Toward Real-World Adoption of Portrait Relighting via Hybrid Domain Knowledge Fusion

2026-04-24

AI Summary

This work addresses the challenges of portrait relighting in real-world scenarios—namely, domain discrepancies, camera sensitivity, and high computational cost—by introducing a hybrid-domain knowledge fusion paradigm. The proposed method efficiently integrates synthetic data, single-light (OLAT) captures, and real-world images for the first time, leveraging a domain-aware adaptive prior model and an enhanced knowledge distillation framework to train a lightweight multi-domain expert network. The resulting compact model achieves state-of-the-art visual quality while delivering 6× to 240× inference speedup. To further advance multi-domain generalization, the authors also release a large-scale, high-fidelity synthetic dataset accompanying the method.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

The real-world adoption of portrait relighting is hindered by dataset domain gaps, camera sensitivity, and computational costs. We address these challenges with Hybrid Domain Knowledge Fusion, a paradigm that fuses the specialized strengths of synthetic, One-Light-at-A-Time (OLAT), and real-world datasets into a compact model. Our approach features specialized prior models hardened by domain-aware adaptation, followed by augmented knowledge distillation into a lightweight student model with multi-domain expertise. Our method demonstrates a 6x to 240x inference speedup while maintaining state-of-the-art (SOTA) visual quality in the experiments. Additionally, we construct a massive, high-fidelity synthetic dataset with diverse ground-truth intrinsics to support our training pipeline.

Problem

• portrait relighting
• domain gap
• camera sensitivity
• computational cost
• real-world adoption

Innovation

• Hybrid Domain Knowledge Fusion
• Portrait Relighting
• Knowledge Distillation
• Domain Adaptation
• Synthetic Dataset