Flash-Split: 2D Reflection Removal with Flash Cues and Latent Diffusion Separation

TL;DR

Flash-Split is a novel framework that effectively separates reflections from transmitted images using flash cues and latent diffusion, even with misaligned flash/no-flash pairs, achieving state-of-the-art results.

Contribution

It introduces a two-stage latent-space reflection separation method leveraging diffusion models and flash cues, addressing misalignment issues and restoring high-resolution details.

Findings

Achieves state-of-the-art reflection separation performance.

Effectively handles misaligned flash/no-flash image pairs.

Restores high-resolution, faithful details in separated images.

Abstract

Transparent surfaces, such as glass, create complex reflections that obscure images and challenge downstream computer vision applications. We introduce Flash-Split, a robust framework for separating transmitted and reflected light using a single (potentially misaligned) pair of flash/no-flash images. Our core idea is to perform latent-space reflection separation while leveraging the flash cues. Specifically, Flash-Split consists of two stages. Stage 1 separates apart the reflection latent and transmission latent via a dual-branch diffusion model conditioned on an encoded flash/no-flash latent pair, effectively mitigating the flash/no-flash misalignment issue. Stage 2 restores high-resolution, faithful details to the separated latents, via a cross-latent decoding process conditioned on the original images before separation. By validating Flash-Split on challenging real-world scenes, we demonstrate state-of-the-art reflection separation performance and significantly outperform the baseline methods.