When we think about image generative models, the most notable one that comes to mind would be DALL-E2, Stable Diffusion or MidJourney. A lot of these underlying techniques were initially invented by physicists, which have become very important in machine learning.

Some of those physicists are Jascha Sohl-Dickstein, Yang Song and Jonathan Ho on diffusion model algorithm.

Let's find out what they did!

https://www.quantamagazine.org/the-physics-principle-that-inspired-modern-ai-art-20230105/

The Physics Principle That Inspired Modern AI Art | Quanta Magazine

Diffusion models generate incredible images by learning to reverse the process that, among other things, causes ink to spread through water.

Quanta Magazine