Diffusion Model
A diffusion model is the AI architecture behind most image and video generators: it starts from random noise and gradually denoises it into a coherent result guided by your prompt.
Diffusion models learn by adding noise to training images and then learning to reverse it. At generation time they begin with pure noise and iteratively 'denoise' toward an image that matches the prompt — which is why generation happens in steps.
Nearly all modern text-to-image and text-to-video systems are diffusion-based. Understanding this explains related controls: the seed sets the starting noise, more steps can add detail, and negative prompts steer the denoising away from unwanted features.