Deploy Stable Diffusion WebUI on Alibaba Cloud PAI‑EAS in Minutes
This guide walks you through using Alibaba Cloud's PAI‑EAS to quickly deploy the open‑source Stable Diffusion text‑to‑image model with a low‑code WebUI, covering preparation, service configuration, inference testing, and common troubleshooting tips for performance and file management.
AIGC (AI‑Generated Content) refers to content created automatically by AI. Stable Diffusion is a popular open‑source cross‑modal model that generates images from text prompts, and its WebUI offers a low‑code interactive interface.
Preparation
Ensure you have an active PAI‑EAS account and a default workspace.
New users can claim a free trial of PAI‑EAS as described in the "Free Experience" section.
Deploy the EAS Service
Follow these steps to launch Stable Diffusion as an AI‑Web application on PAI‑EAS:
Log in to the PAI console and navigate to the workspace where you want to deploy.
Open Model Deployment → Model Online Service (EAS) and click Deploy Service .
Configure the service parameters:
Service Name : e.g., sdwebui_demo Deployment Method : select Image Deployment for AI‑Web Application .
Image Selection : choose stable-diffusion-webui version 2.2.
Resource Group : use a public resource group.
Resource Configuration : select a regular GPU configuration, recommended ml.gu7i.c8m30.1‑gu30.
After confirming the configuration, click Deploy . When the service status changes to "Running", the deployment is successful.
Start the WebUI for Inference
In the service list, click View Web Application to open the WebUI. On the "Text‑to‑Image" tab, enter a prompt (e.g., "cute dog") and click Generate to see the generated image.
Common Issues and Solutions
4.1 Image Generation Is Slow
Enable Blade or xFormers to accelerate the model:
Blade (provided by PAI) can speed up generation up to 3.06× depending on image size and steps.
xFormers is an open‑source acceleration tool bundled with SD‑WebUI.
To enable Blade, add --blade to the run command and redeploy. To enable xFormers, add --xformers similarly.
4.2 Mounting Custom Models and Output Directories
Create an OSS bucket and empty folder (e.g., oss://bucket-test/data-oss/). In the service configuration, set the model mount to this OSS path and map it to /code/stable-diffusion-webui/data-oss. Add --data-dir data-oss to the run command, then redeploy.
4.3 Managing the File System
Add the --filebrowser flag to the run command (e.g.,
./webui.sh --listen --port=8000 --blade --data-dir data-oss --filebrowser) to enable the FileBrowser tab in the WebUI, allowing direct upload and download of files.
Free Trial
Since April 2023, Alibaba Cloud offers a free trial of PAI‑EAS for verified new users. Access the trial via the Alibaba Cloud trial center, select "AI & Machine Learning" → "Machine Learning Platform PAI" → "Model Online Service PAI‑EAS".
Alibaba Cloud Big Data AI Platform
The Alibaba Cloud Big Data AI Platform builds on Alibaba’s leading cloud infrastructure, big‑data and AI engineering capabilities, scenario algorithms, and extensive industry experience to offer enterprises and developers a one‑stop, cloud‑native big‑data and AI capability suite. It boosts AI development efficiency, enables large‑scale AI deployment across industries, and drives business value.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
