# AI图片编辑

**Repository Path**: martin8/batch-image

## Basic Information

- **Project Name**: AI图片编辑
- **Description**: 图片编辑工具，使用Ai的方法。
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-05-17
- **Last Updated**: 2026-09-09

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

<details open>
<summary>???? 中文</summary>

# AI Image Editor

基于 Electron 和 Vue 3 的跨平台桌面应用，支持批量 AI 图像处理。兼容 OpenAI、OpenRouter、阿里灵积模型广场、火山引擎等多个 AI 平台。

## 功能特点

- **批量处理**：单次操作处理多张图片，支持进度跟踪
- **多平台支持**：OpenAI、OpenRouter、OkRouter、灵积模型广场、火山引擎、FreeModel 及自定义 OpenAI 兼容 API
- **OpenAI Responses API**：支持 `gpt-image-2`、`gpt-4.1-mini` 等图像生成模型，可配置输出尺寸
- **智能图片压缩**：自动压缩、调整图片大小后上传，降低 API 费用
- **反幻觉提示词**：内置提示词前缀确保产品细节在编辑背景时保持不变
- **重试与恢复**：网络错误时自动重试（指数退避），支持仅重试失败文件
- **随时取消**：批量处理过程中可以优雅地停止
- **代理支持**：支持 HTTP/HTTPS 代理配置，解决部分地区访问限制
- **模板系统**：内置常见场景的提示词模板，支持变量替换
- **配置持久化**：API Key、偏好设置等通过 `electron-store` 本地加密存储
- **实时日志**：处理过程可视化，支持导出日志
- **跨平台**：支持 Windows、macOS、Linux

## 支持的平台

| 平台 | API 类型 | 支持模型 | 备注 |
|------|----------|------------------|------|
| **OpenAI 官方** | Images Edits / Responses | `gpt-image-2` `gpt-image-1` `dall-e-2`, `dall-e-3` | 完整尺寸支持 |
| **OpenRouter** | Chat Completions | 多模态模型 | OpenAI 兼容接口 |
| **OkRouter** | Chat Completions | 多模态模型 | OpenAI 兼容接口 |
| **灵积模型广场** | 多模态生成 | `qwen-image-*`, `wanx2.1-image-edit` | 国内优化 |
| **火山引擎** | Images Generations | 火山图像模型 | 支持 2K/4K 输出 |
| **FreeModel** | Responses API | `gpt-image-2` | Responses API 生成 |
| **自定义** | OpenAI 兼容 | 任意兼容模型 | 手动配置 Base URL |

## 环境要求

- Node.js v18 或更高
- npm 或 yarn
- 对应平台的 API Key

## 安装

```bash
npm install
```

（可选）在项目根目录创建 `app-config.json` 配置文件：
```json
{
  "openai": {
    "baseUrl": "https://api.openai.com/v1",
    "apiKey": "your_key_here",
    "defaultModel": "gpt-image-1"
  }
}
```

> 应用主要使用 UI 中的设置面板配置。`app-config.json` 中的值仅在未通过 UI 配置时作为后备。

## 使用方法

### 开发调试

```bash
npm run dev
```

### 打包发布

```bash
# 打包安装程序（Windows: NSIS，macOS: DMG，Linux: AppImage）
npm run dist

# 只打包免安装目录
npm run build
```

打包结果位于 `dist-installer/` 目录。

### 配置说明

1. 打开应用 → **Settings**
   - **Provider**：选择 AI 平台
   - **API Key**：填入平台对应的 API Key
   - **Model**：选择模型（切换 Provider 时会自动加载模型列表）
   - **Base URL**：切换 Provider 时自动填充，仅在"自定义"模式下需手动填写
   - **Proxy**：如需要代理请填写（如 `http://127.0.0.1:7890`）
   - **Test Connection**：处理前先测试连接可用性

### 操作流程

1. 选择包含图片的源文件夹
2. 选择输出文件夹
3. 选择或编写提示词（可使用模板快速开始）
4. 选择图片尺寸
5. 点击 **Start** 开始批量处理
6. 查看进度、日志，必要时重试失败文件

## 图片尺寸选项

| 尺寸 | 说明 |
|------|------|
| `auto` | 由 OpenAI 自动决定|
| `1024x1024` | 正方形 (1:1) |
| `1024x1536` | 竖屏 (2:3) |
| `1536x1024` | 横屏 (3:2) |
| `2048x2048` | 2K 正方形（OpenAI） |
| `2048x1152` | 2K 宽屏（OpenAI） |
| `3840x2160` | 4K（OpenAI） |
| `2160x3840` | 4K 竖屏（OpenAI） |
| `2K` | 火山引擎 2K |
| `4K` | 火山引擎 4K |

## 项目结构

```
batch-ai-image-editor/
├── src/
│   ├── main/
│   │   ├── index.js          # 主进程：IPC、配置、批量调度
│   │   └── processImage.js   # 图像处理：各平台 API 调用
│   └── renderer/
│       ├── index.html        # 界面
│       ├── index.js          # 渲染进程逻辑
│       └── styles.css        # 样式
├── assets/                   # 资源文件
├── package.json              # 依赖与构建配置
└── README.md                 # 本文件
```

## 技术栈

- **Electron** (^30.0.0)、**Vue 3** (^3.4.0)、**Element Plus** (^2.6.0)
- **OpenAI SDK** (^6.36.0)、**Sharp** (^0.33.5)、**Axios** (^1.6.0)
- **electron-store** (^8.1.0)

## 安全说明

- API Key 通过 `electron-store` 加密存储
- 使用 Electron 上下文隔离和安全 IPC 通信
- `app-config.json` 仅作为后备配置，非必需

## 开源协议

MIT License

</details>

---

<details>
<summary>???? English</summary>

# AI Image Editor

A cross-platform desktop application for batch AI-powered image editing, built with Electron and Vue 3. Supports multiple AI providers including OpenAI, OpenRouter, Alibaba DashScope, Volcengine, and more.

## Features

- **Batch Processing**: Process multiple images in a single operation with progress tracking
- **Multi-Provider Support**: Compatible with OpenAI, OpenRouter, OkRouter, Alibaba DashScope, Volcengine, FreeModel, and custom OpenAI-compatible APIs
- **OpenAI Responses API**: Supports `gpt-image-1`, `gpt-4.1-mini`, and other image generation models via the Responses API with configurable size parameters
- **Smart Image Compression**: Automatically compresses and resizes images before uploading to reduce API costs and avoid size limits
- **Anti-Hallucination Prompts**: Built-in prompt prefix ensures product details are preserved during background editing
- **Retry & Resume**: Automatically retries on network errors (5xx, 429) with exponential backoff; supports retrying failed files only
- **Cancel Support**: Gracefully stop batch processing mid-operation
- **Proxy Support**: HTTP/HTTPS proxy configuration for regions with restricted API access
- **Template System**: Built-in prompt templates for common use cases
- **Variable Replacement**: Supports `[Product Name]` and other variables in prompt templates
- **Persistent Config**: Settings, API keys, and preferences are saved locally via `electron-store`
- **Log Output**: Real-time processing logs with export capability
- **Cross-Platform**: Runs on Windows, macOS, and Linux

## Supported Providers

| Provider | API Type | Supported Models | Notes |
|----------|----------|------------------|-------|
| **OpenAI Official** | Images Edits / Responses | `dall-e-2`, `dall-e-3`, `gpt-image-1`, `gpt-4.1-mini` | Full size support (auto, 1024x1024, 2048x2048, 3840x2160) |
| **OpenRouter** | Chat Completions | Various multimodal models | OpenAI-compatible endpoint |
| **OkRouter** | Chat Completions | Various multimodal models | OpenAI-compatible endpoint |
| **Alibaba DashScope** | Multimodal Generation | `qwen-image-*`, `wanx2.1-image-edit` | Chinese region optimized |
| **Volcengine** | Images Generations | Volcengine image models | Supports 2K/4K output |
| **FreeModel** | Responses API | `gpt-image-2` | Image generation via Responses API |
| **Custom** | OpenAI-compatible | Any compatible model | Manual base URL configuration |

## Prerequisites

- Node.js (v18 or higher)
- npm or yarn
- OpenAI API key (or equivalent for your chosen provider)

## Installation

```bash
npm install
```

(Optional) Create an `app-config.json` file in the root directory:
```json
{
  "openai": {
    "baseUrl": "https://api.openai.com/v1",
    "apiKey": "your_key_here",
    "defaultModel": "gpt-image-1"
  }
}
```

> The app primarily uses settings configured via the UI (Settings panel). The `app-config.json` values are only used as a fallback when no API key is configured in the app.

## Usage

### Development

```bash
npm run dev
```

### Build for Distribution

```bash
# Build installer (Windows: NSIS, macOS: DMG, Linux: AppImage)
npm run dist

# Build unpacked directory only
npm run build
```

Packaged output is in the `dist-installer/` folder.

### Configuration

1. **Settings Panel** (recommended): Open the app → Settings
   - **Provider**: Select your AI platform
   - **API Key**: Enter your provider's API key
   - **Model**: Choose a supported model (auto-populated by provider)
   - **Base URL**: Auto-filled when switching providers; customize for private deployments
   - **Proxy**: Enter proxy URL if needed (e.g., `http://127.0.0.1:7890`)
   - **Test Connection**: Verify connectivity before processing

### Workflow

1. Select source folder containing images
2. Select output folder for processed images
3. Choose or write a prompt (use templates for quick start)
4. Select image size (provider-specific options available)
5. Click **Start** to begin batch processing
6. Monitor progress, view logs, and retry failed files if needed

## Image Size Options

| Size | Description |
|------|-------------|
| `auto` | Let OpenAI decide (OpenAI Official-1) |
| `1024x1024` | Square (1:1) |
| `1024x1536` | Portrait (2:3) |
| `1536x1024` | Landscape (3:2) |
| `2048x2048` | 2K Square (OpenAI) |
| `2048x1152` | 2K Widescreen (OpenAI) |
| `3840x2160` | 4K (OpenAI) |
| `2160x3840` | 4K Portrait (OpenAI) |
| `2K` | Volcengine 2K |
| `4K` | Volcengine 4K |

## Project Structure

```
batch-ai-image-editor/
├── src/
│   ├── main/
│   │   ├── index.js          # Main process: IPC, config, batch orchestration
│   │   └── processImage.js   # Image processing: API calls per provider
│   └── renderer/
│       ├── index.html        # Main UI
│       ├── index.js          # Renderer process logic
│       └── styles.css        # Styles
├── assets/                   # Application assets (icons, etc.)
├── package.json              # Dependencies and build config
└── README.md                 # This file
```

## Technologies Used

- **Electron** (^30.0.0), **Vue 3** (^3.4.0), **Element Plus** (^2.6.0)
- **OpenAI SDK** (^6.36.0), **Sharp** (^0.33.5), **Axios** (^1.6.0)
- **electron-store** (^8.1.0)

## Security

- API keys are stored in `electron-store` (encrypted on supported platforms)
- Application uses Electron context isolation and secure IPC communication
- `app-config.json` is optional and acts as a fallback only

## License

MIT License

</details>