1. What is Stable Diffuson AI

Stable Diffusion is a deep learning, text-to-image model released in 2022 based on diffusion techniques.

But what actually is it? Is Stable Diffusion an Image generator? Or is it Text to image (text-2-img) generator or is Stable Diffusion image to image (img-2-img) generator? Well it is everything! Let's take a look, what you can make if you're using Stable Diffusion.

2. What you can make with Stable Diffusion AI?

Almost everything that you can imagine and you can describe in words you can generate with Stable diffusion: you can imagine an animal, nature, another planet environment or a person who does not exist in real, but can be shown as a real person on an image or even a video. Stable Diffusion makes it possible by using one of the existing model.

3. What is a Model in Stable Diffusion AI?

A model is a special file with special numbers of the special Matrix (from Math) that is responsible for generating resulting image or video. A lot of depends on what exactly model do you use in Stable Diffusion. If you choose another model - you will get another result, even by using the same prompt.

4. What is prompt?

Prompt - is an English (or any other language) sentence that is understood by Stable Diffusion to generate image or video based on. You can enter anything you want and Stable Diffusion will try to imagine it and show you what it "think". Be aware that Stable Diffusion "thinks" according what it knows from the Model (see above). If it does not know about something (for example a word that you have written into the prompt) - it will not understand you but anyway Stable Diffusion will generate something for you by guessing :)

5. What is the difference between Stable Diffusion, MidJourney, Dall-E, etc. tools out there?

Not a simple question. But in general all tools aim to generate images based on a prompt. The different is in the models and some advanced techniques that are used to generate images. For example, if you will try to ask Stable Diffusion to generate something complex - it will not be able to generate it, because a Model may not include these data in it. MidJourney and Dall-E have more data to generate more complex images.

Some people sees the difference when the same prompt is used for Stable Diffusion, MidJourney and Dall-E. But we can change the model in Stable Diffusion to make our result more precise - we cannot do this in MidJourney and Dall-E.

Stable Diffusion is a free tool (with limitations for some models), ModJourney and Dall-E is included into paid packages of the companies that are developing such tools.

At the same time to run MidJourney and Dall-E you don't need to have powerful hardware such as GPU. StDiff.io does not require you to have such powerful GPU as well.

6. What do you need to run Stable Diffusion on you machine?

To run Stable Diffusion you need to have modern GPU (Graphics Card) with at least 4 GB VRAM (Virtual RAM on this card). Nvidia Graphics cards is recommended. You can try to train yourself with prompting by using stdiff.io for free.