Normalizing Flow as a Flexible Fidelity Objective for Photo-Realistic Super-resolution
This work addresses the challenge of generating photo-realistic super-resolution images for applications like image enhancement, offering a novel method to resolve a known bottleneck in existing approaches.
The paper tackles the problem of super-resolution by addressing the conflict between pixel-wise losses and adversarial losses, which degrades image quality, and proposes using normalizing flows as a fidelity objective to improve visual quality and consistency, outperforming state-of-the-art methods in user studies across three datasets and scale factors.
Super-resolution is an ill-posed problem, where a ground-truth high-resolution image represents only one possibility in the space of plausible solutions. Yet, the dominant paradigm is to employ pixel-wise losses, such as L_1, which drive the prediction towards a blurry average. This leads to fundamentally conflicting objectives when combined with adversarial losses, which degrades the final quality. We address this issue by revisiting the L_1 loss and show that it corresponds to a one-layer conditional flow. Inspired by this relation, we explore general flows as a fidelity-based alternative to the L_1 objective. We demonstrate that the flexibility of deeper flows leads to better visual quality and consistency when combined with adversarial losses. We conduct extensive user studies for three datasets and scale factors, where our approach is shown to outperform state-of-the-art methods for photo-realistic super-resolution. Code and trained models will be available at: git.io/AdFlow