Decoder. plain-English AI glossary

Stable Diffusion

● Core

Open-source text-to-image diffusion model that runs on consumer GPUs and sparked the generative AI art boom.

Think of it like

The Linux of generative AI—free, open, customizable, and community-driven.

Example

Local SD 1.5 with 4GB VRAM generates a 512x512 image in 20 seconds; no API costs or privacy concerns.

How it actually works

Latent diffusion: compress images to latent space via VAE, diffuse in latent space, decode back to pixel space. Much faster than pixel diffusion. Released open-source in 2022, sparking community fine-tunes (LoRA, textual inversion, ControlNet). Model weights available; enables fine-tuning and local deployment.

For product teams

Democratized image generation; eliminated dependency on proprietary APIs.

For engineers

VAE encoder compresses image to latent; UNet with attention in latent space; CLIP encoder guides; VAE decoder reconstructs.

Diffusion — Example of.
Enables.

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open Decoder Add to Chrome