# What is Diffusion Model? Definition & Technical Mechanics

> **Source:** [SPIDITS AI AI Glossary](https://spidits.com/ai-glossary)  
> **Category:** Generative AI  
> **Canonical Citation:** [https://spidits.com/ai-glossary/diffusion-model](https://spidits.com/ai-glossary/diffusion-model)

## Definition
A Diffusion Model is a class of generative AI models that generate data by learning to reverse a process of gradual noise addition. By starting with random noise and iteratively removing it, the model can generate high-resolution images, video, or audio.

## Architectural Mechanics & Deep Dive
A diffusion model is a class of generative models that generates high-quality images, audio, or video by reversing a gradual noise process. During training, the model learns to gradually denoise an image that has been corrupted with random Gaussian noise. During generation, the model starts with pure noise and iteratively refines it over multiple steps based on user text prompts, forming the technology behind platforms like Midjourney and Stable Diffusion.

## Industry Applications & Real-World Use Cases
Image generation (Stable Diffusion, Midjourney), video generation, and audio synthesis

## Frequently Asked Questions

### How do diffusion models compare to GANs?
Diffusion models are generally more stable to train than GANs and produce higher quality/diversity, though they are computationally slower due to iterative sampling.

### What are the two phases of diffusion models?
The forward process (adding noise to an image until it is pure noise) and the reverse process (learning to subtract noise step-by-step).

## Related Terminology
- [generative-ai](https://spidits.com/ai-glossary/generative-ai)
- [gpu](https://spidits.com/ai-glossary/gpu)

---
*From the SPIDITS AI Knowledge Base (336+ terms indexed). Explore full technical definitions: [https://spidits.com/ai-glossary](https://spidits.com/ai-glossary)*