Skip to content

ComfyUI to Vix Workflows Migration

If you want to adopt your ComfyUI workflow to use in Visionatrix, you can use this guide to help you do so. There are a few steps you need to follow.

1. Install ComfyUI-Visionatrix custom nodes

First, it is recommended to install our custom ComfyUI-Visionatrix nodes. Otherwise, you will have to use custom nodes titles which will be parsed by Visionatrix.

git clone https://github.com/Visionatrix/ComfyUI-Visionatrix.git

Note

You can do the required migration via nodes titles, which is less convenient. The node title must be like this: input;Display Name;optional;advanced;order=1;custom_id=custom_name.

2. Define the input params

Visionatrix UI aims simplicity and clarity. Define the most important input params of your ComfyUI workflow to extract them to the Visionatrix UI as inputs, for example:

  • prompt (textarea)
  • negative prompt (textarea)
  • prompt strength (range)
  • some logic toggles (checkbox)
  • input files (file)

For that you will need to attach our custom nodes as adapters to your nodes receiving these inputs that will be filled by the user from the Visionatrix UI.

As example, you can have a look at our list of worklows adopted to the new format.

Note

The list of available nodes can be found in the readme of the ComfyUI-Visionatrix repository.

2.1 Node to Input mapping via title string

Alternatively, Visionatrix supports other Nodes mapping as an input param via node title string separated by semicolon.

The nodes titles starting with input; keyword are considered input parameters to Visionatrix.

The parameters list:

  • input - keyword to define the input param
  • Display Name - positional parameter, the name of the input field displayed in the UI
  • optional - if present, the optional field is set to True
  • advanced - if present, the advanced field is set to True
  • order=1 - the order of the input param in the UI
  • custom_id=custom_name - the custom id of the input param

2.2 Defining file inputs

  • LoadImage - default ComfyUI image loader node as image file input field. As required title: input;Input image;order=1, or optional advanced: input;Optional helper image;optional;advanced;order=20;

    Note

    We recommend to always define custom_id value as it greatly helps to have a constant name for input parameter for API. Like: input;Person's face;order=1;custom_id=person_face

  • You can make a mask from LoadImage using mask word. When you do so you need to specify source_input_name which point to custom_id of input for which this mask belongs.

    Full input for inpainting and mask definition will be:

    `input;Image to redraw;order=1;custom_id=source_image;`
`input;Mask to redraw;order=2;custom_id=mask_redraw;mask;source_input_name=source_image`

3. Map the models for automatic download

Visionatrix simplifies and automates the process of downloading the models.

You need to ensure that models used in workflow are known to Visionatrix, see models catalog

4. Build the list of available flows

The last step is to build the list of available flows in the Visionatrix UI. Follow the steps described in options.py file for FLOWS_URL and MODELS_CATALOG_URL to enable Visionatrix local workflows development mode:

Create a zip with adjusted/new flows:

cd ../VixFlowsDocs && zip -r ../Visionatrix/flows.zip flows && cd ../Visionatrix

And uncomment appropriate code lines in options.py file to use local versions of the flows.

5. Verify and test the workflow

Last step is to run Visionatrix and set up your workflow to verify that everything works as expected.