Updated README.md
This commit is contained in:
@@ -1,2 +1,51 @@
|
||||
# ComfyUI-SendToDiscord
|
||||
A custom node for ComfyUI that allows sending preview images directly to Discord via webhooks, with support for batch mode and metadata.
|
||||
|
||||
ComfyUI-SendToDiscord is a custom node for [ComfyUI](https://github.com/comfyanonymous/ComfyUI) that simplifies sending preview images to Discord via webhooks. It supports both single-image uploads and batch mode, making it an efficient tool for sharing your generated images directly with your Discord server.
|
||||
|
||||
## Features
|
||||
- Send individual or batches of images directly to a Discord channel via webhooks.
|
||||
- Includes metadata (e.g., prompts and additional PNG information) in saved images.
|
||||
- Configurable batch size and compression level.
|
||||
- Simple setup with a `config.ini` file for webhook URL configuration.
|
||||
|
||||
## Requirements
|
||||
- Python 3.10+
|
||||
- `Pillow`, `requests`, and `numpy` libraries (automatically installed via `requirements.txt`).
|
||||
|
||||
## Installation
|
||||
1. Clone the repository into your ComfyUI custom nodes directory:
|
||||
```
|
||||
git clone https://github.com/yourusername/ComfyUI-SendToDiscord.git
|
||||
```
|
||||
2. Install dependencies:
|
||||
```
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
3. Add your Discord webhook URL to the config.ini file located in the root directory:
|
||||
```
|
||||
[Discord]
|
||||
webhook_url = https://discord.com/api/webhooks/your-webhook-id
|
||||
```
|
||||
Usage
|
||||
|
||||
Open ComfyUI.
|
||||
Add the "Preview Image (with Discord option)" node to your workflow.
|
||||
Configure the node parameters:
|
||||
Send to Discord: Enable or disable image uploads to Discord.
|
||||
Batch Mode: Enable to accumulate images and send them in a single batch.
|
||||
Generate your images, and they will be uploaded automatically to the specified Discord channel.
|
||||
|
||||
Configuration
|
||||
|
||||
The node uses the following parameters:
|
||||
|
||||
images: The list of images to process.
|
||||
send_to_discord: Enable or disable sending images to Discord (enable/disable).
|
||||
batch_mode: Accumulate images and send them as a batch (enable/disable).
|
||||
|
||||
Example Workflow
|
||||
|
||||
You can add this node after your image generation process to preview and share your results directly.
|
||||
License
|
||||
|
||||
This project is licensed under the MIT License.
|
||||
Reference in New Issue
Block a user