Stable Diffusion
Stable Diffusion is an open-source text-to-image model that creates pictures by gradually removing noise inside a compressed latent space.
June 16, 2026
Stable Diffusion is an open-source latent diffusion model that generates images from a text prompt by repeatedly denoising a compressed representation until a clear picture appears.
How it works
Stable Diffusion is built on the diffusion process: during training, the model learns how images dissolve into random noise, then learns to reverse that process. To generate a new image, it starts from pure noise and removes a little of it at each step, steered by your text prompt, until a finished image forms.
The key innovation is where this happens. Rather than working directly on millions of pixels, Stable Diffusion compresses images into a much smaller latent space using an autoencoder. Denoising in that compact space is dramatically cheaper, which is why the model can run on consumer hardware. A text encoder converts your prompt into numbers the model can follow, and settings like the seed, CFG scale, and number of steps control the output. A final decoder turns the cleaned-up latent back into a full-resolution image.
Why it matters
Stable Diffusion mattered because it was openly released. That openness sparked an entire ecosystem: custom checkpoints, LoRA fine-tunes, ControlNet, and countless tools all grew around it. It made high-quality image generation accessible to people without large compute budgets, and it established many of the terms — latent space, denoising, CFG scale — that are now standard vocabulary across the field.
In eaxy
You do not need to install or tune any of this to use eaxy. eaxy wraps modern generation models behind a clean studio so you can type a prompt, pick from 30+ style packs, and get a polished image in seconds — then bring it to life with Kling 3 video. The diffusion machinery runs for you; you just describe what you want.
Related terms
Frequently asked questions
What is Stable Diffusion?+
It is an open-source AI model that turns a text prompt into an image. It works by starting from random noise and refining it step by step until a coherent picture matching the prompt emerges.
Why is it called 'latent' diffusion?+
Instead of denoising full-resolution pixels, Stable Diffusion does the work in a smaller compressed representation called the latent space. This makes it far faster and lighter to run than pixel-space diffusion.
Is Stable Diffusion free to use?+
The model weights are released under open licenses, so the technology itself is free. Running it still requires capable hardware or a hosted service, and license terms vary by model version.
How is Stable Diffusion different from other image models?+
Its defining traits are being open-source and operating in latent space. Many other models are closed or proprietary, while Stable Diffusion can be downloaded, fine-tuned, and extended by anyone.
Make it with eaxy
Describe anything and generate stunning images in seconds — then bring them to motion with Kling 3.