Skip to content
ImageQuarry

Question

What is a diffusion model?

A diffusion model is a type of generative AI model that learns to create images by reversing a process that gradually turns training images into noise. When generating, it starts from random noise and removes it step by step until an image matching the prompt appears.

A widely cited 2020 paper, "Denoising Diffusion Probabilistic Models" by Ho, Jain and Abbeel, showed that this method could produce high-quality images. In text-to-image systems, the prompt guides each denoising step toward the picture you described.

Not every current image model is a diffusion model, and model makers do not always publish their architecture. For choosing a tool, the architecture matters less than what a model does well for you: following instructions, keeping details during edits, rendering text and matching a style.

In ImageQuarry you describe the image or the change in plain words and compare the versions that come back.

Source: Ho, Jain and Abbeel (2020), arXiv. Checked October 20, 2026.

Next step

Why do different AI models give different results from the same prompt?Why models answer the same prompt differently.Read more

Related questions

Why do different AI models give different results from the same prompt?Different training and goals produce different pictures.Read more Which AI image model is best?It depends on the task; compare on your own request.Read more What is generative fill?Adobe's selection-based fill in Photoshop.Read more How does text-to-image AI work?A model turns a text description into a new image.Read more

Last reviewed October 20, 2026 by ImageQuarry

Create an image. Edit it with words. Keep refining.

Skip the theory and describe the image

In ImageQuarry, you describe what you want in plain words and refine it version by version.

Start creatingAlready have an account? Sign in