MetaShadow: Object-Centered Shadow Detection, Removal, and Synthesis
Tianyu Wang, Jianming Zhang, Haitian Zheng, Zhihong Ding, Scott Cohen, Zhe Lin, Wei Xiong, Chi-Wing Fu, Luis Figueroa, Soo Ye Kim
Abstract
Figure 1. MetaShadow is a versatile three-in-one framework designed for shadow-related tasks, enabling shadow manipulation in various object-centered image editing operations such as: [I] Object Relocation: Our model can detect and remove the shadow of an existing object, then synthesize the shadow in the new location consistent with the original shadow. [II] Remove an object and its shadow: (1) Based on the mask of the unwanted object, our model can directly remove its shadow (2). After removing the object (3), we can eliminate any remaining shadows for a cleaner background (4) if we do not specify which shadow to remove. [III] Insert an object and synthesize its shadow: When inserting the person in (b) to another image (a) with similar lighting, our model can generate a realistic shadow, enhancing the final compositing quality.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Your agent calls
Luneget_paper_fulltext
Free to start. No credit card required.
Terminal
Install the CLIlune papers fulltext 2e38b58c-37db-4d26-a03c-53071e580ff5Cited by top-tier papers4
- MV-CoLight: Efficient Object Compositing with Consistent Lighting and Shadow GenerationKerui Ren, Jiayang Bai, Linning Xu, Lihan Jiang et al.NeurIPS 2025 · 9 citations
- Omnipaint: Mastering Object-Oriented Editing Via Disentangled Insertion-Removal InpaintingYongsheng Yu, Ziyun Zeng, Haitian Zheng, Jiebo LuoICCV 2025 · 3 citations
- You Only Erase Once: Erasing Anything without Bringing Unexpected ContentYixing Zhu, Qing Zhang, Wenju Xu, Wei-Shi ZhengCVPR 2026
- Generative Image Layer Decomposition with Visual EffectsJinrui Yang, Qing Liu, Yijun Li, Soo Ye Kim et al.CVPR 2025
Builds on33
- Learning Transferable Visual Models From Natural Language SupervisionAlec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh et al.ICML 2021 · 47,906 citations
- Denoising Diffusion Probabilistic ModelsJonathan Ho, Ajay Jain, Pieter AbbeelNeurIPS 2020 · 35,902 citations
- High-Resolution Image Synthesis with Latent Diffusion ModelsRobin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser et al.CVPR 2022 · 13,123 citations
- Adding Conditional Control to Text-to-Image Diffusion ModelsLvmin Zhang, Anyi Rao, Maneesh AgrawalaICCV 2023 · 6,759 citations
- Towards Ghost-Free Shadow Removal via Dual Hierarchical Aggregation Network and Shadow Matting GANXiaodong Cun, Chi-Man Pun, Cheng ShiAAAI 2020 · 272 citations
Related papers
- ShadowMover: Automatically Projecting Real Shadows onto Virtual ObjectPiaopiao Yu, Jie Guo, Fan Huang, Zhenyu Chen et al.IEEE VR 2023 · 6 citations
- No Shadow Left Behind: Removing Objects and Their Shadows Using Approximate Lighting and GeometryEdward Zhang, Ricardo Martin-Brualla, Janne Kontkanen, Brian L. CurlessCVPR 2021
- Shadow Generation for Composite Image in Real-World ScenesYan Hong, Li Niu, Jianfu ZhangAAAI 2022 · 55 citations
- Shadow Generation with Decomposed Mask Prediction and Attentive Shadow FillingXinhao Tao, Junyan Cao, Yan Hong, Li NiuAAAI 2024
- ObjectMover: Generative Object Movement with Video PriorXin Yu, Tianyu Wang, Soo Ye Kim, Paul Guerrero et al.CVPR 2025
