Paper Detail
Lezhong Wang, Mehmet Onurcan Kaya, Siavash Bigdeli, Jeppe Revall Frisvad
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.
No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.
No ranking explanation is available yet.
No tags.
@misc{wang2026wildrelight,
title = {WildRelight: A Real-World Benchmark and Physics-Guided Adaptation for Single-Image Relighting},
author = {Lezhong Wang and Mehmet Onurcan Kaya and Siavash Bigdeli and Jeppe Revall Frisvad},
year = {2026},
abstract = {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},
url = {https://huggingface.co/papers/2605.11696},
keywords = {single-image relighting, generative models, domain shift, diffusion posterior sampling, test-time adaptation, physics-guided inference, temporal Sampling-Aware Test-Time Adaptation, huggingface daily},
eprint = {2605.11696},
archiveprefix = {arXiv},
}
{}