-driven image customization excels in fidelity, yet fine-grained instance-level spatial control remains an elusive challenge, hindering real-world applications. This limitation stems from two factors: a scarcity of scalable, position-annotated datasets, and the entanglement of identity and layout by global attention mechanisms. To this end, we introduce PositionIC, a unified framework for high-fidelity, spatially controllable multi-subject customization. First, we present BMPDS, the first automatic data-synthesis pipeline for position-annotated multi-subject datasets, effectively providing crucial spatial supervision. Second, we design a lightweight, layout-aware diffusion framework that integrates a novel visibility-aware attention mechanism. This mechanism explicitly models spatial relationships via an NeRF-inspired volumetric weight regulation to effectively decouple instance-level spatial embeddings from semantic identity features, enabling precise, occlusion-aware placement of multiple subjects. Extensive experiments demonstrate PositionIC achieves state-of-the-art performance on public benchmarks, setting new records for spatial precision and identity consistency. Our work represents a significant step towards truly controllable, high-fidelity image customization in multi-entity scenarios. Code and data: https://github.com/MeiGen-AI/PositionIC. |
| title | PositionIC: Unified Position and Identity Consistency for Image Customization |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.13861 |