| title | RelightFormer: Feed-forward Generative Transformer for Multiview Object Relighting |
|---|---|
| authors | Hejun Wang, Jinxi Li, Junwei Jiang, Shiwei Mao, Hu Cheng, Shouwang Huang, Bo Yang |
| date | 2026-09-07 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2609.07414 |
| affiliation | The Hong Kong Polytechnic University |
| upvotes | 2 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf code |
| abstract | Image relighting is traditionally tackled via complex inverse rendering pipelines, which suffer from ill-posed optimization, or single-image generative models that ignore crucial multi-view cues necessary for understanding 3D geometry and material interactions. To address these limitations, we introduce a feed-forward generative Transformer for direct single- and multi-view image relighting that entirely bypasses explicit intrinsic property estimation. Adapted from a video foundation model, our architecture features a latent illumination module that dynamically injects target environment maps into spatial features via cross-attention. Furthermore, we employ permutation-invariant positional encodings to symmetrically process unordered multi-view inputs without sequential bias. To train this robust data-driven model, we construct the massive Laval Objaverse Dataset (LOD), comprising 90K objects and 39K unique illuminations. Extensive experiments demonstrate state-of-the-art visual quality, photorealistic relighting quality, and strong zero-shot generalization across single-view, multi-view, and novel-view relighting tasks. |
cite
@misc{wang2026relightformer,
title={RelightFormer: Feed-forward Generative Transformer for Multiview Object Relighting},
author={Hejun Wang and Jinxi Li and Junwei Jiang and Shiwei Mao and Hu Cheng and Shouwang Huang and Bo Yang},
year={2026},
eprint={2609.07414},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2609.07414}
}