foreword
This article uses the AutoDL platform for Stable Diffusion web UI cloud deployment
AutoDL official website: AutoDL Computing Cloud | Flexible, easy to use, and cost-effective. Rent a GPU and use AutoDL
Stable Diffusion web UI 官网：AUTOMATIC1111/stable-diffusion-webui: Stable Diffusion web UI (github.com)
step
Leased instance
Select the community mirror AUTOMATIC1111/stable-diffusion-webui/NovelAI-Consolidation-Package-3.1, and test the RTX 3080 with the v12 version available
startup steps
Click on JupyterLab
Execute according to the instructions in the figure, after "moving complete" appears, click F5 to refresh the browser (very important)
After switching the environment
Then follow the execution of the second code block, and then perform the following operations (each operation waits for the success of the previous step before proceeding to the next step)
Step 1: Click the button "Click me to automatically accelerate academic acceleration", after the acceleration is successful, it will be displayed as "Acceleration successful, click again to cancel"
Step 2: Click the button "Click me to install the downloader", the installation process will execute a series of codes, wait for a while, after the installation is successful, the button will be displayed as "The downloader has been successfully installed"
Step 3: Click the button "Click me to move to the data disk", after clicking, it will display moving..., after a while, the button will display "Moved to the data disk"
When we see that these 3 buttons have turned green, it means that the operation is successful.
Click "Start WebUI", here we introduce from top to bottom
1. Username and password: After enabling Academic Acceleration, you will need to enter a username and password when accessing the WebUI. This is not a mandatory item, and it will be randomly generated if you don’t fill it in. You can fill in a simple username and password (for example: 123, 123456, etc. of)
2. Operation directory: The data disk should be selected by default here, and there is no need to modify it.
3. Operation mode: choose the normal version here, and the other two versions will be more difficult to operate
4. Enable parameters: It involves the functions in WebUI, 3 are checked by default, according to my usage suggestions, check "Picture Reverse Push Text", "xformers greatly improves memory consumption and speed", "Allow WebUI to use installation extensions "Function" these 3 are enough.
The configuration is as follows
We click the "Run WebUI" button, the running code will automatically start executing, and finally click the custom service
Use the SSH Tunnel Broker documentation at this time
Native Windows PowerShell execution
ssh -CNg -L 6006:127.0.0.1:6006 [email protected] -p 42151
Among them, [email protected] and 42151 are the access address and port of the SSH command in the instance respectively. Please find the ssh command of your own instance and replace it accordingly. 6006:127.0.0.1:6006 refers to the port 6006 in the proxy instance to the local port 6006.
Visit http://127.0.0.1:6006 in the local browser to open the service. Note that the port 6006 here should be consistent with the port in 6006:127.0.0.1:6006 above.
Let's take this as an example to simply generate a dog
The blogger will not repeat the other functions here, and leave it to the readers to explore by themselves
reference article
Deploying stable-diffusion-webui locally allows more expansion of AI painting - Minority (sspai.com)
Related Links
This article is published by OpenWrite, a multi-post platform for blogging !