Private deployment of Stable Diffusion to play with AI drawing

Stable Diffusion is an open source deep learning model. Its main function is to generate high-quality images through text descriptions, and supports functions such as graph generation, model merging, and model training. The operating interface of the model can be seen in the figure below.

How to generate pictures

The following is an introduction to the process of generating a picture of a deer drinking water. When generating a picture, it is divided into prompt words and negative prompt words. When entering the prompt words, you must describe it clearly and try to describe the scene, object, style and color you want in detail. . For example, instead of just saying "the deer drinks water", it says "a creek, next to dense trees, and there are deer drinking water next to the creek". The negative prompt words are in the opposite direction. For example: no buildings, no people , no bridges, no fences, and too vague a description may lead to results that are not as expected.

The core advantages of Stable Diffusion

Advantage

There are many applications for AI painting, so what are the core advantages of Stable Diffusion?

Compared with many other AI painting tools, Stable Diffusion is completely open source, which means that anyone can use, modify and distribute it for free, and can train the model by themselves. There are many styles of paintings, and there are many styles of models in civitai (overseas) and LibuLibu AI (domestic), animation games, architectural space design, two-dimensional, 3D, logo icons, etc. . .

I saw an article before that said that payment is automatic, and Stable Diffusion is equivalent to manual transmission.

Then Stable Diffusion stands out among many AI painting applications, mainly due to the following core advantages:

Compared with many other AI painting tools, Stable Diffusion is completely open source, which means that anyone can use it for free, and it can be developed on the original basis for users to register.

Stable Diffusion supports a high degree of customization, and users can adjust model parameters to generate images that meet specific styles or needs. Competent users can even train models.

Stable Diffusion is capable of producing high-resolution, visually impressive images. It leverages advanced deep learning technology to generate detailed, complex images based on text descriptions.

Stable Diffusion is particularly good at generating images based on text prompts, and you can use simple descriptions to create the images you want.

Stable Diffusion is involved in many fields such as artistic creation, entertainment industry, advertising, education, and side business.

scenes to be used

You can publish some works through a certain music, a certain hand and a certain book. AI painting is not very common in self-media yet, and the traffic is pretty good. The key to monetizing on these platforms is to continuously produce high-quality and attractive content, actively interact with fans, and build a fan base.

One thing I have been trying is to use GPT to generate copy to make graphic videos, generate multiple pictures through carefully created copy, and then combine these pictures into a video. In this process, picture quality plays a decisive role in the traffic of the video.

There are many picture tools in Mouyin (I use Blue Cat Wallpaper). This picture tool allows fans to download pictures here. You must watch the advertisement before downloading. You can convert the generated high-quality avatars and wallpapers. , logo into the tool to earn advertising revenue.

Well enough:

After you have traffic, you can customize it and customize unique paintings according to customer needs. For mothers, you can customize baby avatars and so on.

AI painting stalls can be set up offline. You only need a laptop and a printer to support a stall, such as in a university town or a night market with a lot of traffic.

You can draw portraits based on the ability of drawings, and then sell them for cash.

Installation and deployment

Prerequisite

You need to install Rainbond. Rainbond is a cloud-native application management platform that does not require knowledge of Kubernetes. Deployment of applications does not require complex configuration.

The machine configuration for deploying Stable Diffusion is no less than CPU: 8 cores, memory: 16GB.

step

Enter Rainbond platform management, select the project team in the sidebar, and create your first team.

After clicking to enter the team – Create a new application – External application market – Select the open source application store – Search for Stable-diffusion.

You can choose the application version to install the Stable-diffusion application.

After successful creation, you can see the Stable-diffusion application in the topology diagram of the application view.

Click on the gateway on the left - click on the current component gateway - edit - turn on WebSocket support.

Enter the component and wait for the initialization file - check the log (the decompression process is slow, please wait).

Seeing this log means that the startup is successful and you can access it.

Click the access button in the upper right corner to access Stable-diffusion.

Stable-diffusion interface, a Chinese language pack was also added when making this application.

The installation has been completed here, you can try to generate the desired picture.

The above is the detailed content of Private deployment of Stable Diffusion to play with AI drawing. For more information, please follow other related articles on the PHP Chinese website!