Platform For AI:Deploy AI video generation application in EAS

Background information

As generative AI gains more attention, AI-powered video generation has become a popular application industry. There are lots of open source foundation models for video generation in the market. You can select a foundation model based on the model performance and the specific scenario. ComfyUI, a node-based web UI tool for generative AI, divides AI-powered content generation into worker nodes to achieve precise workflow customization and reproducibility. This topic describes how to deploy an AI video generation application and related inference services in the following steps:

1. Deploy model service in EAS

   Deploy an AI video generation application in EAS.

2. Use ComfyUI to perform model inference

   After you start ComfyUI, you can perform model inference on a web UI page that generates images and videos based on text prompts.

3. FAQ

   This section describes how to load other open source models or custom models that you trained, and perform model inference.

Prerequisites

• EAS is activated and the default workspace is created. For more information, see Activate PAI and create the default workspace.

• If you use a RAM user to deploy the model, make sure that the RAM user is granted the management permissions on EAS. For more information, see Grant the permissions that are required to use EAS.

Limits

This feature is available only in the China (Hangzhou) and Singapore regions.

Deploy model service in EAS

1. Go to the EAS-Online Model Services page.

   1. Log on to the PAI console.

   2. In the left-side navigation pane, click Workspaces. On the Workspaces page, click the name of the workspace to which the model service that you want to manage belongs.

   3. In the left-side navigation pane, choose Model Deployment>Elastic Algorithm Service (EAS) to go to the EAS-Online Model Services page.

2. On the PAI-EAS Model Online Service page, click Deploy Service. In the dialog box that appears, select Custom Deployment and click OK.

3. On the Create Service page, configure the required parameters. The following table describes key parameters.

   Parameter	Description
   Service Name	The name of the service. The name comfyui_svd_demo is used in this example.
   Deployment Method	Select Deploy Web App by Using Image.
   Select Image	Click PAI Image. Select comfyui from the image drop-down list, and select 0.1 from the image version drop-down list. Note You can select the latest version of the image when you deploy the model service.
   Command to Run	After you configure the image version, the system automatically sets the parameter to python main.py --listen --port 8000. Port number: 8000.
   Resource Group Type	Select Public Resource Group.
   Resource Configuration Mode	Select General.
   Resource Configuration	Select an Instance Type on the GPU tab. In terms of cost-effectiveness, we recommend that you use the ml.gu7i.c16m60.1-gu30 instance type. If the resources are insufficient, you can also select the ecs.gn6i-c16g1.4xlarge instance type.

4. Click Deploy. The deployment requires approximately five minutes to complete.

   When the Model Status changes to Running, the service is deployed.

Use ComfyUI to perform model inference

1. Find the service that you want to manage and click View Web App in the Service Type column.

2. Perform model inference on the web UI page.

   Select a model for text-to-image generation and a model for image-to-video generation based on your business requirements. This example uses the default settings. Enter text prompts in the CLIP Text Encode (Prompt) section, for example, Rocket takes off from the ground, fire, sky, airplane and click Queue Prompt. The system starts to run the workflow and generate the video.

3. Right-click the generated video and select Save Image to save the generated video to your on-premises machine.

   The following section provides an example of the generated video:

FAQ

How do I mount a custom model?

Assume that you obtained a model such as SDXL, LoRA, or SVD from the open-source community, or you have a custom model that you trained, if you want to save the generated model files to your Object Storage Service (OSS) bucket directory and load the model files by mounting the file, perform the following steps:

1. Log on to the OSS console. Create a bucket and an empty directory.

   Example: oss://bucket-test/data-oss/, where: bucket-test is the name of the OSS bucket, and data-oss is the empty file directory in the bucket. For more information about how to create a bucket, see Create buckets. For more information about how to create an empty directory, see Manage directories.

2. On the Elastic Algorithm Service (EAS) page, find the service for which you want to add a version and click Update Service in the Actions column.

3. In the Model Service Information section, configure the following parameters.

   Parameter	Description
   Model Settings	Click Specify Model Settings to configure the model. Select OSS Mount and set OSS Path to the path of the OSS bucket that you created in Step 1. Example: oss://bucket-test/data-oss/. Mount Path: Mount the OSS file directory to the /code/data-oss path of the image. Example: /code/data-oss. Enable Read-only Mode: turn off the read-only mode.
   Command to Run	Add the --data-dir mount directory in the Command to Run field, where mount directory must be the same as the Mount Path in the Model Settings section. Example: python main.py --listen --port 8000 --data-dir /code/data-oss.

4. Click Update to update the model service.

   The following figure shows the directory that PAI automatically creates in the empty OSS directory you specified and copies the required data to the directory. We recommend that you upload data to the specified directory after the service is started.

5. Upload the on-premises model file to the OSS path ~/models/checkpoints/ that is generated in Step 4. For more information, see the "Upload an object" section in the Get started by using the OSS console topic.

6. Load the model and perform model inference.

   1. On the Elastic Algorithm Service (EAS) page, find the service that you want to manage and choose > Restart Service in the Actions column.

   2. After the service is started, click View Web App in the Service Type column.

   3. On the web UI page, in the Load Checkpoint drop-down list, select the model that you want to use. Perform model inference based on the procedure described in the Use ComfyUI to perform model inference section of this topic.

References

For more information about EAS, see EAS overview.