ComfyOne - Capture One Integration with ComfyUI API

This script, ComfyOne, integrates the ComfyUI API directly within Capture One, offering extensive control over AI-powered image generation through a highly customizable interface. ComfyOne provides control over prompts, workflows, resolution, dual ControlNet support, IP-Adapter, and inpainting settings—all managed within Capture One.

Work In Progress

Key Features

• Server Configuration: Easily connect to the ComfyUI server by specifying its URL, with a default of 127.0.0.1:8188.
• Workflow Management: Dynamically loads available workflows from a specified folder, allowing users to select and execute a workflow from within Capture One.
• Prompting System: Define positive and negative prompts to guide content in the generated image and set seed values to control randomness.
• Resolution and Denoising: Choose from predefined resolution settings (Low, Mid, High) and set denoising levels to refine image clarity.
• Masking Options: Configure mask prompts to target specific areas of the image with options for blur, expand, invert, or fill.
• Dual ControlNet Integration: Supports two ControlNets (ControlNet A and B) simultaneously, each with customizable parameters and custom reference images.
• IP-Adapter Functionality: Customize the IP-Adapter for stylistic refinement with separate reference images.
• Inpainting Mode: Inpainting settings enable targeted editing with blur and blending options for masks.

Node Naming and Automated Workflow Management

To fully automate workflow management through ComfyOne, users need to follow a naming convention for nodes within the JSON workflow. By adhering to the ComfyUI API JSON structure, a separate Python script can dynamically manage parameters based on these identifiers.

Node Identifiers for Automation

• Prompt Control:
  • PositivePromptAPI
  • NegativePromptAPI
• Image Generation:
  • SeedAPI
  • DenoiseAPI
  • SrcImgAPI
  • SaveImgAPI
• Advanced Settings:
  • turboOnAPI
  • inpaintOnAPI
  • resolutionAPI
• Masking and Inpainting:
  • maskAPI
  • inpaintAPI
• ControlNet Nodes:
  • ctrlNetAAPI
  • ctrlNetBAPI
  • uploadImgCtrlNetA
  • uploadImgCtrlNetB
• IP-Adapter Node:
  • ipAdapterAPI
  • uploadImgipAdapter

Workflow Tree with Adjustable Variables per Node

• PositivePromptAPI: Adjust the positive prompt for image generation.
• NegativePromptAPI: Adjust the negative prompt to filter out unwanted elements.
• SeedAPI: Set the seed value to control randomness.
• DenoiseAPI: Adjust the denoising level.
• SrcImgAPI: Source image input path.
• SaveImgAPI: Output image save path.
• turboOnAPI: Boolean for Turbo mode.
• inpaintOnAPI: Boolean for inpainting mode.
• resolutionAPI: Set the resolution (Low, Mid, High).
• maskAPI: Configure mask settings.
• inpaintAPI: Control blur and blend for inpainting.
• ctrlNetAAPI / ctrlNetBAPI: ControlNet A and B settings.
• uploadImgCtrlNetA / uploadImgCtrlNetB: Upload custom images for ControlNet.
• ipAdapterAPI: IP-Adapter settings.
• uploadImgipAdapter: Reference image upload for IP-Adapter.

Script Execution and Generation from Capture One

A secondary script enables direct image generation from the Capture One window using saved parameters, allowing quick iterations without reconfiguring settings.

Installation Script

ComfyOne includes an installation script that sets up a virtual environment (venv) and installs necessary libraries, ensuring proper configuration.

User Interface and Configuration

Using AppleScript’s Dialog Toolkit Plus and PrefsStorageLib, this script provides a user-friendly interface within Capture One for managing inputs and saving preferences.

Workflow Execution

Once configured, users can run the script to:

• Apply chosen parameters to the diffusion model on the ComfyUI server.
• Save configurations for repeatability.

ComfyOne empowers professionals in image generation with ControlNet for compositional guidance, IP-Adapter for styling, and complete node-based automation using the ComfyUI API JSON format.