PulseAugur
EN
LIVE 06:32:20

New diffusion models tackle in-the-wild video shadow removal

Researchers have developed two new methods for removing shadows from videos using diffusion models. WildShadowRemover adapts a pre-trained video diffusion model with a detail injection module and a frequency-decomposed modulation module to preserve fine details and suppress shadow artifacts. FreeShadow, a training-free approach, utilizes illumination transfer attention to recover illumination and selectively preserves illumination-invariant components for content fidelity, further enhanced by local texture-preserving relighting. Both methods aim to improve shadow removal quality, temporal consistency, and generalization across diverse real-world scenarios. AI

IMPACT These methods could improve video editing and visual effects by enabling more realistic and automated shadow removal in diverse real-world footage.

RANK_REASON Two research papers published on arXiv detailing new methods for video shadow removal using diffusion models.

Read on arXiv cs.CV →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New diffusion models tackle in-the-wild video shadow removal

COVERAGE [2]

  1. arXiv cs.CV TIER_1 English(EN) · Jiamin Xu, Cong Wang, Zheng Dong, Chi Wang, Renshu Gu, Weiwei Xu, Gang Xu ·

    WildShadowRemover: In-the-Wild Video Shadow Removal via Detail-Preserving Video Diffusion Models

    arXiv:2607.26203v1 Announce Type: new Abstract: Video shadow removal in the wild remains challenging due to complex illumination, diverse shadow appearances, and limited training data. Despite its importance to numerous vision and graphics applications, it remains largely unexplo…

  2. arXiv cs.CV TIER_1 English(EN) · Yinan Wang, Yan Huang, Yong Xu, Patrick Le Callet ·

    FreeShadow: Training-Free Shadow Removal via Illumination Transfer and Selective Content Preservation in Diffusion Models

    arXiv:2607.26715v1 Announce Type: new Abstract: Existing supervised and unsupervised shadow removal methods often suffer from limited generalization due to the insufficient diversity of available training datasets, while zero-shot methods tend to produce artifacts and require tim…