About Lesson
Diffusion models encompass a broader category of generative models. They work by gradually adding noise (randomness) to an existing image until it becomes pure noise.
During the generation process, the model learns to reverse this noise addition step by step, ultimately creating a new image that resembles the training data.