node-red-contrib-replicate

A Node-RED node for generating images using the Replicate API. This node supports various Replicate models, including standard base models and LoRA fine-tuned models.

Prerequisites

• Node-RED installation
• Replicate API key (get one at replicate.com)

Installation

Manual Installation

1. Download the Code:

   • Clone the repository or download the ZIP file from GitHub.

   git clone https://github.com/ashakoen/node-red-contrib-replicate.git
   

2. Navigate to the Node-RED User Directory:

   • Typically, this is located at ~/.node-red.

   cd ~/.node-red
   

3. Copy the Node Files:

   • Copy the downloaded files into the ~/.node-red directory.

4. Install Dependencies:

   • Run the following command to install the necessary dependencies.

   npm install ./node-red-contrib-replicate
   

5. Restart Node-RED:

   • Restart your Node-RED instance to load the new node.

Features

• Support for standard Replicate models and LoRA fine-tuned models
• Automatic parameter validation and adjustment
• Supports image inputs via URL, file path, or base64 data
• Comprehensive error handling and status feedback
• Configurable output formats (webp, jpg, png)
• Support for various model-specific parameters
• Include config node for secure storage of Replicate API key
• Image preview functionality inspired by node-red-contrib-image-output by @rikukissa

Usage

1. Add your Replicate API key in the configuration node.
2. Choose your model type (Standard or LoRA).
3. Configure model-specific parameters.
4. Send input via msg.payload.

Input Parameters

The node accepts parameters via the node configuration or msg.payload. Parameters in msg.payload override node configuration.

Common parameters include:

• prompt - Text description of the desired image
• num_outputs - Number of images to generate (1-4)
• num_inference_steps - Number of denoising steps
• guidance_scale - How closely to follow the prompt (0-10)
• aspect_ratio - Image aspect ratio (e.g., "1:1", "16:9")
• output_format - Image format ("webp", "jpg", "png")
• seed - Random seed for reproducible results

Output

The node outputs a msg.payload object containing:

• output - Array of generated image URLs
• original_output - Original API response
• logs - Generation process logs
• metrics - Performance metrics
• id - Prediction ID
• status - Final status
• created_at - Creation timestamp
• completed_at - Completion timestamp

Example Flow

[
  {
    "id": "1ad8eec921654237",
    "type": "tab",
    "label": "Replicate",
    "disabled": false,
    "info": "",
    "env": []
  },
  {
    "id": "96bd7d88790b2536",
    "type": "inject",
    "z": "1ad8eec921654237",
    "name": "",
    "props": [
      {
        "p": "payload"
      }
    ],
    "repeat": "",
    "crontab": "",
    "once": false,
    "onceDelay": 0.1,
    "topic": "",
    "payload": "{\"prompt\":\"a fast racecar\",\"num_outputs\":1,\"guidance_scale\":5,\"num_inference_steps\":35,\"model\":\"black-forest-labs/flux-dev\",\"useLoRAModel\":false,\"image\":\"\",\"mask\":\"\",\"prompt_strength\":0.8,\"format\":\"png\",\"aspect_ratio\":\"4:5\"}",
    "payloadType": "json",
    "x": 650,
    "y": 380,
    "wires": [
      [
        "34302b79787f7c6e"
      ]
    ]
  },
  {
    "id": "34302b79787f7c6e",
    "type": "replicate-node",
    "z": "1ad8eec921654237",
    "name": "",
    "replicateConfig": "de8b4314b35b9d1e",
    "modelType": "standard",
    "prompt": "",
    "model": "",
    "version": "",
    "showPreview": true,
    "previewWidth": "200",
    "x": 820,
    "y": 380,
    "wires": [
      []
    ]
  },
  {
    "id": "de8b4314b35b9d1e",
    "type": "replicate-config"
  }
]

Supported Models

Standard Models

• FLUX Dev
• FLUX Pro
• FLUX Pro Ultra
• FLUX Schnell
• Other Replicate-compatible models

LoRA Models

• Any LoRA fine-tuned model on Replicate

Error Handling

The node provides detailed error messages and status updates for:

• Missing required parameters
• Invalid parameter values
• API errors
• Network issues
• Image processing errors

License

MIT

Support

For issues and feature requests, please use the GitHub issue tracker.

Contributing

Contributions are welcome! Please submit pull requests or open issues for any improvements.