Stable Diffusion is an open-source latent diffusion model developed by Stability AI that generates images from text descriptions. Unlike DALL-E or Midjourney, it operates in a compressed latent space rather than pixel space, making it computationally efficient enough to run on consumer GPUs. The model uses a VAE encoder-decoder architecture with a U-Net noise predictor and CLIP text encoder to transform text prompts into detailed images through an iterative denoising process. Understanding its parameters—from CFG scale to sampling methods—unlocks precise control over composition, style, and quality, while extensions like ControlNet and LoRA enable advanced customization without retraining the entire model.