Unsupervised HDR Imaging: What Can Be Learned from a Single 8-bit Video?

TL;DR

This paper introduces a zero-shot method for HDR video generation from a single SDR video, demonstrating that extensive datasets are not always necessary for high-quality HDR reconstruction.

Contribution

The work presents a novel zero-shot approach that learns from only one SDR video to produce HDR videos, challenging the reliance on large datasets in deep learning methods.

Findings

Single SDR video can be sufficient for high-quality HDR reconstruction

The proposed method outperforms some state-of-the-art inverse tone-mapping techniques

Zero-shot learning reduces the need for large training datasets

Abstract

Recently, Deep Learning-based methods for inverse tone-mapping standard dynamic range (SDR) images to obtain high dynamic range (HDR) images have become very popular. These methods manage to fill over-exposed areas convincingly both in terms of details and dynamic range. Typically, these methods, to be effective, need to learn from large datasets and to transfer this knowledge to the network weights. In this work, we tackle this problem from a completely different perspective. What can we learn from a single SDR video? With the presented zero-shot approach, we show that, in many cases, a single SDR video is sufficient to be able to generate an HDR video of the same quality or better than other state-of-the-art methods.