Lossy Compression with Pretrained Diffusion Models
This work addresses the challenge of efficient image compression for applications requiring low bitrates, though it is incremental as it builds on existing algorithms with simple workarounds.
The paper tackled the problem of implementing a principled algorithm for lossy image compression using pretrained diffusion models, and the result was a complete implementation of DiffC that compresses and decompresses images in under 10 seconds, achieving competitive performance with state-of-the-art generative compression methods at ultra-low bitrates.
We apply the DiffC algorithm (Theis et al. 2022) to Stable Diffusion 1.5, 2.1, XL, and Flux-dev, and demonstrate that these pretrained models are remarkably capable lossy image compressors. A principled algorithm for lossy compression using pretrained diffusion models has been understood since at least Ho et al. 2020, but challenges in reverse-channel coding have prevented such algorithms from ever being fully implemented. We introduce simple workarounds that lead to the first complete implementation of DiffC, which is capable of compressing and decompressing images using Stable Diffusion in under 10 seconds. Despite requiring no additional training, our method is competitive with other state-of-the-art generative compression methods at low ultra-low bitrates.