Toward Real-World Adoption of Portrait Relighting via Hybrid Domain Knowledge Fusion
It addresses the practical deployment of portrait relighting for real-world applications by overcoming domain gaps and computational costs.
The paper tackles real-world portrait relighting by fusing synthetic, OLAT, and real-world datasets into a compact model, achieving 6x to 240x inference speedup while maintaining state-of-the-art visual quality.
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.