ComfyUI S3 Connector

A ComfyUI custom node pack for seamless image uploading and downloading between local workflows and S3 buckets. Supports AWS S3 and S3-compatible services (MinIO, DigitalOcean Spaces, etc.).

Installation

1. Clone this repository into your ComfyUI custom_nodes folder:

   cd ComfyUI/custom_nodes
   git clone https://github.com/your-repo/comfyui-s3-connector.git
   

2. Install dependencies:

   cd comfyui-s3-connector
   pip install -r requirements.txt
   

3. Configure S3 credentials:

   cp .env.example .env
   

   Edit .env and fill in your S3 credentials.

4. Restart ComfyUI

Configuration

Copy .env.example to .env and configure the following:

Variable	Required	Description
S3_ACCESS_KEY_ID	Yes	Your AWS/S3 access key
S3_SECRET_ACCESS_KEY	Yes	Your AWS/S3 secret key
S3_BUCKET_NAME	Yes	Target bucket name
S3_REGION	No	AWS region (default: us-east-1)
S3_ENDPOINT_URL	No	Custom endpoint for S3-compatible services
S3_PREFIX	No	Prefix path for all uploads

Nodes

S3 Upload Image

Uploads images from your ComfyUI workflow to an S3 bucket.

Inputs:

• images (IMAGE): Image tensor from ComfyUI workflow
• folder_path (STRING): S3 folder path to upload to
• file_name (STRING): File name to save as

Outputs:

• s3_url (STRING): Full URL to the uploaded image
• s3_key (STRING): S3 key/path of the uploaded image

S3 Load Image

Loads images from an S3 bucket into your ComfyUI workflow.

Inputs:

• folder_path (STRING): S3 folder path (e.g., uploaded/)
• file_name (STRING): File name to load (e.g., image.png)

Outputs:

• image (IMAGE): Image tensor for use in ComfyUI workflow
• mask (MASK): Alpha mask (if image has transparency)

S3 Upload Image (Full Path)

Uploads images using a single full path string instead of separate folder and filename.

Inputs:

• images (IMAGE): Image tensor from ComfyUI workflow
• full_path (STRING): Complete S3 path including filename (e.g., folder/subfolder/image.png)

Outputs:

• s3_url (STRING): Full URL to the uploaded image
• path (STRING): S3 key/path of the uploaded image

S3 Load Image (Full Path)

Loads images using a single full path string.

Inputs:

• full_path (STRING): Complete S3 path including filename

Outputs:

• image (IMAGE): Image tensor for use in ComfyUI workflow
• mask (MASK): Alpha mask (if image has transparency)

Get Job ID

Returns the unique job/prompt ID for the currently running ComfyUI workflow. Useful for tracking outputs, organizing files by job, or logging.

Inputs:

• None (uses hidden ComfyUI context)

Outputs:

• job_id (STRING): The unique identifier for the current job/prompt execution

Notes:

• The node automatically refreshes on each queue to ensure you get the current job ID
• Can be connected to filename inputs to organize S3 uploads by job

Usage Examples

Upload workflow output to S3

1. Connect your image output to the S3 Upload Image node
2. Set a filename prefix (e.g., "generated_art")
3. Optionally set a custom path (e.g., "portraits/2024")
4. Run the workflow - images will be uploaded to your S3 bucket

Load image from S3 for img2img

1. Add an S3 Load Image node
2. Enter the S3 key (e.g., "images/reference.png")
3. Connect the output to your workflow (KSampler, ControlNet, etc.)

S3-Compatible Services

This plugin works with any S3-compatible storage service:

• AWS S3: Leave S3_ENDPOINT_URL empty
• MinIO: Set S3_ENDPOINT_URL=http://localhost:9000
• DigitalOcean Spaces: Set S3_ENDPOINT_URL=https://nyc3.digitaloceanspaces.com
• Cloudflare R2: Set S3_ENDPOINT_URL=https://<account-id>.r2.cloudflarestorage.com

License

MIT License - See LICENSE file for details.