7 terms
A diffusion model is a generative AI model that creates images, audio or video by learning to reverse a gradual noising process step by step.
A GAN is a generative AI model in which a generator and a discriminator compete, so the generator learns to produce realistic synthetic data.
Generative AI is the branch of artificial intelligence whose models learn patterns in data and create new text, images, audio, video or code from prompts.
Latent space in generative AI is the compressed internal representation in which a model encodes data and from which it generates new samples.
Stable Diffusion is an open-weight latent diffusion model for generative AI that creates and edits images from text prompts by denoising in a latent space.
Text-to-image is a generative AI task in which a model creates a new image from a written description, most often using a latent diffusion model.
Text-to-video is a generative AI technique that creates video clips from a written prompt by generating frames that stay consistent over time and motion.