The first in-the-wild dataset for evaluating single-image relighting models</p>\n","updatedAt":"2026-05-13T08:43:42.835Z","author":{"_id":"63be9021da08ed0544f36c38","avatarUrl":"/avatars/42511d8b1d3a3ef99bc154c98b72dfba.svg","fullname":"onurcan","name":"monurcan","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.7642694115638733},"editors":["monurcan"],"editorAvatarUrls":["/avatars/42511d8b1d3a3ef99bc154c98b72dfba.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2605.11696","authors":[{"_id":"6a042ecb86b054ce2fa4105f","name":"Lezhong Wang","hidden":false},{"_id":"6a042ecb86b054ce2fa41060","name":"Mehmet Onurcan Kaya","hidden":false},{"_id":"6a042ecb86b054ce2fa41061","name":"Siavash Bigdeli","hidden":false},{"_id":"6a042ecb86b054ce2fa41062","name":"Jeppe Revall Frisvad","hidden":false}],"publishedAt":"2026-05-12T00:00:00.000Z","submittedOnDailyAt":"2026-05-13T00:00:00.000Z","title":"WildRelight: A Real-World Benchmark and Physics-Guided Adaptation for Single-Image Relighting","submittedOnDailyBy":{"_id":"63be9021da08ed0544f36c38","avatarUrl":"/avatars/42511d8b1d3a3ef99bc154c98b72dfba.svg","isPro":false,"fullname":"onurcan","user":"monurcan","type":"user","name":"monurcan"},"summary":"Recent single-image relighting methods, powered by advanced generative models, have achieved impressive photorealism on synthetic benchmarks. However, their effectiveness in the complex visual landscape of the real world remains largely unverified. A critical gap exists, as current datasets are typically designed for multi-view reconstruction and fail to address the unique challenges of single-image relighting. To bridge this synthetic-to-real gap, we introduce WildRelight, the first in-the-wild dataset specifically created for evaluating single-image relighting models. WildRelight features a diverse collection of high-resolution outdoor scenes, captured under strictly aligned, temporally varying natural illuminations, each paired with a high-dynamic-range environment map. Using this data, we establish a rigorous benchmark revealing that state-of-the-art models trained on synthetic data suffer from severe domain shifts. The strictly aligned temporal structure of WildRelight enables a new paradigm for domain adaptation. We demonstrate this by introducing a physics-guided inference framework that leverages the captured natural light evolution as a self-supervised constraint. By integrating Diffusion Posterior Sampling (DPS) with temporal Sampling-Aware Test-Time Adaptation (TTA), we show that the dataset allows synthetic models to align with real-world statistics on-the-fly, transforming the intractable sim-to-real challenge into a tractable self-supervised task. The dataset and code will be made publicly available to foster robust, physically-grounded relighting research.","upvotes":2,"discussionId":"6a042ecb86b054ce2fa41063","projectPage":"https://lez-s.github.io/wildrelight_proj/","ai_summary":"WildRelight dataset addresses the gap between synthetic and real-world single-image relighting by providing high-resolution outdoor scenes with aligned natural illumination, enabling physics-guided domain adaptation through diffusion posterior sampling and test-time adaptation.","ai_keywords":["single-image relighting","generative models","domain shift","diffusion posterior sampling","test-time adaptation","physics-guided inference","temporal Sampling-Aware Test-Time Adaptation"],"organization":{"_id":"65206b4136008ecc886b085d","name":"dtudk","fullname":"Technical University of Denmark","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/xJq65HwJYs32ove1JVx3n.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"63be9021da08ed0544f36c38","avatarUrl":"/avatars/42511d8b1d3a3ef99bc154c98b72dfba.svg","isPro":false,"fullname":"onurcan","user":"monurcan","type":"user"},{"_id":"69ffa09cf204ce9d1821803a","avatarUrl":"/avatars/caca3cc632dc36ba8babc753d81cb4fd.svg","isPro":false,"fullname":"PMendes","user":"peteofthewole","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"65206b4136008ecc886b085d","name":"dtudk","fullname":"Technical University of Denmark","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/xJq65HwJYs32ove1JVx3n.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2605/2605.11696.md"}">
WildRelight: A Real-World Benchmark and Physics-Guided Adaptation for Single-Image Relighting
Abstract
WildRelight dataset addresses the gap between synthetic and real-world single-image relighting by providing high-resolution outdoor scenes with aligned natural illumination, enabling physics-guided domain adaptation through diffusion posterior sampling and test-time adaptation.
AI-generated summary
Recent single-image relighting methods, powered by advanced generative models, have achieved impressive photorealism on synthetic benchmarks. However, their effectiveness in the complex visual landscape of the real world remains largely unverified. A critical gap exists, as current datasets are typically designed for multi-view reconstruction and fail to address the unique challenges of single-image relighting. To bridge this synthetic-to-real gap, we introduce WildRelight, the first in-the-wild dataset specifically created for evaluating single-image relighting models. WildRelight features a diverse collection of high-resolution outdoor scenes, captured under strictly aligned, temporally varying natural illuminations, each paired with a high-dynamic-range environment map. Using this data, we establish a rigorous benchmark revealing that state-of-the-art models trained on synthetic data suffer from severe domain shifts. The strictly aligned temporal structure of WildRelight enables a new paradigm for domain adaptation. We demonstrate this by introducing a physics-guided inference framework that leverages the captured natural light evolution as a self-supervised constraint. By integrating Diffusion Posterior Sampling (DPS) with temporal Sampling-Aware Test-Time Adaptation (TTA), we show that the dataset allows synthetic models to align with real-world statistics on-the-fly, transforming the intractable sim-to-real challenge into a tractable self-supervised task. The dataset and code will be made publicly available to foster robust, physically-grounded relighting research.
Community
The first in-the-wild dataset for evaluating single-image relighting models
Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.
Tap or paste here to upload images
Cite arxiv.org/abs/2605.11696 in a model README.md to link it from this page.
Cite arxiv.org/abs/2605.11696 in a Space README.md to link it from this page.
Discussion (0)
Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.
Sign in →No comments yet. Sign in and be the first to say something.