A generative model that creates images (or other data) by learning to reverse a gradual noising process, producing high-quality outputs from random noise.
The AI that makes pictures — it starts with TV static and slowly turns it into a beautiful image, like magic.
The type of AI behind image generators like DALL-E and Midjourney. It learns by adding noise to images, then learns to remove it to create new ones.
A generative model that creates images (or other data) by learning to reverse a gradual noising process, producing high-quality outputs from random noise.
A generative architecture that learns a denoising process over a forward diffusion (noising) trajectory, enabling high-fidelity image, video, and audio synthesis via iterative refinement from Gaussian noise.
A latent-variable model defining a forward Markov chain that gradually adds Gaussian noise and a learned reverse process that denoises — optimized via a variational bound equivalent to denoising score matching, with classifier-free guidance steering generation.
Want to explore Diffusion Model in depth?
Ask SeekBox and get answers from 7 AI engines at once.
Try it in SeekBox →