Modern Computer Vision with PyTorch - Second Edition

17

Applications of Stable Diffusion

In the previous chapter, we learned about how diffusion models work, the architecture of Stable Diffusion, and diffusers – the library.

While we learned about generating images, unconditional and conditional (from a text prompt), we still did not learn about having the ability to control the images – for example, I might want to replace a cat in an image with a dog, make a person stand in a certain pose, or replace the face of a superhero with a subject of interest. In this chapter, we will learn about the model training process and coding some of the applications of diffusion that help in achieving the above. In particular, we will cover the following topics:

• In-painting to replace objects within an image ...