This topic describes how to use the ComfyUI, Kohya, and Stable Diffusion WebUI tools.
Log on to the PAI ArtLab console.
Tool types and versions
The Toolbox includes the following tools:
-
Stable Diffusion WebUI: Supports features such as text-to-image, image-to-image, and animated image generation. For more information, see PAI ArtLab Stable Diffusion WebUI.
-
Kohya: Supports features such as dataset image annotation, batch generation, and model fine-tuning. For more information, see PAI ArtLab Kohya.
-
ComfyUI: Allows you to build node-based workflows. For more information, see PAI ArtLab ComfyUI.
The differences between the versions are as follows:
|
Difference |
Shared edition |
Exclusive edition |
|
Resource difference |
Shared compute resources across multiple users. |
Dedicated compute resources for exclusive use. |
|
Billing difference |
No charge for tool startup. You are billed only for actual GPU time consumed during image generation tasks. For example, if you start a tool, initiate generation on the tool's WebUI, and an image renders in 10 seconds, you are charged for 10 seconds of GPU usage. |
Billing begins immediately after successful tool launch and continues until the tool is stopped. You are billed hourly based on the instance type selected at startup. For example, billing starts when the tool's WebUI becomes accessible. If the tool remains active for 20 minutes, you are billed for 20 minutes regardless of actual usage. Closing the browser tab terminates the tool and stops billing. |
|
Difference in custom installation of models and node plug-ins |
Does not support custom plug-in installation. Only essential pre-installed plug-ins are available, including text-to-image, image-to-image, and ControlNet. |
Supports custom installation of models, nodes, and plug-ins. For more information, see Upload a local model to the PAI ArtLab platform. |
|
Difference in generation task queuing |
Queuing may occur during peak hours. |
No queuing. |