PerCo (SD): Open Perceptual Compression

TL;DR

PerCo (SD) introduces an open perceptual image compression method based on Stable Diffusion v2.1, offering a competitive alternative to proprietary methods by emphasizing perceptual quality at ultra-low bit rates.

Contribution

This work adapts PerCo to the Stable Diffusion ecosystem, providing an open-source, comprehensive comparison, and analyzing its perceptual advantages and limitations.

Findings

PerCo (SD) shows improved perceptual quality at higher distortion levels.

The model's capacity differences partly explain the perceptual-distortion trade-off.

Open-source release facilitates further research and development in perceptual compression.

Abstract

We introduce PerCo (SD), a perceptual image compression method based on Stable Diffusion v2.1, targeting the ultra-low bit range. PerCo (SD) serves as an open and competitive alternative to the state-of-the-art method PerCo, which relies on a proprietary variant of GLIDE and remains closed to the public. In this work, we review the theoretical foundations, discuss key engineering decisions in adapting PerCo to the Stable Diffusion ecosystem, and provide a comprehensive comparison, both quantitatively and qualitatively. On the MSCOCO-30k dataset, PerCo (SD) demonstrates improved perceptual characteristics at the cost of higher distortion. We partly attribute this gap to the different model capacities being used (866M vs. 1.4B). We hope our work contributes to a deeper understanding of the underlying mechanisms and paves the way for future advancements in the field. Code and trained models will be released at https://github.com/Nikolai10/PerCo.