Stable Diffusion（InvokeAI）
Overview#
Compared to Midjourney, Stable Diffusion's advantages are: free of charge, supports switching models, higher operability. Its disadvantages are: it mostly requires local deployment.
Suggestion: If you are new to AIGC, it's recommended to find ready-to-use services instead of wasting your passion on local deployment.
InvokeAI is an open-source implementation based on Stable Diffusion，Currently it does not have any online services and requires users to deploy locally on their own (MacOS/Win).
InvokeAI GitHub Repo：https://github.com/invoke-ai/InvokeAI
Preview#
|
|
|
|---|
Usage#
- First, follow the official tutorial，choose the correct version for your computer and deploy InvokeAI locally
- Go to the InvokeAI installed folder，execute
./invoke.sh，Select 2 to open the prompted URL and get a WebUI
- Try inputting "a girl with long hair" in the Prompt text box in the top left. Then adjust the parameters below and click "Invoke" to generate an image
We can click "Invoke" multiple times to generate multiple images. We can also adjust the Prompt to generate other content.
- The default Stable Diffusion model may not meet everyone's needs. In that case, you can adjust the model as needed
You can go to https://civitai.com to see artwork generated by others.
- We can click on a artwork we like to go to the model details, then click "Download" to download the model.
- Place the downloaded model in your local InvokeAI models folder (your-invokeai-path/models), and load the model through the WebUI.
Input the model path, search for the model, select the model, and add it.
- Go back to civitai, select an artwork you like from the model, and the artwork information will pop up.
Below there will be some parameters for the artwork such as Prompt, configurations, etc.
- Go back to the WebUI, switch the model, and configure it as needed.
- Iteratively refine parameters and invoke image generation multiple times to obtain high-quality samples for the overview section.
Summary#
Stable Diffusion allows styling through model adjustments. Works generated from different prompts also vary.
Using Stable Diffusion is like conducting experiments, requiring control over different variables to achieve aesthetic results.
In addition to InvokeAI, you can also use other WebUIs for AIGC. The operations are largely similar, such as:
- https://github.com/AUTOMATIC1111/stable-diffusion-webui （MacOS / Win）
- https://diffusionbee.com/ （MacOS ）
- https://drawthings.ai （MacOS）
More information#
Learn Stable Diffusion
Prompt