feat: 添加AI模板管理系统和Docker配置

- 实现面向对象的模板管理系统 - 添加N8n模板抽象类和具体实现 - 创建老照片修复上色模板 - 创建修人物表情闭眼模板 - 集成BowongAI服务 - 添加完整的Docker配置（多阶段构建） - 添加docker-compose配置 - 完善项目文档和API设计
2025-09-03 17:37:30 +08:00
parent 7e42c9e756
commit e30d7cc4da
25 changed files with 6734 additions and 24 deletions
--- a/docs/bowongai-integration-summary.md
+++ b/docs/bowongai-integration-summary.md
@@ -0,0 +1,350 @@
+# BowongAI 集成总结
+
+## 🎯 **集成概述**
+
+基于您提供的BowongAI API接口，我已经完整更新了模板管理系统，实现了图生图和图生视频功能的集成。
+
+## 🔗 **API接口信息**
+
+### Webhook URL
+```
+https://n8n.bowongai.com/webhook/76f92dbf-785f-4add-96b5-a108174b7c14
+```
+
+### 环境地址
+```
+https://bowongai-test--text-video-agent-fastapi-app.modal.run
+```
+
+## 📋 **支持的工作流**
+
+### 1. 图生图工作流
+```json
+{
+    "workflow": "图生图",
+    "environment": "https://bowongai-test--text-video-agent-fastapi-app.modal.run",
+    "image_generation": {
+        "model": "gemini-2.5-flash-image-preview",
+        "prompt": "详细的提示词...",
+        "image_url": "输入图片URL"
+    }
+}
+```
+
+### 2. 图生图+生视频工作流
+```json
+{
+    "workflow": "图生图+生视频",
+    "environment": "https://bowongai-test--text-video-agent-fastapi-app.modal.run",
+    "image_generation": {
+        "model": "gemini-2.5-flash-image-preview",
+        "prompt": "详细的提示词...",
+        "image_url": "输入图片URL"
+    },
+    "video_generation": {
+        "model": "302/MiniMax-Hailuo-02",
+        "prompt": "do anything you want",
+        "duration": "6",
+        "aspect_ratio": "9:16"
+    }
+}
+```
+
+## 🏗️ **技术实现**
+
+### 1. BowongAI基础服务类（通用服务）
+```typescript
+// src/services/bowongai.service.ts
+@Injectable()
+export class BowongAIService {
+  private readonly webhookUrl = 'https://n8n.bowongai.com/webhook/76f92dbf-785f-4add-96b5-a108174b7c14';
+  private readonly environment = 'https://bowongai-test--text-video-agent-fastapi-app.modal.run';
+
+  // 通用图生图接口 - 各模板传入自己的参数
+  async imageToImage(options: {
+    prompt: string;
+    imageUrl: string;
+    model?: string;
+  }): Promise<ImageToImageResponse>
+
+  // 通用图生视频接口 - 各模板传入自己的参数
+  async imageToVideo(options: {
+    imagePrompt: string;
+    videoPrompt: string;
+    imageUrl: string;
+    duration?: number;
+    aspectRatio?: string;
+    imageModel?: string;
+    videoModel?: string;
+  }): Promise<ImageToVideoResponse>
+}
+```
+
+### 2. 具体模板实现（各自定制）
+
+#### 换装模板（图生图）
+```typescript
+@Injectable()
+export class OutfitChangeTemplate extends ImageGenerateTemplate {
+  readonly metadata = {
+    code: 'outfit_change_v1',
+    name: '智能换装',
+    category: '换装',
+    creditCost: 15,
+  };
+
+  constructor(private readonly bowongAI: BowongAIService) {}
+
+  async execute(input, context) {
+    // 换装模板专用提示词
+    const prompt = this.buildOutfitChangePrompt(input);
+
+    return await this.bowongAI.imageToImage({
+      prompt,
+      imageUrl: input.inputImage,
+      model: 'gemini-2.5-flash-image-preview', // 换装专用模型
+    });
+  }
+
+  private buildOutfitChangePrompt(input) {
+    return `Please convert this photo into a highly detailed character model wearing ${input.clothingDescription}. Place in front of the model a delicate, colorful box featuring the printed image of the person from the photo...`;
+  }
+}
+```
+
+#### 人像增强模板（图生图）
+```typescript
+@Injectable()
+export class PortraitEnhanceTemplate extends ImageGenerateTemplate {
+  readonly metadata = {
+    code: 'portrait_enhance_v1',
+    name: '人像增强',
+    category: '人像处理',
+    creditCost: 12,
+  };
+
+  constructor(private readonly bowongAI: BowongAIService) {}
+
+  async execute(input, context) {
+    // 人像增强专用提示词
+    const prompt = this.buildPortraitPrompt(input);
+
+    return await this.bowongAI.imageToImage({
+      prompt,
+      imageUrl: input.inputImage,
+      model: 'gemini-2.5-flash-image-preview',
+    });
+  }
+
+  private buildPortraitPrompt(input) {
+    return `Enhance this portrait photo to professional quality. Improve skin texture, lighting, and overall clarity while maintaining natural appearance. ${input.enhanceLevel === 'subtle' ? 'Apply subtle enhancements' : 'Apply dramatic improvements'}...`;
+  }
+}
+```
+
+#### 动态视频模板（图生视频）
+```typescript
+@Injectable()
+export class DynamicVideoTemplate extends VideoGenerateTemplate {
+  readonly metadata = {
+    code: 'dynamic_video_v1',
+    name: '动态视频',
+    category: '视频生成',
+    creditCost: 45,
+  };
+
+  constructor(private readonly bowongAI: BowongAIService) {}
+
+  async execute(input, context) {
+    // 动态视频专用提示词
+    const imagePrompt = this.buildImagePrompt(input);
+    const videoPrompt = this.buildVideoPrompt(input);
+
+    return await this.bowongAI.imageToVideo({
+      imagePrompt,
+      videoPrompt,
+      imageUrl: input.inputImage,
+      duration: input.duration || 4,
+      aspectRatio: input.aspectRatio || '9:16',
+      imageModel: 'gemini-2.5-flash-image-preview',
+      videoModel: '302/MiniMax-Hailuo-02',
+    });
+  }
+
+  private buildImagePrompt(input) {
+    return `Convert this image for dynamic video generation with ${input.motionType} motion effects...`;
+  }
+
+  private buildVideoPrompt(input) {
+    return `Create ${input.motionType} animation with ${input.intensity} intensity. ${input.customPrompt || 'Natural movement'}`;
+  }
+}
+```
+
+## 🎨 **模板化提示词设计**
+
+### 换装模板提示词
+```typescript
+private buildOutfitChangePrompt(input: OutfitChangeInput): string {
+  const styleMap = {
+    casual: 'casual everyday wear',
+    formal: 'formal business attire',
+    vintage: 'vintage retro style',
+    modern: 'modern contemporary fashion',
+    elegant: 'elegant sophisticated style'
+  };
+
+  const styleText = styleMap[input.style] || 'casual';
+
+  return `Please convert this photo into a highly detailed character model wearing ${input.clothingDescription}.
+Style: ${styleText}.
+Place in front of the model a delicate, colorful box featuring the printed image of the person from the photo.
+Behind the box, show a computer screen actively displaying the modeling process.
+Set the entire scene in a bright, stylish indoor environment.
+Ensure the lighting is crisp and luminous, highlighting both the model and its packaging.`;
+}
+```
+
+### 人像增强模板提示词
+```typescript
+private buildPortraitPrompt(input: PortraitEnhanceInput): string {
+  const enhancementLevels = {
+    subtle: 'Apply subtle professional enhancements',
+    moderate: 'Apply moderate quality improvements',
+    dramatic: 'Apply dramatic professional transformation'
+  };
+
+  const enhancement = enhancementLevels[input.enhanceLevel] || 'moderate';
+
+  return `${enhancement} to this portrait photo.
+Improve skin texture, lighting, and overall clarity while maintaining natural appearance.
+Focus on professional photography quality with ${input.lightingStyle || 'natural'} lighting.
+Preserve the person's unique features and expressions.`;
+}
+```
+
+### 动态视频模板提示词
+```typescript
+private buildVideoPrompt(input: DynamicVideoInput): string {
+  const motionTypes = {
+    gentle: 'gentle subtle movements',
+    dynamic: 'dynamic energetic motion',
+    cinematic: 'cinematic smooth transitions',
+    artistic: 'artistic creative effects'
+  };
+
+  const motion = motionTypes[input.motionType] || 'gentle';
+
+  return `Create ${motion} with ${input.intensity || 'medium'} intensity.
+${input.customPrompt || 'Natural lifelike movement'}.
+Maintain character consistency throughout the animation.`;
+}
+```
+
+## 💰 **模板积分消耗**
+
+### 图生图模板
+- **换装模板**: 15积分
+- **人像增强**: 12积分
+- **风格转换**: 18积分
+- **背景替换**: 20积分
+
+### 图生视频模板
+- **动态视频**: 45积分
+- **角色动画**: 50积分
+- **场景视频**: 55积分
+- **特效视频**: 60积分
+
+## 🔧 **环境配置**
+
+```env
+# BowongAI 配置
+BOWONGAI_WEBHOOK_URL=https://n8n.bowongai.com/webhook/76f92dbf-785f-4add-96b5-a108174b7c14
+BOWONGAI_ENVIRONMENT=https://bowongai-test--text-video-agent-fastapi-app.modal.run
+```
+
+## 📱 **前端调用示例**
+
+### 图生图调用
+```typescript
+const response = await fetch('/api/templates/outfit_change_v1/execute', {
+  method: 'POST',
+  headers: {
+    'Authorization': `Bearer ${token}`,
+    'Content-Type': 'application/json',
+  },
+  body: JSON.stringify({
+    inputImage: 'https://cdn.roasmax.cn/upload/e9066bf20fb546b09ece65d6fbcf2dd3.png',
+    clothingDescription: '红色连衣裙',
+    style: 'elegant'
+  })
+});
+```
+
+### 图生视频调用
+```typescript
+const response = await fetch('/api/templates/image_to_video_v1/execute', {
+  method: 'POST',
+  headers: {
+    'Authorization': `Bearer ${token}`,
+    'Content-Type': 'application/json',
+  },
+  body: JSON.stringify({
+    inputImage: 'https://cdn.roasmax.cn/upload/e9066bf20fb546b09ece65d6fbcf2dd3.png',
+    videoPrompt: '让角色动起来',
+    duration: 6,
+    resolution: '9:16'
+  })
+});
+```
+
+## ✅ **基础服务架构优势**
+
+### 🏗️ **分层设计**
+1. **BowongAI基础服务层**: 提供通用的API调用能力
+2. **模板业务层**: 各模板实现自己的业务逻辑和提示词
+3. **参数定制层**: 每个模板可以传入不同的参数配置
+
+### 🎯 **模板独立性**
+- ✅ 每个模板有独立的提示词逻辑
+- ✅ 每个模板有独立的参数验证
+- ✅ 每个模板有独立的积分消耗
+- ✅ 每个模板可以选择不同的AI模型
+
+### 📊 **当前实现状态**
+- ✅ BowongAI基础服务实现
+- ✅ 换装模板（图生图）- 15积分
+- ✅ 图生视频模板 - 50积分
+- ✅ 人像增强模板 - 12积分
+- ✅ 模板工厂和模块配置
+- ✅ 环境配置和前端示例
+
+## 🚀 **扩展新模板步骤**
+
+1. **继承对应的抽象类**
+2. **实现execute方法，调用BowongAI基础服务**
+3. **定义模板专用的提示词构建逻辑**
+4. **配置模板特定的参数和积分**
+5. **在工厂类中注册新模板**
+
+### 示例：添加风格转换模板
+```typescript
+@Injectable()
+export class StyleTransferTemplate extends ImageGenerateTemplate {
+  async execute(input, context) {
+    const prompt = this.buildStylePrompt(input); // 风格转换专用提示词
+
+    return await this.bowongAI.imageToImage({
+      prompt,
+      imageUrl: input.inputImage,
+      model: 'gemini-2.5-flash-image-preview',
+    });
+  }
+
+  private buildStylePrompt(input) {
+    return `Transform this image into ${input.targetStyle} style...`;
+  }
+}
+```
+
+现在BowongAI作为基础服务，支持各种模板的个性化定制！🎉