Image generation with Shortest Path Diffusion

International Conference on Machine Learning (ICML), 2023
Generative Models

Ayan Das

MediaTek Research UK

Stathi Fotiadis

Imperial College London, UK

MediaTek Research UK

Anil Batra

University of Edinburgh UK

MediaTek Research UK

Farhang Nabiei

MediaTek Research UK

FengTing Liao

MediaTek Research TW

Sattar Vakili

MediaTek Research UK

Da-shan Shiu

MediaTek Research

Alberto Bernacchia

MediaTek Research UK


April 25, 2023

Paper (official) Paper (axXiv) Code


The field of image generation has made significant progress thanks to the introduction of Diffusion Models, which learn to progressively reverse a given image corruption. Recently, a few studies introduced alternative ways of corrupting images in Diffusion Models, with an emphasis on blurring. However, these studies are purely empirical and it remains unclear what is the optimal procedure for corrupting an image. In this work, we hypothesize that the optimal procedure minimizes the length of the path taken when corrupting an image towards a given final state. We propose the Fisher metric for the path length, measured in the space of probability distributions. We compute the shortest path according to this metric, and we show that it corresponds to a combination of image sharpening, rather than blurring, and noise deblurring. While the corruption was chosen arbitrarily in previous work, our Shortest Path Diffusion (SPD) determines uniquely the entire spatiotemporal structure of the corruption. We show that SPD improves on strong baselines without any hyperparameter tuning, and outperforms all previous Diffusion Models based on image blurring. Furthermore, any small deviation from the shortest path leads to worse performance, suggesting that SPD provides the optimal procedure to corrupt images. Our work sheds new light on observations made in recent works, and provides a new approach to improve diffusion models on images and other types of data.


BibTeX citation:
  author = {Das, Ayan and Fotiadis, Stathi and Batra, Anil and Nabiei,
    Farhang and Liao, FengTing and Vakili, Sattar and Shiu, Da-shan and
    Bernacchia, Alberto},
  title = {Image Generation with {Shortest} {Path} {Diffusion}},
  booktitle = {International Conference on Machine Learning (ICML),
  date = {2023-04-25},
  url = {},
  langid = {en}
For attribution, please cite this work as:
Das, Ayan, Stathi Fotiadis, Anil Batra, Farhang Nabiei, FengTing Liao, Sattar Vakili, Da-shan Shiu, and Alberto Bernacchia. 2023. “Image Generation with Shortest Path Diffusion.” In International Conference on Machine Learning (ICML), 2023.