Interface: ComfyUIImageModelOptions

Defined in: src/comfyui/ComfyUIImageModel.ts:11

Properties

apiHost

> apiHost: string

Defined in: src/comfyui/ComfyUIImageModel.ts:15

The ComfyUI server address, e.g. "127.0.0.1:8188" or "my-comfyui.example.com".

promptInputName?

> optional promptInputName?: string

Defined in: src/comfyui/ComfyUIImageModel.ts:37

The name of the input on promptNodeId that holds the prompt text.

Default

"text"

promptNodeId

> promptNodeId: string

Defined in: src/comfyui/ComfyUIImageModel.ts:32

The ID of the node (as it appears in workflow) whose input holds the positive prompt text (typically a CLIPTextEncode node).

ssl?

> optional ssl?: boolean

Defined in: src/comfyui/ComfyUIImageModel.ts:20

Whether to connect over SSL/WSS.

Default

false

timeoutMs?

> optional timeoutMs?: number

Defined in: src/comfyui/ComfyUIImageModel.ts:42

Milliseconds to wait for the workflow to finish before giving up.

Default

300000 (5 minutes)

workflow

> workflow: Record<string, unknown>

Defined in: src/comfyui/ComfyUIImageModel.ts:27

The ComfyUI workflow to run, exported in API format (ComfyUI menu: "Save (API Format)"). Pixi'VN AI has no way to know how your workflow is structured (checkpoint, sampler, custom nodes, ...), so you provide it, along with where to inject the prompt via promptNodeId/promptInputName.