Stable Diffusion, how to install and use it for free

1. Home
2. artificial intelligence
3. Stable Diffusion, how to install and use it for free

steady spread is a generative AI created by research labs mid journey, located in San Francisco. Creating images with AI requires a lot of computing resources; For this reason, good quality tools found online are often either very expensive or very complicated to use. this software can Install locally free, allows you to create high resolution images without any limitation, from text messages,

The software is based on the Diffusion Model, which is a generative model that creates images by gradually adding details to an image of random noise. Steady Diffusion has been trained on a huge dataset of images and text, which allows it to generate realistic and creative images.

how does the steady spread work

To use the still spread, you must provide a text prompt of your choice, such as a description of an image, scene, or object. Stationary diffusion will use the given signal to produce a matching image. The quality of the image generated will depend on the quality of the signal provided.

There are minimum system requirements for a local installation of Stable Propagation. The speed at which Stable Diffusion will be able to generate material will depend on the hardware at our disposal, mainly on the processing power of the CPU and GPU.

In this step-by-step guide, we’ll take a look at How to install Stable Diffusion on Windows and Mac for free, and how to write the first prompt.

How to install Stable Diffusion on Windows for free

Usually, in order to use Generative AI on your computer you need to install the programming language Python And perform many operations from command line which are not within everyone’s reach.

easy spread Stable is a distribution of the spread that is easy to install and free to use. The tool is open source and allows you to install for free all the software components needed to run a stable spread from its web interface.

The minimum hardware requirements to easily install and run Stable Diffusion on Windows are:

• GPU: Nvidia graphics card with 4GB VRAM or more
• RAM: 8GB or more
• Storage Space: 20GB or more
• Operating System: Windows 10/11

The installation steps are simple to follow:

1. download executable from here
2. run the installer and let it run
3. Click on “Install”
4. Choose the installation path (if you have Windows 10, install Easy Diffusion on the top level of the drive, for example C:\EasyDiffusion, to avoid known problems due to path length limitations)
5. Read the terms of service carefully and accept them
6. At the end of the installation, click “Finish”

After this process, Static Propagation will start automatically by opening a window of your default browser to access the graphical interface. If the software doesn’t start automatically, you’ll also find a Quick Launch icon on your desktop and in the Start menu.

Now you can go directly to the paragraph “How to use StableDiffusion”.

How to install Stable Diffusion on Mac for free

Stable Diffusion works especially well on the latest generation of Macs. It was created to avoid setting up a development environment and using the command line. spread beeis an open source software that reduces the required operations to a simple click.

Before we get started with the actual installation, let’s take a look at some of the minimum system requirements to get Stable Propagation running smoothly:

• Chip: Apple Silicon (all M1 and M2 versions)
• RAM: 8GB or more
• Storage Space: 20GB or more
• Operating System: macOS 12.5.1 or later

These are the steps to install Stable Diffusion on your Mac

1. Download DiffusionB from here.
2. open downloaded file
3. Drag the package to your default “Applications” folder
4. Start DiffusionB and wait for the default template to download. This operation takes a variable amount of time, depending on the speed of your Internet connection.

Once the download is complete, you can use the app. Start with the “Text to Image” tab, enter the prompt in the appropriate text field and press “Generate.” The DiffusionBee app will generate an image based on your description. Generation may take several minutes depending on your hardware configuration. Once the image is generated, you can choose to save it, or improve its quality with the upscaling option available in the menu accessible via the three bars to the left of the image.

how to use static diffusion

If you’ve already tried to generate some images immediately after installation, you’ve probably noticed that the result is a far cry from the spectacular images you can find on lexica.art. The reason is simple: Behind every good image is an analysis to choose the most suitable generative model and create the ideal signal,

In general, the more details you provide in the prompt, the better results the system will generate for you. But finding the right sign can be difficult.

An easy way to familiarize yourself with making signs for a static spread is to go to lexica.art above and copy the signs from images you like, perhaps piecing them together to find the right combination to get the desired image. Including

Here are some examples of text messages you can use with static spreads:

• “an astronaut in a beach chair, vibrant lighting, elegant, highly detailed, smooth, sharp focus, illustration, graceful, geometric, trending at ArtStation, full body, cinematic”
• Motorcycle designed by Teenage Engineering + Simone Stellenhag, styled by Laurie Greasley, Studio Ghibli, Akira Toriyama, James Illard, Genshin Impact, 8k resolution, hyper realistic, Dieter Rams. Detailed render. Smooth Cam de Leon Eric Jenner Dramatic, Mark Raiden and Pixar and Hayao Miyazaki”
• “Vintage 90’s anime style. A lonely astronaut walking down the street; by Hajime Sorayama, Greg Tochini, Virgil Finley, Sci-Fi. Line art. Environment arcade art”

If you’re sending a long prompt and want the AI ​​to focus on a specific word, you can use parentheses to highlight them, which changes their weight for the AI. In this way, the control over the final result is increased, directly choosing which will be the key elements of the graphic composition.

To fully automate the prompt writing operation, it is possible for other text-generative AIs, such as Ask chat gpt Or google bardto make the correct signal.

Configure neural network parameters

After understanding the mechanism of the prompt generation, you can proceed to configure the advanced parameters of the neural network. Using AI settings properly requires a thorough understanding of them. So let’s look at the main points:

• Stages: Think of stages as iterations of the image creation process. During the first few passes, the image will look more like fuzzy random noise. With each cycle, the AI ​​will modify the image by adding new details.
• Size: By default, static diffusion produces images of 512 x 512 pixels. You can resize to get larger images, but this will require more computing power and therefore more time.
• CFG Scale: This setting indicates how well the static spread will follow your signal. Reducing the scale to zero means the AI ​​will only consider parts of the signal; Will be more creative. Conversely, by bringing the scale to the maximum value, the result will faithfully follow the description.
• Mode: In addition to the Text2Image mode, which generates an image from the text prompt, other features can be used. With img2img, for example, it is possible to create an image from another image, perhaps a rough sketch. with mode inpainting You can “overwrite” a portion of the image, recreating it. This work is indispensable for correcting small defects. Together outpaintingHowever, the AI ​​will complete a partial image, generating a new adjacent content that corresponds to the content presented in the image provided as input.

conclusion

Static diffusion is still under development, but represents a class of generative systems that will become a powerful tool for artists, designers, and creative professionals in the future. With steady dissemination, it will be possible to create original images for free, which can be used for both entertainment and marketing.

Of course, this is not a perfect system: The results are not always as desired, and in some cases the retouching of the face and hands can still be a problem. Also, depending on the hardware used, image creation may still be especially slow, In any case, it represents a fundamental step towards an AI-driven future, where being able to understand and use tools capable of automating a large portion of our work will be essential.