Reflection Removal through Efficient Adaptation of Diffusion Transformers

TL;DR

This paper presents a novel diffusion-transformer framework for single-image reflection removal that leverages pre-trained models, synthetic data, and efficient adaptation to achieve state-of-the-art results.

Contribution

It introduces a diffusion-transformer approach that repurposes pre-trained models with synthetic data and LoRA-based adaptation for reflection removal.

Findings

Achieves state-of-the-art performance on benchmarks.

Effective zero-shot reflection removal results.

Synthetic data improves model generalization.

Abstract

We introduce a diffusion-transformer (DiT) framework for single-image reflection removal that leverages the generalization strengths of foundation diffusion models in the restoration setting. Rather than relying on task-specific architectures, we repurpose a pre-trained DiT-based foundation model by conditioning it on reflection-contaminated inputs and guiding it toward clean transmission layers. We systematically analyze existing reflection removal data sources for diversity, scalability, and photorealism. To address the shortage of suitable data, we construct a physically based rendering (PBR) pipeline in Blender, built around the Principled BSDF, to synthesize realistic glass materials and reflection effects. Efficient LoRA-based adaptation of the foundation model, combined with the proposed synthetic data, achieves state-of-the-art performance on in-domain and zero-shot benchmarks. These results demonstrate that pretrained diffusion transformers, when paired with physically grounded data synthesis and efficient adaptation, offer a scalable and high-fidelity solution for reflection removal. Project page: https://hf.co/spaces/huawei-bayerlab/windowseat-reflection-removal-web