Generative AI, especially image generation, has become widely familiar today. Before diffusion models, other generative approaches like GANs, VAEs, and Flow-based models already existed. Diffusion models are conceptually similar to VAEs but differ in the details of how they reconstruct data. They work by gradually adding Gaussian noise to data and training a model to remove that noise. During training, text embeddings are incorporated, enabling the model to turn prompts into high-quality images.