Generative AI refers to models that can create new content, such as text, images, audio, or video, based on patterns learned from existing data. This is in contrast to discriminative models, which categorize or predict labels for existing data.
Diffusion Models
Modern image generation (like Midjourney and Stable Diffusion) is largely powered by Diffusion Models. These models learn to generate data by reversing a process that gradually adds noise to an image.
The Diffusion Process
- Forward Diffusion: Gradually adding Gaussian noise to an image until it becomes pure noise.
- Reverse Diffusion (Reverse Process): Learning to predict the noise added at each step to reconstruct the original image.
- Conditioning: Using text prompts or other inputs to guide the reverse process towards a specific output.
Generative Adversarial Networks (GANs)
GANs consist of two neural networks competing against each other: a Generator and a Discriminator. The generator tries to create realistic data, while the discriminator tries to distinguish between real data and the generator's fakes.