BlueNeg: A 35mm Negative Film Dataset for Restoring Channel-Heterogeneous Deterioration

Abstract

While digitally acquired photographs have been dominating since around 2000, there remains a huge amount of legacy photographs being acquired by optical cameras and are stored in the form of film negatives. In this paper, we address the unique challenge of channel-heterogeneous deterioration in film negatives and introduce BlueNeg, the first high-quality 35mm negative film dataset specifically designed for restoration in this context. Our work aims to spotlight this underexplored area of image restoration on channel-heterogeneous deterioration. A key challenge in evaluating restoration performance is the absence of ground truth, as many negatives are already contaminated. To overcome this, we utilize printed photographs from the same negatives as proxies for quantitative assessment. These prints, while affected by spatially invariant color fading, are less affected from channel-heterogeneous degradation. We propose a reverse-printing process to estimate pseudo-ground truth from the prints and introduce a dedicated evaluation protocol. Our empirical analysis shows that existing restoration methods struggle with this dataset, highlighting the need for specialized techniques. We hope that BlueNeg and our evaluation framework will inspire further research in legacy photograph restoration, contributing to the digital preservation of historical moments for archival use. Our dataset is publicly available at https://huggingface.co/datasets/ttgroup/blueneg-release and https://blueneg.github.io