POLAR: A Portrait OLAT Dataset and Generative Framework for Illumination-Aware Face Modeling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Zhuo, Yang, Chengqun, Su, Zhuo, Lv, Zheng, Gao, Jingnan, Zhang, Xiaoyuan, Yang, Xiaokang, Yan, Yichao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908714299031552
author Chen, Zhuo
Yang, Chengqun
Su, Zhuo
Lv, Zheng
Gao, Jingnan
Zhang, Xiaoyuan
Yang, Xiaokang
Yan, Yichao
author_facet Chen, Zhuo
Yang, Chengqun
Su, Zhuo
Lv, Zheng
Gao, Jingnan
Zhang, Xiaoyuan
Yang, Xiaokang
Yan, Yichao
contents Face relighting aims to synthesize realistic portraits under novel illumination while preserving identity and geometry. However, progress remains constrained by the limited availability of large-scale, physically consistent illumination data. To address this, we introduce POLAR, a large-scale and physically calibrated One-Light-at-a-Time (OLAT) dataset containing over 200 subjects captured under 156 lighting directions, multiple views, and diverse expressions. Building upon POLAR, we develop a flow-based generative model POLARNet that predicts per-light OLAT responses from a single portrait, capturing fine-grained and direction-aware illumination effects while preserving facial identity. Unlike diffusion or background-conditioned methods that rely on statistical or contextual cues, our formulation models illumination as a continuous, physically interpretable transformation between lighting states, enabling scalable and controllable relighting. Together, POLAR and POLARNet form a unified illumination learning framework that links real data, generative synthesis, and physically grounded relighting, establishing a self-sustaining "chicken-and-egg" cycle for scalable and reproducible portrait illumination. Our project page: https://rex0191.github.io/POLAR/.
format Preprint
id arxiv_https___arxiv_org_abs_2512_13192
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle POLAR: A Portrait OLAT Dataset and Generative Framework for Illumination-Aware Face Modeling
Chen, Zhuo
Yang, Chengqun
Su, Zhuo
Lv, Zheng
Gao, Jingnan
Zhang, Xiaoyuan
Yang, Xiaokang
Yan, Yichao
Computer Vision and Pattern Recognition
Face relighting aims to synthesize realistic portraits under novel illumination while preserving identity and geometry. However, progress remains constrained by the limited availability of large-scale, physically consistent illumination data. To address this, we introduce POLAR, a large-scale and physically calibrated One-Light-at-a-Time (OLAT) dataset containing over 200 subjects captured under 156 lighting directions, multiple views, and diverse expressions. Building upon POLAR, we develop a flow-based generative model POLARNet that predicts per-light OLAT responses from a single portrait, capturing fine-grained and direction-aware illumination effects while preserving facial identity. Unlike diffusion or background-conditioned methods that rely on statistical or contextual cues, our formulation models illumination as a continuous, physically interpretable transformation between lighting states, enabling scalable and controllable relighting. Together, POLAR and POLARNet form a unified illumination learning framework that links real data, generative synthesis, and physically grounded relighting, establishing a self-sustaining "chicken-and-egg" cycle for scalable and reproducible portrait illumination. Our project page: https://rex0191.github.io/POLAR/.
title POLAR: A Portrait OLAT Dataset and Generative Framework for Illumination-Aware Face Modeling
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.13192