diff --git a/README.md b/README.md index 1a04d8c..80d40fa 100644 --- a/README.md +++ b/README.md @@ -37,6 +37,12 @@ English | [简体中文](README_CN.md) - **1080p resolution**: Only supports 8-second duration for Veo 3.1 models - **First-last frame interpolation**: Only available for Veo 3.1 models with 8-second duration +### Watermark Removal +- Automatically detects and removes Gemini watermarks from generated images +- Supports both 48×48 and 96×96 watermark sizes +- Uses reverse alpha blending algorithm for clean removal +- Simple one-click operation - just pass the image through the node + ## Installation 1. Clone or download this repository to your ComfyUI custom_nodes directory: @@ -261,8 +267,45 @@ For issues and feature requests, please visit the [GitHub repository](https://gi - **1080p resolution** is only supported with **8-second duration** for Veo 3.1 models - **First-last frame interpolation** requires **Veo 3.1 models** and **8-second duration only** +## Watermark Remover Node Usage + +### Node: JM Gemini Watermark Remover + +This node removes the Gemini watermark that appears on generated images. + +#### Input: +- **image**: The Gemini-generated image with watermark (IMAGE tensor) + +#### Output: +- **image**: The same image with watermark removed (IMAGE tensor) + +#### Usage: + +1. Add "JM Gemini Watermark Remover" node to your workflow +2. Connect the output from "JM Gemini Image Generator" (or any Load Image node with a Gemini image) to the image input +3. Connect the output to Preview Image or Save Image node +4. Run the workflow + +The node will: +- Automatically detect the watermark size (48×48 or 96×96) based on image dimensions +- Calculate the watermark position (bottom-right corner) +- Remove the watermark using reverse alpha blending algorithm +- Save the cleaned image to the ComfyUI output directory + +**How it works:** +- Gemini adds watermarks to images larger than 512×512 pixels +- Images larger than 1024×1024 use a 96×96 watermark +- Smaller images use a 48×48 watermark +- The node uses pre-calculated alpha maps from background captures to reverse the watermark blending + ## Changelog +### Version 1.2.0 +- Added JM Gemini Watermark Remover node +- Automatically removes Gemini watermarks from generated images +- Supports both 48×48 and 96×96 watermark sizes +- Uses reverse alpha blending algorithm for clean removal + ### Version 1.1.0 - Added video generation support with JM Gemini Video Generator node - Support for Veo 3.1 and Veo 3.0 models diff --git a/README_CN.md b/README_CN.md index 4347f1c..c49aa3e 100644 --- a/README_CN.md +++ b/README_CN.md @@ -37,6 +37,12 @@ - **1080p分辨率**:Veo 3.1 模型的1080p分辨率仅支持8秒时长 - **首尾帧插值**:仅支持Veo 3.1模型,且时长只能为8秒 +### 水印去除 +- 自动检测并去除Gemini生成图像上的水印 +- 支持48×48和96×96两种水印尺寸 +- 使用反向alpha混合算法实现干净去除 +- 简单的一键操作 - 只需将图像传递给节点即可 + ## 安装说明 ### 1. 克隆或下载仓库 @@ -311,8 +317,45 @@ MIT License - **1080p分辨率**仅支持**8秒时长**(Veo 3.1模型) - **首尾帧插值**功能仅支持**Veo 3.1模型**,且**时长只能为8秒** +## 水印去除节点使用说明 + +### 节点:JM Gemini Watermark Remover(JM Gemini水印去除器) + +该节点用于去除Gemini生成图像上显示的水印。 + +#### 输入: +- **image**:带有水印的Gemini生成图像(IMAGE tensor格式) + +#### 输出: +- **image**:去除水印后的图像(IMAGE tensor格式) + +#### 使用方法: + +1. 在工作流中添加"JM Gemini Watermark Remover"节点 +2. 将"JM Gemini Image Generator"的输出(或任何包含Gemini图像的Load Image节点)连接到image输入 +3. 将输出连接到Preview Image或Save Image节点 +4. 运行工作流 + +节点将会: +- 根据图像尺寸自动检测水印大小(48×48或96×96) +- 计算水印位置(右下角) +- 使用反向alpha混合算法去除水印 +- 将处理后的图像保存到ComfyUI的output目录 + +**工作原理:** +- Gemini会为大于512×512像素的图像添加水印 +- 大于1024×1024的图像使用96×96水印 +- 较小的图像使用48×48水印 +- 节点使用从背景捕获预计算的alpha映射来反向混合水印 + ## 更新日志 +### 版本 1.2.0 +- 新增JM Gemini Watermark Remover水印去除节点 +- 自动去除Gemini生成图像上的水印 +- 支持48×48和96×96两种水印尺寸 +- 使用反向alpha混合算法实现干净去除 + ### 版本 1.1.0 - 新增视频生成功能,添加JM Gemini Video Generator节点 - 支持Veo 3.1和Veo 3.0模型