Nodes Browser

ComfyDeploy: How Together Vision Node works in ComfyUI?

What is Together Vision Node?

A custom ComfyUI node for generating AI-powered image descriptions using Together AI's Vision models (both free and paid versions). Features include customizable prompts, advanced generation parameters, and robust image handling with comprehensive error management.

How to install it in ComfyDeploy?

Head over to the machine page

  1. Click on the "Create a new machine" button
  2. Select the Edit build steps
  3. Add a new step -> Custom Node
  4. Search for Together Vision Node and select it
  5. Close the build step dialig and then click on the "Save" button to rebuild the machine

ComfyUI-TogetherVision

A custom node for ComfyUI that enables image description using Together AI's Vision models. This node allows you to generate detailed descriptions of images using either the paid or free version of Together AI's Llama Vision models.

Together Vision Node

Features

🖼️ Image Description & Text Generation:

  • Generate detailed descriptions of images using state-of-the-art vision models
  • Toggle vision processing on/off for flexible usage
  • Use as a text-only LLM when vision is disabled

🤖 Multiple Models:

  • Paid Version: Llama-3.2-11B-Vision-Instruct-Turbo
  • Free Version: Llama-Vision-Free

⚙️ Customizable Parameters:

  • Temperature control
  • Top P sampling
  • Top K sampling
  • Repetition penalty

🔑 Flexible API Key Management:

  • Direct input in the node
  • Environment variable through .env file

📝 Custom Prompting:

  • System prompt customization
  • User prompt customization

Getting Started

1. Get Together AI API Key

  1. Go to Together AI API Settings
  2. Sign up or log in to your Together AI account
  3. Click "Create API Key"
  4. Copy your API key for later use

2. Installation

  1. Clone this repository into your ComfyUI custom_nodes directory:
cd ComfyUI/custom_nodes
git clone https://github.com/mithamunda/ComfyUI-TogetherVision.git
  1. Restart ComfyUI - it will automatically install the required dependencies from requirements.txt

  2. Set up your Together AI API key using one of these methods:

    • Option 1: Create a .env file in the node directory:
      TOGETHER_API_KEY=your_api_key_here
      
    • Option 2: Input your API key directly in the node

Usage

  1. Add the "Together Vision 🔍" node to your workflow
  2. Configure Vision Mode:
    • Enable Vision (Default): Connect an image output to the node's image input
    • Disable Vision: Skip image input for text-only generation
  3. Select your preferred model (Paid or Free)
  4. Configure the parameters:
    • Temperature (0.0 - 2.0)
    • Top P (0.0 - 1.0)
    • Top K (1 - 100)
    • Repetition Penalty (0.0 - 2.0)
  5. Customize the prompts:
    • System prompt: Sets the behavior of the AI
    • User prompt: Specific instructions for image description or text generation

Parameters

| Parameter | Description | Default | Range | |-----------|-------------|---------|--------| | Vision Enable | Toggles vision processing | True | True/False | | Temperature | Controls randomness | 0.7 | 0.0 - 2.0 | | Top P | Nucleus sampling | 0.7 | 0.0 - 1.0 | | Top K | Top K sampling | 50 | 1 - 100 | | Repetition Penalty | Prevents repetition | 1.0 | 0.0 - 2.0 |

Image Resolution Limits

The node automatically handles high-resolution images:

  • Images larger than 2048x2048 pixels will be automatically resized
  • Aspect ratio is preserved during resizing
  • High-quality LANCZOS resampling is used

For best results:

  1. Keep image dimensions under 2048 pixels
  2. Use ComfyUI's built-in resize nodes before this node
  3. For very large images, consider splitting them into sections

Rate Limits

Free Model (Llama-Vision-Free)

  • Limited to approximately 100 requests per day
  • Rate limit resets every 24 hours
  • Hourly limits may apply (typically 20-30 requests per hour)

Paid Model (Llama-3.2-11B-Vision)

  • Higher rate limits based on your Together AI subscription
  • Better performance and reliability
  • Priority API access

Handling Rate Limits

When you hit a rate limit:

  1. Wait for the specified time (usually 1 hour for hourly limits)
  2. Switch to a different Together AI account
  3. Upgrade to the paid model for higher limits
  4. Consider batching your requests during off-peak hours

Tips to Avoid Rate Limits

  1. Cache results for repeated images
  2. Use the paid model for production workloads
  3. Monitor your API usage through Together AI dashboard
  4. Space out your requests when possible

Operating Modes

Vision Mode (Default)

  • Requires connected image input
  • Generates detailed image descriptions
  • Full vision + language capabilities

Text-Only Mode

  • No image input required
  • Functions as a standard LLM
  • Useful for text generation and chat

Error Handling

The node includes comprehensive error handling and logging:

  • API key validation
  • Rate limit notifications
  • Image processing errors
  • API response errors
  • Vision mode validation

Examples

Here are some example prompts you can try:

  1. Vision Mode - Detailed Description:
Describe this image in detail, including colors, objects, and composition.
  1. Vision Mode - Technical Analysis:
Analyze this image from a technical perspective, including lighting, composition, and photographic techniques.
  1. Text-Only Mode - Creative Writing:
Write a creative story about a magical forest.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

  • Together AI for providing the Vision API
  • ComfyUI community for the framework and support

Support

If you encounter any issues or have questions:

  1. Check the error logs in ComfyUI
  2. Ensure your API key is valid
  3. Check Together AI's service status
  4. Open an issue on GitHub

Note: This node requires a Together AI account and API key. You can get one at Together AI's website.

Updated README.md to reflect automatic mode switching based on image connection

The node now automatically switches between Vision Mode and Text-Only Mode based on the presence of an image input connection. When an image is connected, the node will generate detailed image descriptions. When no image is connected, the node will function as a text generation model.

Flexible Processing Modes

  • Image + Text Mode: When an image is connected, generates descriptions and responses about the image
  • Text-Only Mode: When no image is connected, functions as a text generation model
  • Seamlessly switches between modes based on input connections