How diffusion models create images

Learn the simple picture: during training, the model learns to remove noise from images step by step. To create an image, it starts from pure noise and removes noise in many small steps, guided by your text (a text encoder turns the prompt into numbers). Stable Diffusion does this in a small compressed 'latent' space, which makes it fast. (Before diffusion, GANs made images with two networks competing — a generator and a discriminator; they are fast but harder to train, and diffusion replaced them for most uses.) Define the product: brief → 4 image options → choose and edit.