Fashion AI glossary
Diffusion model
A diffusion model is a class of generative AI that creates images by progressively denoising a random noise pattern, guided by a prompt or reference image. Diffusion models power most modern AI fashion photography tools.
Stable Diffusion, FLUX, DALL-E, Imagen, Midjourney, and most fashion-specific AI engines are diffusion models. The technique starts with random noise and iteratively refines it toward an output that matches a guiding signal (text prompt, reference image, or both).
For fashion product photography, diffusion models are fine-tuned on apparel-specific data — drape, fabric, model poses, garment construction. This is what separates a fashion-tuned platform from a general-purpose image model: the same architecture, very different training.
Brands don't typically need to understand the technology, but knowing 'fashion-fine-tuned diffusion' is what's under the hood explains why catalog outputs from a fashion platform look different from outputs from a general-purpose tool.
