The model encodes the semantic meaning of the prompt to guide the image generation process. Latent Space Representation: Stable Diffusion uses a Variational Autoencoder (VAE) to compress images ...