VSDiffusion: Taming Ill-Posed Shadow Generation via Visibility-Constrained Diffusion

• URL: http://arxiv.org/abs/2603.08020v1
• Date: Mon, 09 Mar 2026 06:52:53 GMT
• Title: VSDiffusion: Taming Ill-Posed Shadow Generation via Visibility-Constrained Diffusion
• Abstract summary: We propose VSDiffusion to generate realistic cast shadows for inserted foreground objects.<n>A visibility-constrained two-stage framework designed to narrow the solution space by incorporating visibility priors.<n> Experiments on widely used public DESOBAv2 dataset demonstrated that our proposed VSDiffusion can generate accurate shadow.
• Score: 10.547624877131122
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Generating realistic cast shadows for inserted foreground objects is a crucial yet challenging problem in image composition, where maintaining geometric consistency of shadow and object in complex scenes remains difficult due to the ill-posed nature of shadow formation. To address this issue, we propose VSDiffusion, a visibility-constrained two-stage framework designed to narrow the solution space by incorporating visibility priors. In Stage I, we predict a coarse shadow mask to localize plausible shadow generated regions. And in Stage II, conditional diffusion is performed guided by lighting and depth cues estimated from the composite to generate accurate shadows. In VSDiffusion, we inject visibility priors through two complementary pathways. First, a visibility control branch with shadow-gated cross attention that provides multi-scale structural guidance. Then, a learned soft prior map that reweights training loss in error-prone regions to enhance geometric correction. Additionally, we also introduce high-frequency guided enhancement module to sharpen boundaries and improve texture interaction with the background. Experiments on widely used public DESOBAv2 dataset demonstrated that our proposed VSDiffusion can generate accurate shadow, and establishes new SOTA results across most evaluation metrics.

Related papers

• Revisiting Shadow Detection from a Vision-Language Perspective [88.46763399969217]
  Shadow Vision--Language (SVL) uses language as an explicit semantic reference to disambiguate shadows from visually similar dark regions.<n>SVL applies local patch-level constraints with text embeddings to improve fine-grained discrimination under challenging appearance conditions.<n>Built on a frozen DINOv3 image encoder, the framework learns only lightweight projection and decoding modules, yielding a parameter-efficient design with less than $1%$ trainable parameters.
  arXiv  Detail & Related papers  (2026-05-12T08:42:30Z)
• deSEO: Physics-Aware Dataset Creation for High-Resolution Satellite Image Shadow Removal [36.90985364531957]
  Shadows cast by terrain and tall structures remain a major obstacle for high-resolution satellite image analysis.<n>Most Earth-observation datasets are designed for shadow detection or 3D modelling rather than removal.<n>We develop a geometry-aware and physics-informed methodology that is the first to derive paired supervision for satellite shadow removal.
  arXiv  Detail & Related papers  (2026-05-05T10:35:59Z)
• Joint Shadow Generation and Relighting via Light-Geometry Interaction Maps [51.82696819319878]
  We propose Light-Geometry Interaction maps, a novel representation that encodes light-aware occlusion from monocular depth.<n>LGI captures essential light-shadow interactions reliably and accurately, computed from off-the-shelf 2.5D depth map predictions.<n>By embedding LGI into a bridge-matching generative backbone, we reduce ambiguity and enforce physically consistent light-shadow reasoning.
  arXiv  Detail & Related papers  (2026-02-25T11:47:26Z)
• Contrast-Prior Enhanced Duality for Mask-Free Shadow Removal [12.417806583744134]
  Existing shadow removal methods often rely on shadow masks, which are challenging to acquire in real-world scenarios.<n> Exploring intrinsic image cues, such as local contrast information, presents a potential alternative for guiding shadow removal in the absence of explicit masks.<n>We propose the Adaptive Gated Dual-Branch Attention (AGBA) mechanism, which filters and re-weighs the contrast prior to effectively disentangle shadow features.
  arXiv  Detail & Related papers  (2025-07-29T16:00:42Z)
• Detail-Preserving Latent Diffusion for Stable Shadow Removal [24.18957090960958]
  We propose a two-stage fine-tuning pipeline to adapt the Stable Diffusion model for stable and efficient shadow removal.<n> Experimental results show that our method outperforms state-of-the-art shadow removal techniques.
  arXiv  Detail & Related papers  (2024-12-23T15:06:46Z)
• Test-Time Intensity Consistency Adaptation for Shadow Detection [35.03354405371279]
  TICA is a novel framework that leverages light-intensity information during test-time adaptation to enhance shadow detection accuracy. A basic encoder-decoder model is initially trained on a labeled dataset for shadow detection. During the testing phase, the network is adjusted for each test sample by enforcing consistent intensity predictions.
  arXiv  Detail & Related papers  (2024-10-10T08:08:32Z)
• Single-Image Shadow Removal Using Deep Learning: A Comprehensive Survey [78.84004293081631]
  The patterns of shadows are arbitrary, varied, and often have highly complex trace structures. The degradation caused by shadows is spatially non-uniform, resulting in inconsistencies in illumination and color between shadow and non-shadow areas. Recent developments in this field are primarily driven by deep learning-based solutions.
  arXiv  Detail & Related papers  (2024-07-11T20:58:38Z)
• Cross-Modal Spherical Aggregation for Weakly Supervised Remote Sensing Shadow Removal [22.4845448174729]
  We propose a weakly supervised shadow removal network with a spherical feature space, dubbed S2-ShadowNet, to explore the best of both worlds for visible and infrared modalities. Specifically, we employ a modal translation (visible-to-infrared) model to learn the cross-domain mapping, thus generating realistic infrared samples. We contribute a large-scale weakly supervised shadow removal benchmark, including 4000 shadow images with corresponding shadow masks.
  arXiv  Detail & Related papers  (2024-06-25T11:14:09Z)
• SIRe-IR: Inverse Rendering for BRDF Reconstruction with Shadow and Illumination Removal in High-Illuminance Scenes [51.50157919750782]
  We present SIRe-IR, an implicit neural rendering inverse approach that decomposes the scene into environment map, albedo, and roughness. By accurately modeling the indirect radiance field, normal, visibility, and direct light simultaneously, we are able to remove both shadows and indirect illumination. Even in the presence of intense illumination, our method recovers high-quality albedo and roughness with no shadow interference.
  arXiv  Detail & Related papers  (2023-10-19T10:44:23Z)
• SDDNet: Style-guided Dual-layer Disentanglement Network for Shadow Detection [85.16141353762445]
  We treat the input shadow image as a composition of a background layer and a shadow layer, and design a Style-guided Dual-layer Disentanglement Network to model these layers independently.<n>Our model effectively minimizes the detrimental effects of background color, yielding superior performance on three public datasets with a real-time inference speed of 32 FPS.
  arXiv  Detail & Related papers  (2023-08-17T12:10:51Z)
• R2D: Learning Shadow Removal to Enhance Fine-Context Shadow Detection [64.10636296274168]
  Current shadow detection methods perform poorly when detecting shadow regions that are small, unclear or have blurry edges. We propose a new method called Restore to Detect (R2D), where a deep neural network is trained for restoration (shadow removal) We show that our proposed method R2D improves the shadow detection performance while being able to detect fine context better compared to the other recent methods.
  arXiv  Detail & Related papers  (2021-09-20T15:09:22Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.