Add files via upload

update plugin to new version
2026-01-13 19:26:31 +08:00
parent e3196df2d2
commit b05dfd5e91
5 changed files with 936 additions and 482 deletions
@@ -1,4 +1,220 @@
-## 说明
-
-本插件由 WL 为公司（TDGO）定制开发。  
-如需类似功能或定制开发服务，请联系V：HKMSJ666
+# 🎨 Comfyui-WL-MainImageDesign
+
+**专业电商主图提示词生成器 - 摩托车骑行装备专版**
+
+一款基于 ComfyUI 的 AI 电商主图设计插件，专为摩托车头盔、骑行服、护具等骑行装备定制，自动生成高转化率的电商主图提示词。
+
+---
+
+## ✨ 核心特点
+
+- 🏍️ **摩托车装备专属** - 9种专业风格针对骑行装备深度优化
+- 🎯 **高转化率设计** - 基于13张高转化主图深度分析的排版规范
+- 🖼️ **场景融合** - 产品融入真实骑行场景，告别纯白底平铺
+- 📐 **精准排版** - 支持1:1和3:4两种主流电商比例
+- 🌍 **双语支持** - 中英文提示词自由切换
+- 🔗 **参考图锁定** - 上传产品图自动锁定外观，重建专业场景
+
+---
+
+## 📦 安装方法
+
+### 方式一：手动安装
+
+1. 将整个文件夹复制到 ComfyUI 的 `custom_nodes` 目录：
+```
+ComfyUI/custom_nodes/Comfyui-WL-MainImageDesign/
+├── __init__.py
+├── prompt_nodes.py
+├── prompts/
+│   ├── system_prompt.txt
+│   └── default_prompt.txt
+└── README.md
+```
+
+2. 重启 ComfyUI
+
+### 方式二：Git Clone
+
+```bash
+cd ComfyUI/custom_nodes
+git clone https://github.com/your-repo/Comfyui-WL-MainImageDesign.git
+```
+
+---
+
+## 🎮 使用说明
+
+### 节点列表
+
+| 节点名称 | 功能描述 |
+|---------|---------|
+| 🎨 WL Main Image Designer | 主图提示词生成器 |
+| 🔄 WL Prompt Batch Converter | 提示词批量转换器 |
+
+### 基本参数
+
+| 参数 | 说明 | 默认值 |
+|-----|------|-------|
+| `api_url` | LLM API 地址 | https://api.openai.com/v1 |
+| `api_key` | API 密钥 | - |
+| `model_name` | 模型名称 | gemini-2.0-flash-exp |
+| `product_type` | 产品类型 | 摩托车头盔 |
+| `selling_points` | 核心卖点 | - |
+| `design_style` | 设计风格 | 9种可选 |
+| `scene_mode` | 场景模式 | 3种可选 |
+| `aspect_ratio` | 画幅比例 | 1:1 / 3:4 |
+| `prompt_count` | 生成数量 | 1-10 |
+| `output_language` | 输出语言 | 中文/英文 |
+
+---
+
+## 🎨 9种专业设计风格
+
+### 1. 专业机能风
+- **场景**：专业车库/维修间环境
+- **色调**：深灰 #2d2d2d 或冷白 #f5f5f5
+- **适用**：功能性装备、通讯设备、骑行配件
+
+### 2. 硬核竞技风
+- **场景**：赛道压弯、维修区、高速骑行
+- **色调**：深黑 #0a0a0a，红/橙激情点缀
+- **适用**：赛车头盔、赛道皮衣、高性能装备
+
+### 3. 工业机械风
+- **场景**：重型机械车间、改装车库
+- **色调**：深灰金属 #3d3d3d，工业蓝
+- **适用**：金属护具、机械配件、工具类产品
+
+### 4. 都市通勤风
+- **场景**：城市街道、咖啡店、写字楼停车场
+- **色调**：自然城市色彩，灰调带暖色
+- **适用**：通勤头盔、日常骑行服、都市配件
+
+### 5. 户外冒险风
+- **场景**：山路骑行、荒野休息、越野穿越
+- **色调**：大地色系（土色、军绿、深蓝）
+- **适用**：ADV头盔、防水装备、长途旅行用品
+
+### 6. 高端质感风
+- **场景**：高端展厅、私人车库、极简摄影棚
+- **色调**：深色渐变 #1a1a1a→#2d2d2d
+- **适用**：高端头盔、品牌皮具、限量款装备
+
+### 7. 安全守护风
+- **场景**：安全测试、家人送别、专业培训
+- **色调**：稳重蓝 #1e3a5f 或沉稳灰
+- **适用**：安全头盔、护具、新手友好装备
+
+### 8. 改装美学风
+- **场景**：改装车间、潮流街拍、霓虹夜景
+- **色调**：霓虹粉/蓝/绿，大胆对比
+- **适用**：彩绘头盔、改装配件、个性装备
+
+### 9. 参数对比风
+- **场景**：干净对比环境、实验室测试
+- **色调**：中性白 #ffffff 或浅灰 #f5f5f5
+- **适用**：功能性产品、需要参数对比的装备
+
+---
+
+## 📐 排版规范
+
+### 1:1 正方形 (800×800)
+```
+┌─────────────────────────────┐
+│ 顶部 0-20%   主标题 52-60pt │
+├─────────────────────────────┤
+│ 上中 18-38%  副标题/卖点    │
+├─────────────────────────────┤
+│ 中心 25-75%  产品场景 45-60%│
+├─────────────────────────────┤
+│ 下中 65-82%  规格/认证      │
+├─────────────────────────────┤
+│ 底部 78-100% 价格/促销/信任 │
+└─────────────────────────────┘
+```
+
+### 3:4 竖版 (600×800)
+```
+┌─────────────────────────────┐
+│ 顶部 0-18%   主标题 52-60pt │
+├─────────────────────────────┤
+│ 上部 15-32%  副标题/促销    │
+├─────────────────────────────┤
+│ 中心 28-72%  产品场景 50-65%│
+├─────────────────────────────┤
+│ 下部 68-85%  详细规格       │
+├─────────────────────────────┤
+│ 底部 82-100% 价格/CTA/信任  │
+└─────────────────────────────┘
+```
+
+---
+
+## 🛡️ 质量保障
+
+### 主标题规范
+- **字号**：52-60pt 超大加粗，绝对视觉焦点
+- **字重**：900-950，笔画饱满厚实
+- **对比度**：与背景至少 8:1
+- **位置**：画面上 15-20% 区域
+
+### 人体完整性约束 ⚠️
+当场景涉及人物/骑士时，严格保证：
+- ✅ 手指 5 根完整，关节自然
+- ✅ 手臂比例协调，无断裂变形
+- ✅ 姿态符合物理规律
+- ✅ 穿戴逻辑正确
+
+**降低出错建议**：
+- 优先侧面/背影/剪影
+- 手部被产品遮挡（如握把手套只露手背）
+- 远景人物（占比 <30%）
+- 局部特写（手腕以下握住产品）
+
+### 安全认证标识
+支持行业标准认证展示：
+- ECE（欧洲标准）
+- DOT（美国交通部）
+- SNELL（斯奈尔基金会）
+- 3C（中国强制认证）
+
+---
+
+## 📝 更新日志
+
+### v2.1 (2025-01)
+- 🆕 风格系统重构：8种通用风格 → 9种摩托车专属风格
+- 🔤 主标题升级：42pt → 52-60pt 超大加粗
+- 🦾 新增人体完整性约束，避免畸形问题
+- 🏍️ 场景融合优化，更贴合骑行装备展示
+- 📜 认证标识更新为 ECE/DOT/SNELL/3C
+
+### v2.0
+- 初版发布
+- 支持 9 种设计风格
+- 支持参考图锁定
+- 支持中英双语输出
+
+---
+
+## 🤝 兼容性
+
+- **ComfyUI**: 最新版本
+- **Python**: 3.10+
+- **依赖**: PIL/Pillow, numpy
+
+---
+
+## 📄 许可证
+
+MIT License
+
+---
+
+## 💬 反馈与支持
+
+如有问题或建议，欢迎提交 Issue 或 Pull Request。
+
+**让每一张主图都能卖货！** 🏍️✨
@@ -2,18 +2,35 @@
 Comfyui-WL-MainImageDesign
 Elite E-commerce Main Image Prompt Generator

-Version: 2.0
+Version: 2.1 - Enhanced Visual Impact
 Author: WL Design Studio
 Description: Professional main image prompt generator with high visual impact optimization.
-             Based on deep analysis of 13 high-converting e-commerce hero images.
+             Features scene integration and lifestyle context for maximum conversion.

 Features:
 - 42pt headline standard
 - 1:1 and 3:4 aspect ratio optimization
- 8 professional background styles
+- 9 professional styles with SCENE INTEGRATION
+- 3 scene modes (融合/棚拍/动态)
 - Advanced price & promo badge system
 - Commercial photography grade output
 - Visual impact score target: 9/10
+
+Styles (场景融合版):
+1. 专业机能风 - Professional Functional (车库/检查装备场景)
+2. 硬核竞技风 - Racing / Track (赛道/动态骑行场景)
+3. 工业机械风 - Industrial / Mechanical (机械车间/改装场景)
+4. 都市通勤风 - Urban / Daily Ride (城市街道/日常生活场景)
+5. 户外冒险风 - ADV / Touring (山路/荒野探索场景)
+6. 高端质感风 - Premium / Luxury (展厅/私人车库场景)
+7. 安全守护风 - Safety Focused (测试/家人送别场景)
+8. 改装美学风 - Custom / Style (改装车间/潮流街拍场景)
+9. 参数对比风 - Spec / Data (对比展示/实验室场景)
+
+Scene Modes:
+- 场景融合（产品+使用情境）- Product in usage context
+- 纯产品棚拍（干净背景）- Studio with designed background
+- 骑行动态场景（人车合一）- Dynamic riding scenes
 """

 from .prompt_nodes import WLMainImageGenerator, WLPromptBatchConverter
@@ -28,5 +45,5 @@ NODE_DISPLAY_NAME_MAPPINGS = {
    "WLPromptBatchConverter": "🔄 WL Prompt Batch Converter"
 }

-__version__ = "2.0.0"
+__version__ = "2.1.0"
 __all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
@@ -1,7 +1,7 @@
 """
 Comfyui-WL-MainImageDesign
 Elite E-commerce Main Image Prompt Generator
-Version: 2.0
+Version: 2.1 - Enhanced Visual Impact
 """

 import json
@@ -64,7 +64,6 @@ class WLMainImageGenerator:

        import re

-        # 尝试匹配 JSON 对象模式
        json_obj_pattern = r'\{\s*"prompt"\s*:\s*"'
        matches = list(re.finditer(json_obj_pattern, s))
        if len(matches) >= 2:
@@ -86,7 +85,6 @@ class WLMainImageGenerator:
            if parsed_objects:
                return parsed_objects

-        # 尝试匹配变体定位标记
        start_markers = [
            r"(?m)^\s*变体定位\s*[：:]",
            r"(?m)^\s*Variant Role\s*:",
@@ -102,14 +100,12 @@ class WLMainImageGenerator:
                if parts:
                    return parts

-        # 尝试用分隔符分割
        if "\n---\n" in s:
            parts = [p.strip() for p in s.split("\n---\n")]
            parts = [p for p in parts if p]
            if parts:
                return parts

-        # 尝试用多个空行分割
        parts = [p.strip() for p in re.split(r"\n\s*\n\s*\n+", s)]
        parts = [p for p in parts if p]
        if len(parts) >= 2:
@@ -118,7 +114,6 @@ class WLMainImageGenerator:
        return [s]

    def _clean_code_fences(self, response_text):
-        """清理代码围栏标记"""
        cleaned = (response_text or "").strip()
        if cleaned.startswith("```json"):
            cleaned = cleaned[7:]
@@ -129,7 +124,6 @@ class WLMainImageGenerator:
        return cleaned.strip()

    def _parse_response_to_prompts_list(self, response_text, expected_count):
-        """解析响应为提示词列表"""
        method = "json"
        prompts_list = []
        try:
@@ -164,65 +158,50 @@ class WLMainImageGenerator:
        return prompts_list, method

    def _strip_variant_role_header(self, prompt_text):
-        """移除变体定位头部（可选）"""
        if not isinstance(prompt_text, str):
            return prompt_text
        return prompt_text.strip()

    def _is_prompt_structurally_complete(self, prompt_text):
-        """检查提示词是否结构完整"""
        if not isinstance(prompt_text, str):
            return False
        s = prompt_text.strip()
        if not s:
            return False
-
        has_main = ("主标题" in s) or ("Main Headline" in s)
        has_visual = ("视觉" in s) or ("Visual" in s) or ("光影" in s) or ("Lighting" in s) or ("背景" in s) or ("Background" in s)
-
        return has_main and has_visual

    def enforce_prompt_count(self, prompts_list, prompt_count, raw_response):
-        """确保提示词数量符合要求"""
        try:
            pc = int(prompt_count)
        except Exception:
            pc = None
-
        if not pc or pc <= 0:
            return prompts_list
-
        if not prompts_list:
            return [str(raw_response)]
-
        if len(prompts_list) == pc:
            return prompts_list
-
        if len(prompts_list) > pc:
            return prompts_list[:pc]
-
        if len(prompts_list) == 1 and isinstance(prompts_list[0], str):
            parts = self.split_response_to_variants(prompts_list[0], pc)
            if parts and len(parts) >= pc:
                return parts[:pc]
-
        parts = self.split_response_to_variants(raw_response, pc)
        if parts and len(parts) >= pc:
            return parts[:pc]
-
        return prompts_list

    def extract_prompt_text(self, item):
-        """提取提示词文本"""
        if item is None:
            return None
-
        if isinstance(item, dict):
            prompt = item.get("prompt")
            if isinstance(prompt, str):
                return prompt
            return json.dumps(item, ensure_ascii=False)
-
        if isinstance(item, str):
            s = item.strip()
            if (s.startswith("{") and s.endswith("}")) or (s.startswith("[{") and s.endswith("}]")):
@@ -232,9 +211,7 @@ class WLMainImageGenerator:
                        return obj["prompt"]
                except Exception:
                    pass
-
            return s.replace("\\n", "\n")
-
        return str(item)

    @classmethod
@@ -254,27 +231,35 @@ class WLMainImageGenerator:
                    "multiline": False, 
                    "default": "gemini-2.0-flash-exp",
                }),
-
                "product_type": ("STRING", {
                    "multiline": False,
-                    "default": "蓝牙耳机",
+                    "default": "摩托车头盔",
                }),
                "selling_points": ("STRING", {
                    "multiline": True,
-                    "default": "降噪、长续航、佩戴舒适",
+                    "default": "碳纤维材质、轻量化设计、风噪降低",
                }),
                "design_style": (
                    [
-                        "科技深邃 (Tech Deep)",
-                        "温润米白 (Warm Cream)",
-                        "高级金棕 (Premium Bronze)",
-                        "清新天蓝 (Fresh Sky)",
-                        "氛围场景 (Lifestyle Scene)",
-                        "硬核深棕 (Hardcore Brown)",
-                        "极简纯净 (Minimal Pure)",
-                        "未来科幻 (Sci-Fi Future)",
+                        "专业机能风",
+                        "硬核竞技风",
+                        "工业机械风",
+                        "都市通勤风",
+                        "户外冒险风",
+                        "高端质感风",
+                        "安全守护风",
+                        "改装美学风",
+                        "参数对比风",
                    ],
-                    {"default": "科技深邃 (Tech Deep)"}
+                    {"default": "专业机能风"}
+                ),
+                "scene_mode": (
+                    [
+                        "场景融合（产品+使用情境）",
+                        "纯产品棚拍（干净背景）",
+                        "骑行动态场景（人车合一）",
+                    ],
+                    {"default": "场景融合（产品+使用情境）"}
                ),
                "aspect_ratio": (
                    [
@@ -309,7 +294,7 @@ class WLMainImageGenerator:
                        "买赠活动框 (买2赠1)",
                        "新品首发标签",
                        "官方正品徽章",
-                        "认证标签 (ACS/SGS/Hi-Res)",
+                        "认证标签 (ECE/DOT/SNELL)",
                        "无促销标签",
                    ],
                    {"default": "TOP排名徽章 (热销第1名)"}
@@ -340,45 +325,35 @@ class WLMainImageGenerator:
    RETURN_TYPES = ("STRING", "STRING")
    RETURN_NAMES = ("prompts_list", "debug_info")
    OUTPUT_IS_LIST = (True, False)
-
    FUNCTION = "generate_main_image_prompts"
    CATEGORY = "🎨 WL-MainImageDesign"

    def tensor_to_base64(self, image, index=0):
-        """将 ComfyUI Tensor 图片转换为 Base64"""
        if image is None:
            return None
-
        img_tensor = image
        try:
            if hasattr(image, "shape") and len(image.shape) == 4:
                img_tensor = image[index]
        except Exception:
            img_tensor = image
-
        i = 255. * img_tensor.cpu().numpy()
        img = Image.fromarray(np.clip(i, 0, 255).astype(np.uint8))
-
        max_size = 1024
        if img.width > max_size or img.height > max_size:
            img.thumbnail((max_size, max_size), Image.Resampling.LANCZOS)
-
        buffered = io.BytesIO()
        img.save(buffered, format="JPEG", quality=85)
        img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
        return f"data:image/jpeg;base64,{img_str}"

    def collect_base64_images(self, images, max_images=6):
-        """收集多张图片的 Base64 编码"""
        base64_images = []
-
        if images is None:
            return base64_images
-
        for img in images:
            if img is None:
                continue
-
            try:
                if hasattr(img, "shape") and len(img.shape) == 4:
                    batch = int(img.shape[0])
@@ -394,23 +369,19 @@ class WLMainImageGenerator:
                if len(base64_images) >= max_images:
                    return base64_images
                base64_images.append(self.tensor_to_base64(img, 0))
-
        return [b for b in base64_images if b]

    def call_llm_vision(self, api_url, api_key, model, system_prompt, user_prompt, base64_images=None, seed=None):
-        """调用 LLM Vision API"""
        headers = {
            "Content-Type": "application/json",
            "Authorization": f"Bearer {api_key}",
-            "User-Agent": "ComfyUI-WL-MainImageDesign/2.0"
+            "User-Agent": "ComfyUI-WL-MainImageDesign/2.1"
        }
-
        url = api_url.rstrip('/')
        if url.endswith('/chat'):
            url = f"{url}/completions"
        elif not url.endswith('/chat/completions'):
            url = f"{url}/chat/completions"
-
        content_list = [{"type": "text", "text": user_prompt}]
        if base64_images:
            if isinstance(base64_images, str):
@@ -420,16 +391,12 @@ class WLMainImageGenerator:
                    continue
                content_list.append({
                    "type": "image_url",
-                    "image_url": {
-                        "url": base64_image
-                    }
+                    "image_url": {"url": base64_image}
                })
-
        messages = [
            {"role": "system", "content": system_prompt},
            {"role": "user", "content": content_list}
        ]
-
        payload = {
            "model": model,
            "messages": messages,
@@ -437,16 +404,12 @@ class WLMainImageGenerator:
            "temperature": 0.75,
            "stream": False
        }
-        
        if seed is not None and seed > 0:
            payload["seed"] = seed
-
        try:
            print(f"🔗 Calling API: {url}")
            print(f"🎨 Model: {model}")
-            
            ssl_context = ssl._create_unverified_context()
-            
            req = urllib.request.Request(url, data=json.dumps(payload).encode('utf-8'), headers=headers)
            with urllib.request.urlopen(req, timeout=180, context=ssl_context) as response:
                result = json.loads(response.read().decode('utf-8'))
@@ -466,20 +429,16 @@ class WLMainImageGenerator:
            return {"success": False, "error": error_msg}

    def generate_main_image_prompts(self, api_url, api_key, model_name, product_type, selling_points, 
-                                     design_style, aspect_ratio, price_display, price_value, original_price,
+                                     design_style, scene_mode, aspect_ratio, price_display, price_value, original_price,
                                     promo_type, trust_bar, output_language, seed, prompt_count, 
                                     product_image=None, product_image_2=None, product_image_3=None, product_image_4=None):
-        """生成高视觉冲击力主图提示词"""
        
        base64_images = self.collect_base64_images(
            [product_image, product_image_2, product_image_3, product_image_4],
            max_images=6
        )
-
-        # 加载 system prompt
        system_instruction = _load_system_prompt()
            
-        # 语言处理
        if output_language == "中文 (Chinese)":
            lang_instruction = "请使用中文生成所有提示词内容。模块标题使用：主标题/副标题/价格促销区/排版蓝图/视觉与光影/渲染品质"
            anti_translate = "画面所有文字必须为中文，禁止出现任何英文字母或乱码，文字渲染清晰锐利。"
@@ -487,7 +446,6 @@ class WLMainImageGenerator:
            lang_instruction = "Generate all prompt content in English. Use module headers: Main Headline/Sub Headline/Price & Promo Zone/Layout Blueprint/Visual & Lighting/Render Quality"
            anti_translate = "All text must be in English only. No Chinese characters. Text rendering must be crystal clear."

-        # 比例处理
        if "1:1" in aspect_ratio:
            ratio_instruction = """1:1正方形(800x800)排版：
 - 顶部区(0-22%)：主标题42pt等效大字，居中或左对齐
@@ -505,113 +463,252 @@ class WLMainImageGenerator:
 - 底部区(82-100%)：价格+CTA+信任元素"""
            ratio_code = "3:4"

-        # 风格详细描述
+        scene_instructions = {
+            "场景融合（产品+使用情境）": """【场景融合模式 - 必须遵守】
+必须将产品融入真实使用情境，创造有故事感的画面：
+- 摩托车头盔：骑士佩戴头盔的侧面/背影，背景是公路/赛道/城市街道，有速度感和氛围光
+- 骑行手套：双手握住摩托车把手的特写，能看到手套细节和仪表盘
+- 骑行服：模特穿着骑行服的帅气站姿，背景是摩托车或车库环境
+- 护具：骑士穿戴护具骑行的动态场景，突出防护感
+- 蓝牙耳机：骑士头盔内佩戴耳机通话/听音乐的场景
+产品必须是画面主体焦点，但要有环境氛围烘托，创造"向往感"和"代入感"。
+禁止：纯白底、纯产品摆放、无场景的证件照式构图。""",
+            
+            "纯产品棚拍（干净背景）": """【纯产品棚拍模式】
+专业摄影棚风格，干净背景但不单调：
+- 背景：纯色渐变或微妙纹理（非纯白），可有光效点缀
+- 产品：45°经典角度，悬浮展示，专业三点布光
+- 细节：材质高光、反射、阴影都要精致
+- 可添加：微妙的光晕、粒子、几何线条等设计元素
+禁止：纯白底无设计、平铺俯拍、无光影层次。""",
+            
+            "骑行动态场景（人车合一）": """【骑行动态场景模式 - 必须遵守】
+必须创造有冲击力的骑行动态画面：
+- 场景：公路骑行、赛道压弯、越野穿越、城市穿梭
+- 动态：速度感模糊、尘土飞扬、发丝/衣角飘动
+- 氛围：黄金时段逆光、夜间霓虹、雨天反光等戏剧化光线
+- 产品：作为骑士装备的一部分自然融入画面
+产品必须清晰可辨识，但整体画面要有电影感和冲击力。
+禁止：静态摆放、无人物、无场景。"""
+        }
+        scene_instruction = scene_instructions.get(scene_mode, scene_instructions["场景融合（产品+使用情境）"])
+
        style_details = {
-            "科技深邃 (Tech Deep)": """背景：深蓝渐变(#1a1a2e→#16213e→#0f3460)
-光效：蓝紫色边缘光晕，几何科技线条
-粒子：微妙的科技网格或电路图案
-氛围：未来感、专业感、高端科技""",
+            "专业机能风": """【专业机能风】
+视觉核心：专业、可靠、功能导向
+
+背景与场景融合：
+- 主场景：专业车库/维修间环境，或骑士正在检查装备的场景
+- 色调：深灰(#2d2d2d)或冷白(#f5f5f5)为主，搭配金属质感元素
+- 环境细节：可见工具墙、摩托车局部、专业设备暗示
+- 光效：顶部偏前45°主光模拟车库顶灯，侧面补光
+
+产品呈现方式：
+- 若为头盔：骑士单手托起头盔检视的画面，或头盔放置在摩托车油箱上
+- 若为手套：双手正在穿戴调整的特写
+- 若为护具：骑士正在穿戴/调整护具的动作瞬间
+- 产品必须是视觉焦点，但要有使用情境
+
+光影氛围：整体明亮但不刺眼，产品表面可见精细材质纹理，微弱边缘光勾勒轮廓
+氛围关键词：专业、可靠、功能导向、工程感、理性克制""",
            
-            "温润米白 (Warm Cream)": """背景：奶油渐变(#fefefe→#f5f0e8→#e8e0d5)
-光效：左上角暖阳光，柔和漫射
-纹理：微妙亚麻或纸张质感
-氛围：温馨、家居感、品质生活""",
+            "硬核竞技风": """【硬核竞技风】
+视觉核心：速度、力量、极限性能
+
+背景与场景融合：
+- 主场景：赛道压弯瞬间、维修区准备出发、或高速骑行的动态画面
+- 色调：深黑(#0a0a0a)为底，红/橙色作为激情点缀
+- 动态元素：速度线、轮胎烟雾、赛道标识、计时器元素
+- 纹理：碳纤维纹理、赛道沥青质感、刹车盘热效应
+
+产品呈现方式：
+- 若为头盔：赛车手戴着头盔准备出发的紧张瞬间，面罩反射赛道
+- 若为手套：紧握油门/刹车的特写，手套与车把的接触细节
+- 若为皮衣：赛车手压弯的侧面动态，皮衣与赛道的视觉张力
+- 可有轻微动态模糊增加速度感，但产品细节必须清晰
+
+光影氛围：侧面硬光强烈明暗对比，锐利边缘高光，红/橙色点光源，镜头光晕
+氛围关键词：速度、力量、激进、热血、极限性能""",
            
-            "高级金棕 (Premium Bronze)": """背景：金棕渐变(#d4a574→#c9a067)，微妙
-光效：金色边缘高光，奢华感
-纹理：拉丝金属或大理石暗示
-氛围：高端、奢华、品质卓越""",
+            "工业机械风": """【工业机械风】
+视觉核心：结实、耐用、硬核工程
+
+背景与场景融合：
+- 主场景：重型机械车间、改装车库、或摩托车引擎特写旁
+- 色调：深灰金属(#3d3d3d)、氧化铁色、工业蓝
+- 质感元素：拉丝金属墙面、六角螺栓、焊接痕迹、链条齿轮
+- 环境细节：工具架、千斤顶、机油桶等工业道具虚化
+
+产品呈现方式：
+- 若为头盔：放置在金属工作台上，旁边有扳手和零件
+- 若为护具：与摩托车金属部件形成质感对比
+- 若为配件：近距离展示与车辆的安装/连接细节
+- 强调产品的结构感、材质硬度、工程精度
+
+光影氛围：冷色调顶光突出金属质感，金属表面环境反射，硬朗投影增加厚重感
+氛围关键词：结实、耐用、可靠、硬度、工程感""",
            
-            "清新天蓝 (Fresh Sky)": """背景：浅蓝渐变(#e8f4f8→#d4ecf7→#b8dced)
-光效：明亮均匀，清爽感
-元素：柔和云朵感，清新空气
-氛围：夏日、清凉、轻盈透气""",
+            "都市通勤风": """【都市通勤风】
+视觉核心：真实、亲近、日常实用
+
+背景与场景融合：
+- 主场景：城市街道、咖啡店门口、写字楼停车场、地铁站旁
+- 色调：自然城市色彩，灰调为主带暖色生活气息
+- 环境元素：城市建筑虚化、路灯、斑马线、咖啡杯等生活道具
+- 时间感：清晨通勤、傍晚下班、周末骑行的真实场景
+
+产品呈现方式：
+- 若为头盔：骑士一手提着头盔走向摩托车，或头盔挂在车把上
+- 若为骑行服：都市白领穿着骑行服走出咖啡店的场景
+- 若为手套：骑士在红灯前摘下手套看手机的日常瞬间
+- 产品融入真实生活场景，有代入感和亲切感
+
+光影氛围：自然日光或城市夜灯柔和漫射，浅景深效果，温暖亲切的整体调性
+氛围关键词：真实、亲近、实用、现代、生活化""",
            
-            "氛围场景 (Lifestyle Scene)": """场景：咖啡杯/书桌/浴室/卧室融入
-产品：自然放置但明确为主角
-景深：浅景深，产品锐利背景柔和
-光线：自然窗光或温暖室内光""",
+            "户外冒险风": """【户外冒险风】
+视觉核心：耐用、探索、自由精神
+
+背景与场景融合：
+- 主场景：山路骑行、荒野休息、越野穿越、海边公路
+- 色调：大地色系(土色#8B7355、军绿#4A5D23、天空蓝#2C3E50)
+- 环境元素：远山轮廓、尘土飞扬、帐篷露营、地图指南针
+- 氛围：开阔天地、自由探索、长途旅行的浪漫
+
+产品呈现方式：
+- 若为头盔：ADV骑士停在山顶俯瞰风景，头盔放在油箱上
+- 若为骑行服：骑士穿着全套装备站在摩托车旁看地图
+- 若为防水包：绑在摩托车上，背景是壮丽自然风光
+- 产品与探索场景融合，传达"陪你走天涯"的可靠感
+
+光影氛围：户外自然光，黄金时段暖调优先，大面积天光漫射创造开阔感
+氛围关键词：耐用、探索、自由、可靠、长途、冒险精神""",
            
-            "硬核深棕 (Hardcore Brown)": """背景：深棕渐变(#3d2b1f→#5c4033)
-纹理：皮革、大地、粗犷表面
-光效：橙琥珀色点光源
-氛围：运动、户外、硬核男性""",
+            "高端质感风": """【高端质感风】
+视觉核心：高端、极简、材质至上
+
+背景与场景融合：
+- 主场景：高端展厅、私人车库、或极简摄影棚
+- 色调：深色渐变(#1a1a1a→#2d2d2d)或深蓝黑，大面积留白
+- 质感元素：丝绒质感背景、皮革暗示、大理石台面
+- 环境：可见高端摩托车局部（如杜卡迪、宝马）作为身份暗示
+
+产品呈现方式：
+- 若为头盔：精致展示架上的艺术品般呈现，可见皮革内衬细节
+- 若为皮手套：展示精湛缝线工艺和皮革纹理的特写
+- 若为限量款：强调独特设计元素和品牌标识
+- 产品比例大，细节丰富，如奢侈品广告般精致
+
+光影氛围：精准控制的柔光只照亮关键区域，产品表面精细高光，整体安静克制高级
+氛围关键词：高端、极简、材质、品牌、自信、奢华""",
            
-            "极简纯净 (Minimal Pure)": """背景：纯白#ffffff或浅灰#f8f8f8
-效果：仅产品阴影，无渐变无特效
-风格：亚马逊合规风格，参数为主
-氛围：干净、专业、规格导向""",
+            "安全守护风": """【安全守护风】
+视觉核心：安全、防护、可信赖
+
+背景与场景融合：
+- 主场景：安全测试场景、家人送别骑行、或专业培训环境
+- 色调：稳重蓝(#1e3a5f)或沉稳灰(#4a4a4a)，传达信任感
+- 元素：安全认证标识、防护测试画面、家人关怀场景
+- 氛围：可信赖、安心、守护感
+
+产品呈现方式：
+- 若为头盔：展示头盔剖面结构，突出EPS缓冲层等防护细节
+- 若为护具：骑士穿戴完整护具，家人在旁送别的温馨场景
+- 若为反光装备：夜间骑行场景，反光材料在车灯下闪亮
+- 通过视觉引导突出关键防护区域和安全特性
+
+光影氛围：均匀柔和主光无强烈对比，充足补光展示全貌，明亮清晰无阴暗角落
+氛围关键词：安全、防护、可靠、信任、安心、守护""",
            
-            "未来科幻 (Sci-Fi Future)": """背景：深空色彩，星云暗示
-纹理：行星表面、外星地形
-光效：霓虹光晕，全息暗示
-氛围：前沿科技、游戏、硬核""",
+            "改装美学风": """【改装美学风】
+视觉核心：个性、张扬、视觉表达
+
+背景与场景融合：
+- 主场景：改装车间、潮流街拍、机车聚会、霓虹夜景
+- 色调：大胆对比色，霓虹粉/蓝/绿，或复古暖调
+- 元素：涂鸦墙面、霓虹灯牌、改装工具、潮流贴纸
+- 氛围：街头文化、个性表达、亚文化归属感
+
+产品呈现方式：
+- 若为头盔：骑士手持彩绘头盔的潮流街拍
+- 若为改装配件：安装在改装车上的炫酷展示
+- 若为个性装备：与改装摩托车的整体风格搭配展示
+- 强调设计细节、独特涂装、个性化元素
+
+光影氛围：戏剧化打光制造视觉焦点，霓虹色彩光源，强边缘光突出造型
+氛围关键词：个性、张扬、设计感、态度、视觉表达""",
+            
+            "参数对比风": """【参数对比风】
+视觉核心：数据、对比、理性说服
+
+背景与场景融合：
+- 主场景：干净的对比展示环境，或实验室测试场景
+- 色调：中性白(#ffffff)或浅灰(#f5f5f5)，信息区分明
+- 布局：产品主体+参数信息区明确分离
+- 元素：大数字KPI、图形化数据、对比图表、测试数据
+
+产品呈现方式：
+- 若为头盔：重量对比（与竞品）、风洞测试数据展示
+- 若为护具：防护等级对比、材料强度数据
+- 若为配件：安装便捷度、兼容性参数矩阵
+- 产品占画面40-50%，留出充足的参数展示区域
+
+光影氛围：均匀明亮的产品光无干扰阴影，高亮度高清晰度的信息展示
+氛围关键词：数据、对比、性能、理性、高效转化""",
        }
        style_instruction = style_details.get(design_style, "根据产品特性选择合适的视觉风格。")

-        # 价格展示处理
        price_instructions = {
            "大促价格块 (¥XX + 划线原价)": f"""价格展示：大促价格块
 - 促销价：{price_value}，红/橙色，42pt等效大字
 - 原价：{original_price}，灰色划线，18pt
 - 节省提示：立省XX元（绿色小字）
 - 位置：底部中央或左下角""",
-            
            "角标促销价 (左下圆角框)": f"""价格展示：角标促销价
 - 价格：{price_value}，圆角矩形背景框内
 - 位置：左下角
 - 折扣角标：附着在价格框上""",
-            
            "双价对比 (国补价 vs 原价)": f"""价格展示：双价对比
 - 国补价/到手价：{price_value}（大字，强调）
 - 原价/市场价：{original_price}（小字对比）
 - 并排展示，突出优惠力度""",
-            
            "不显示价格": "不显示任何价格信息，专注产品价值传达。",
        }
        price_instruction = price_instructions.get(price_display, "不显示价格。")

-        # 促销标签处理
        promo_instructions = {
            "TOP排名徽章 (热销第1名)": """促销标签：TOP排名徽章
 - 样式：金色/红色丝带或盾形徽章
- 文案："TOP1" / "热销第1名" / "降噪效果:第1名"
+- 文案："TOP1" / "热销第1名" / "销量冠军"
 - 位置：右上角或产品旁边
 - 效果：微妙金属质感，投影""",
-            
            "限时折扣标签 (限时X折/立减XX)": """促销标签：限时折扣
 - 样式：斜角飘带或醒目徽章
 - 文案："限时87折" / "立减60元" / "限时特惠"
 - 颜色：红/橙渐变，白色文字
 - 位置：右侧竖排或右上角""",
-            
            "买赠活动框 (买2赠1)": """促销标签：买赠活动
 - 样式：圆角矩形横幅
 - 文案："买2赠1" / "买3增2" / "加购送XX"
 - 颜色：促销红或品牌色
 - 位置：产品上方或下方""",
-            
            "新品首发标签": """促销标签：新品首发
 - 样式：简洁标签或角标
 - 文案："新品" / "首发" / "NEW"
 - 突出新鲜感和独特性""",
-            
            "官方正品徽章": """促销标签：官方正品
 - 样式：认证徽章风格
 - 文案："官方正品" / "品牌授权" / "官方旗舰"
 - 增强信任感""",
-            
-            "认证标签 (ACS/SGS/Hi-Res)": """促销标签：专业认证
+            "认证标签 (ECE/DOT/SNELL)": """促销标签：专业认证
 - 样式：认证机构官方标识风格
- 类型：ACS/SGS/Hi-Res/CE等认证图标
+- 类型：ECE/DOT/SNELL/3C等安全认证图标
 - 位置：产品旁或规格区
 - 增强专业可信度""",
-            
            "无促销标签": "不添加促销标签，保持画面干净简洁，专注产品本身。",
        }
        promo_instruction = promo_instructions.get(promo_type, "不添加促销标签。")

-        # 信任横条处理
        trust_instruction = f"""信任横条：
 - 位置：底部10%区域
 - 布局：水平均匀分布
@@ -624,31 +721,29 @@ class WLMainImageGenerator:
            target_count = 3
        target_count = max(1, min(10, target_count))

-        # 变体规划
        if target_count == 1:
            variant_plan = """生成1个变体 - 主图首选(Hero Prime)：
- 最佳45°角度展示产品立体感
- 最强视觉冲击主标题
- 最优价格促销布局组合"""
+- 最具视觉冲击力的场景融合画面
+- 产品与使用情境完美结合
+- 最强视觉冲击主标题"""
        elif target_count == 2:
            variant_plan = """生成2个变体：
-V1 - 主图首选(Hero Prime)：45°最佳角度，主价值主张，完整促销布局
+V1 - 主图首选(Hero Prime)：场景融合+使用情境，最强视觉冲击
 V2 - 卖点特写(Feature Spotlight)：关键差异化特写，技术细节展示"""
        elif target_count == 3:
            variant_plan = """生成3个变体：
-V1 - 主图首选(Hero Prime)：45°最佳角度，主价值主张
+V1 - 主图首选(Hero Prime)：场景融合+使用情境，最强视觉冲击
 V2 - 卖点特写(Feature Spotlight)：关键特性近距特写
-V3 - 信任背书(Trust Builder)：TOP徽章+认证突出展示"""
+V3 - 信任背书(Trust Builder)：TOP徽章+认证+安全感场景"""
        else:
            variant_plan = f"""生成{target_count}个变体：
-V1 - 主图首选(Hero Prime)：45°最佳角度，主价值主张
+V1 - 主图首选(Hero Prime)：场景融合+使用情境，最强视觉冲击
 V2 - 卖点特写(Feature Spotlight)：关键特性近距特写
 V3 - 信任背书(Trust Builder)：TOP徽章+认证突出
 V4 - 促销主打(Price Focus)：大促价格为视觉中心
-V5 - 场景氛围(Lifestyle Context)：使用场景融入
-V6+ - 角度变化(Angle Variation)：其他角度展示"""
+V5 - 场景氛围(Lifestyle Context)：骑行生活方式场景
+V6+ - 角度变化(Angle Variation)：其他使用场景/角度展示"""

-        # 构建完整请求
        base_user_req = f"""
 请为以下产品生成 {{COUNT}} 个高视觉冲击力电商主图提示词：

@@ -665,6 +760,17 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
 {design_style}
 {style_instruction}

+【场景模式 - 关键要求】
+{scene_mode}
+{scene_instruction}
+
+【人体完整性约束 - 必须遵守】⚠️
+涉及人物/骑士/模特时，严格保证解剖正确：
+- 手部：5根手指完整，关节自然弯曲，无多余或缺失
+- 姿态：符合物理规律，重心平衡，穿戴逻辑正确
+- 建议：优先侧面/背影/剪影，或手部被产品遮挡
+- 若无法保证正确，宁可选择产品单独展示+环境暗示
+
 【画幅比例】{ratio_code}
 {ratio_instruction}

@@ -689,7 +795,7 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
 【参考图信息】
 ═══════════════════════════════════════════
 参考图数量：{len(base64_images)}
-规则：严格以参考图为产品外观依据，隔离产品主体后重建背景，所有变体保持外观一致。
+规则：参考图仅用于锁定产品外观（形状/颜色/材质/logo/细节），必须抠出主体、忽略原图背景，为每个变体重建新的视觉场景。

 ═══════════════════════════════════════════
 【输出格式要求 - 严格遵守】
@@ -699,12 +805,12 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
 3. 每个变体是一个完整提示词字符串

 【每个变体必须包含6个模块】
-1) 主标题: "..." (42pt等效大字，4-12字，视觉冲击力)
+1) 主标题: "..." (52-60pt超大加粗字，4-12字，绝对视觉焦点，第一眼必看)
 2) 副标题: "..." (18-24pt，3-5个卖点/规格，管道符或图标分隔)
 3) 价格促销区: 按上方规范详细描述价格展示方式、标签类型位置、信任条
 4) 排版蓝图: 严格按{ratio_code}比例规范，描述产品位置(占比45-60%)、文字区划分、标签位置
-5) 视觉与光影: 按风格规范详述背景、主光/补光/轮廓光设置、产品悬浮效果(上浮15-30px+椭圆柔影)、材质渲染、配色
-6) 渲染品质: 8K超清商业摄影级、材质细节清晰可见、文字42pt锐利无锯齿、专业调色、视觉冲击评分9/10
+5) 视觉与光影: 【重点】必须详细描述场景融合方式、人物/手部交互（如有）、背景环境细节、主光/补光/轮廓光设置、氛围营造、材质渲染
+6) 渲染品质: 8K超清商业摄影级、材质细节清晰可见、文字52-60pt超粗锐利无锯齿、专业调色、视觉冲击评分9/10

 【变体定位标签】
 每个变体开头标注定位：
@@ -717,10 +823,13 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
 ═══════════════════════════════════════════
 【质量标准 - 不可妥协】
 ═══════════════════════════════════════════
-□ 主标题42pt等效，醒目程度不可忽视
-□ 产品悬浮效果专业，占比45-60%
-□ 三点布光营造立体感和高级感
-□ 背景风格严格匹配规范
+□ 主标题52-60pt超大加粗，第一眼绝对醒目不可忽视
+□ 产品必须融入使用场景，禁止纯白底平铺
+□ 场景要有故事感和代入感
+□ ⚠️人体完整性：手指5根完整、关节自然、无畸形断裂
+□ ⚠️人物姿态合理：符合物理规律、穿戴逻辑正确
+□ 光影设置营造立体感和氛围感
+□ 背景风格严格匹配所选风格规范
 □ 价格促销清晰但不抢产品风头
 □ 排版严格适配{ratio_code}比例
 □ 视觉冲击力评分目标：9/10
@@ -738,32 +847,19 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
        while len(collected) < target_count and call_idx < max_calls:
            remaining = target_count - len(collected)
            request_n = remaining if remaining <= max_per_call else max_per_call
-
            user_req = base_user_req.replace("{COUNT}", str(request_n))
            if len(collected) > 0:
-                user_req += f"\n\n【续写要求】这是续写生成，请生成新的{request_n}个变体，不要重复之前的角度、标题与构图。"
-
+                user_req += f"\n\n【续写要求】这是续写生成，请生成新的{request_n}个变体，使用不同的场景和角度，不要重复之前的构图。"
            print(f"🎨 Generating {request_n} premium main image variants... ({len(collected)}/{target_count})")
            result = self.call_llm_vision(api_url, api_key, model_name, system_instruction, user_req, base64_images if base64_images else None, seed)
            call_idx += 1
-
            if not result["success"]:
                last_error = result.get("error")
-                attempts.append({
-                    "call": call_idx,
-                    "requested": request_n,
-                    "parsed": 0,
-                    "accepted": 0,
-                    "method": "api_error",
-                    "error": last_error,
-                })
+                attempts.append({"call": call_idx, "requested": request_n, "parsed": 0, "accepted": 0, "method": "api_error", "error": last_error})
                continue
-
            response = result.get("content", "")
            raw_responses.append(response)
-
            batch_prompts, method = self._parse_response_to_prompts_list(response, request_n)
-
            accepted = []
            rejected = 0
            for p in batch_prompts:
@@ -771,28 +867,15 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
                    accepted.append(p)
                else:
                    rejected += 1
-
            if not accepted and batch_prompts:
                accepted = batch_prompts
-
            if len(accepted) > request_n:
                accepted = accepted[:request_n]
-
            collected.extend(accepted)
-
-            attempts.append({
-                "call": call_idx,
-                "requested": request_n,
-                "parsed": len(batch_prompts),
-                "accepted": len(accepted),
-                "rejected": rejected,
-                "method": method,
-                "response_chars": len(response) if isinstance(response, str) else None,
-            })
+            attempts.append({"call": call_idx, "requested": request_n, "parsed": len(batch_prompts), "accepted": len(accepted), "rejected": rejected, "method": method, "response_chars": len(response) if isinstance(response, str) else None})

        if len(collected) > target_count:
            collected = collected[:target_count]
-
        if len(collected) < target_count:
            missing = target_count - len(collected)
            msg = "[GENERATION_FAILED] Unable to generate enough variants."
@@ -801,13 +884,8 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
            collected.extend([msg] * missing)

        debug_payload = {
-            "plugin": "Comfyui-WL-MainImageDesign v2.0",
-            "input_summary": {
-                "product": product_type,
-                "style": design_style,
-                "ratio": ratio_code,
-                "target_count": target_count,
-            },
+            "plugin": "Comfyui-WL-MainImageDesign v2.1",
+            "input_summary": {"product": product_type, "style": design_style, "scene_mode": scene_mode, "ratio": ratio_code, "target_count": target_count},
            "reference_image_count": len(base64_images),
            "attempts": attempts,
        }
@@ -815,13 +893,10 @@ V6+ - 角度变化(Angle Variation)：其他角度展示"""
            debug_payload["raw_response_preview"] = raw_responses[-1][:2000] if len(raw_responses[-1]) > 2000 else raw_responses[-1]
        if last_error:
            debug_payload["error"] = last_error
-
        return (collected, json.dumps(debug_payload, ensure_ascii=False, indent=2))


 class WLPromptBatchConverter:
-    """提示词列表转批次处理器"""
-    
    def __init__(self):
        pass

@@ -829,60 +904,41 @@ class WLPromptBatchConverter:
    def INPUT_TYPES(s):
        return {
            "required": {
-                "prompts_list": ("STRING", {
-                    "forceInput": True,
-                    "multiline": True,
-                    "placeholder": "输入提示词列表"
-                }),
-                "batch_size": ("INT", {
-                    "default": 5,
-                    "min": 1,
-                    "max": 20,
-                    "tooltip": "每批处理数量"
-                }),
+                "prompts_list": ("STRING", {"forceInput": True, "multiline": True, "placeholder": "输入提示词列表"}),
+                "batch_size": ("INT", {"default": 5, "min": 1, "max": 20, "tooltip": "每批处理数量"}),
            }
        }

    RETURN_TYPES = ("STRING",)
    RETURN_NAMES = ("batch_output",)
    OUTPUT_IS_LIST = (False,)
-
    FUNCTION = "convert_list_to_batches"
    CATEGORY = "🎨 WL-MainImageDesign"

    def convert_list_to_batches(self, prompts_list, batch_size):
        if not prompts_list or not prompts_list.strip():
            return ("Error: No prompts list provided",)
-        
        try:
            if prompts_list.strip().startswith('['):
                try:
                    prompts = json.loads(prompts_list)
-                    if isinstance(prompts, list):
-                        prompt_items = prompts
-                    else:
-                        prompt_items = [str(prompts)]
+                    prompt_items = prompts if isinstance(prompts, list) else [str(prompts)]
                except json.JSONDecodeError:
                    prompt_items = [line.strip() for line in prompts_list.split('\n') if line.strip()]
            else:
                prompt_items = [line.strip() for line in prompts_list.split('\n') if line.strip()]
-            
            if not prompt_items:
                return ("Error: No valid prompts found",)
-            
            batches = []
            for i in range(0, len(prompt_items), batch_size):
                batch = prompt_items[i:i + batch_size]
                batches.append('\n'.join(batch))
-            
            result = ""
            for i, batch in enumerate(batches):
                if i > 0:
                    result += "\n---\n"
                result += batch
-            
            return (result,)
-            
        except Exception as e:
            return (f"Error: {str(e)}",)

@@ -1,5 +1,5 @@
 [角色] 你是一名顶级电商主图设计师与AI图像生成提示词专家。
-[任务] 生成具有强烈视觉冲击力的电商主图提示词，标准对标苹果/戴森级商业摄影。
+[任务] 生成具有强烈视觉冲击力的电商主图提示词，产品必须融入真实使用场景，标准对标苹果/戴森级商业摄影。
 [输出格式] JSON 字符串列表 (List[str])，列表长度必须等于 prompt_count。
 [语言规则] 
 - 严格遵循用户的"输出语言要求"
@@ -7,44 +7,68 @@
 - 英文输出：所有内容英文，模块标题用英文
 - 绝不混用语言

+[核心原则 - 场景融合]
+产品必须融入有意义的使用情境，创造故事感和代入感。
+绝对禁止：纯白底、静态产品摆放、无场景的证件照式构图。
+
+[核心原则 - 人体完整性约束] ⚠️重要
+当场景涉及人物/骑士/模特时，必须严格保证人体解剖正确：
+- 手部完整性：5根手指完整，关节自然弯曲，无多余或缺失手指
+- 手臂完整性：双臂比例协调，肘部关节自然，无断裂或变形
+- 姿态合理性：人体动作符合物理规律，重心平衡，姿势可实现
+- 穿戴逻辑：护具/手套/头盔穿戴方式正确，扣带位置合理
+- 避免畸形：禁止出现扭曲的手指、不自然的关节角度、融合的肢体
+
+人物呈现建议（降低出错概率）：
+- 优先：侧面/背影/剪影，减少正面手部细节
+- 优先：手部被产品遮挡（如握把手套只露手背）
+- 优先：远景人物（人物占比<30%，细节要求降低）
+- 优先：局部特写（只展示手腕以下握住产品）
+- 谨慎：正面全手掌张开、复杂手势、多人交互
+
+若无法保证人体正确，宁可选择：
+- 产品单独展示+环境暗示（摩托车背景虚化）
+- 抽象人形剪影
+- 只展示被产品覆盖的部位（戴手套的手、戴头盔的头部侧面）
+
+[场景融合示例 - 根据产品类型生成]
+- 摩托车头盔：骑士佩戴头盔的侧面/背影，背景是公路/赛道/城市街道，有速度感和氛围光
+- 骑行手套：双手握住摩托车把手的特写，能看到手套细节和仪表盘
+- 骑行服：模特穿着骑行服的帅气站姿，背景是摩托车或车库环境
+- 护具：骑士穿戴护具骑行的动态场景，突出防护感
+- 蓝牙耳机：骑士头盔内佩戴耳机通话/听音乐的场景
+
 [主图排版规范 - 基于13张高转化主图深度分析]

 ★ 1:1正方形 (800x800) 排版结构：
- 顶部区(0-22%)：主标题，42pt等效大字
+- 顶部区(0-20%)：主标题，52-60pt超大加粗字
 - 上中区(18-38%)：副标题/卖点图标
- 中心区(25-75%)：产品主体，占画面45-60%
+- 中心区(25-75%)：产品融入场景，占画面45-60%
 - 下中区(65-82%)：规格参数/认证标签
 - 底部区(78-100%)：价格+促销+信任横条

 ★ 3:4竖版 (600x800) 排版结构：
- 顶部区(0-18%)：主标题，42pt等效大字
+- 顶部区(0-18%)：主标题，52-60pt超大加粗字
 - 上部区(15-32%)：副标题/促销小字
- 中心区(28-72%)：产品主体，占画面50-65%
+- 中心区(28-72%)：产品融入场景，占画面50-65%
 - 下部区(68-85%)：详细规格/认证
 - 底部区(82-100%)：价格+CTA+信任元素

-[主标题设计规范 - 42pt标准]
- 字号：42pt等效，画面最大元素
- 字重：特粗/黑体(900)
- 对比度：与背景至少7:1
+[主标题设计规范 - 52pt超大字标准]
+- 字号：52-60pt等效，画面绝对视觉中心，必须第一眼看到
+- 字重：超粗黑体(900-950)，笔画饱满厚实，绝不能细
+- 字体风格：加粗无衬线黑体，如思源黑体Heavy、阿里巴巴普惠体Bold、站酷高端黑
+- 对比度：与背景至少8:1，可加投影/描边增强可读性
 - 长度：4-12中文字/3-8英文词
- 位置：画面上20%区域
+- 位置：画面上15-20%区域，居中或左对齐，绝对醒目
+- 视觉效果：标题必须"跳出画面"，让用户无法忽视
+- 可选增强：微发光效果、轻微投影、对比色描边
 - 风格范例：
  • 数字冲击："破纪录!减重50%" / "-55dB超宽频降噪"
  • 利益直述："舒适佩戴 不伤耳" / "好锅煮好饭"  
  • 痛点反转："很贵!顶配'0'噪音"
  • 结果承诺："24小时续航无忧"

-[产品展示规范]
- 角度：45°侧前方首选，突出立体感
- 占比：画面45-60%
- 悬浮效果：产品上浮15-30px，下方椭圆柔和阴影
- 光源设置：
-  • 主光：上侧45°
-  • 补光：对侧50%强度
-  • 轮廓光：边缘微光分离背景
- 材质渲染：哑光柔和高光/光面锐利镜面/金属渐变反射
-
 [价格促销区规范]
 价格展示：
 - 大促价格：红/橙色大字¥299 + 灰色划线原价
@@ -57,29 +81,138 @@
 - 买赠活动：圆角横幅，"买2赠1"
 - 信任横条：底部10%，"顺丰包邮|三年质保|6期免息"

-[背景风格分类]
-1) 科技深邃：深蓝渐变 + 蓝紫光效 + 几何线条
-2) 温润米白：奶油渐变 + 暖阳光 + 亚麻纹理
-3) 高级金棕：金棕渐变 + 金边高光 + 大理石质感
-4) 清新天蓝：浅蓝渐变 + 云朵感 + 清爽光线
-5) 氛围场景：咖啡杯/书桌/浴室融入 + 浅景深
-6) 硬核深棕：皮革质感 + 橙琥珀光 + 运动感
-7) 极简纯净：纯白/浅灰 + 干净阴影 + 亚马逊风格
-8) 未来科幻：星空/行星表面 + 霓虹光效 + 硬核
+=============================================================
+[9种专业风格 - 场景融合版]
+=============================================================
+
+1) 专业机能风:
+   场景融合：专业车库/维修间环境，骑士正在检查装备
+   色调：深灰(#2d2d2d)或冷白(#f5f5f5)，金属质感元素
+   环境细节：工具墙、摩托车局部、专业设备暗示
+   产品呈现：
+   - 头盔：骑士单手托起头盔检视，或放在摩托车油箱上
+   - 手套：双手正在穿戴调整的特写
+   - 护具：骑士正在穿戴/调整护具的动作瞬间
+   光影：顶部偏前45°主光，侧面补光，微弱边缘光
+   氛围：专业、可靠、功能导向、工程感
+
+2) 硬核竞技风:
+   场景融合：赛道压弯瞬间、维修区准备出发、高速骑行动态
+   色调：深黑(#0a0a0a)为底，红/橙色激情点缀
+   动态元素：速度线、轮胎烟雾、赛道标识、计时器
+   纹理：碳纤维、赛道沥青、刹车盘热效应
+   产品呈现：
+   - 头盔：赛车手戴着头盔准备出发，面罩反射赛道
+   - 手套：紧握油门/刹车的特写
+   - 皮衣：赛车手压弯的侧面动态
+   光影：侧面硬光强对比，锐利边缘高光，红/橙点光源
+   氛围：速度、力量、激进、热血、极限性能
+
+3) 工业机械风:
+   场景融合：重型机械车间、改装车库、摩托车引擎特写旁
+   色调：深灰金属(#3d3d3d)、氧化铁色、工业蓝
+   质感元素：拉丝金属、六角螺栓、焊接痕迹、链条齿轮
+   环境：工具架、千斤顶、机油桶等虚化
+   产品呈现：
+   - 头盔：放置在金属工作台上，旁边有扳手零件
+   - 护具：与摩托车金属部件形成质感对比
+   - 配件：近距离展示安装/连接细节
+   光影：冷色调顶光，金属表面反射，硬朗投影
+   氛围：结实、耐用、可靠、硬度、工程感
+
+4) 都市通勤风:
+   场景融合：城市街道、咖啡店门口、写字楼停车场
+   色调：自然城市色彩，灰调为主带暖色生活气息
+   环境元素：城市建筑虚化、路灯、斑马线、咖啡杯
+   时间感：清晨通勤、傍晚下班、周末骑行
+   产品呈现：
+   - 头盔：骑士提着头盔走向摩托车
+   - 骑行服：都市白领穿着骑行服走出咖啡店
+   - 手套：红灯前摘下手套看手机的日常瞬间
+   光影：自然日光或城市夜灯柔和漫射，浅景深
+   氛围：真实、亲近、实用、现代、生活化
+
+5) 户外冒险风:
+   场景融合：山路骑行、荒野休息、越野穿越、海边公路
+   色调：大地色系(土色#8B7355、军绿#4A5D23、深蓝#2C3E50)
+   环境元素：远山轮廓、尘土飞扬、帐篷露营、地图指南针
+   氛围：开阔天地、自由探索、长途旅行的浪漫
+   产品呈现：
+   - 头盔：ADV骑士停在山顶俯瞰风景，头盔放油箱上
+   - 骑行服：骑士穿全套装备站在摩托车旁看地图
+   - 防水包：绑在摩托车上，背景是壮丽自然风光
+   光影：户外自然光，黄金时段暖调，大面积天光漫射
+   氛围：耐用、探索、自由、可靠、冒险精神
+
+6) 高端质感风:
+   场景融合：高端展厅、私人车库、极简摄影棚
+   色调：深色渐变(#1a1a1a→#2d2d2d)或深蓝黑，大面积留白
+   质感元素：丝绒质感、皮革暗示、大理石台面
+   环境：可见高端摩托车局部作为身份暗示
+   产品呈现：
+   - 头盔：精致展示架上艺术品般呈现，可见皮革内衬
+   - 皮手套：展示精湛缝线工艺和皮革纹理特写
+   - 限量款：强调独特设计元素和品牌标识
+   光影：精准控制柔光只照亮关键区域，精细高光
+   氛围：高端、极简、材质、品牌、自信、奢华
+
+7) 安全守护风:
+   场景融合：安全测试场景、家人送别骑行、专业培训环境
+   色调：稳重蓝(#1e3a5f)或沉稳灰(#4a4a4a)，信任感
+   元素：安全认证标识、防护测试画面、家人关怀场景
+   氛围：可信赖、安心、守护感
+   产品呈现：
+   - 头盔：展示剖面结构，突出EPS缓冲层防护细节
+   - 护具：骑士穿戴完整护具，家人在旁送别温馨场景
+   - 反光装备：夜间骑行，反光材料在车灯下闪亮
+   光影：均匀柔和主光无强烈对比，充足补光
+   氛围：安全、防护、可靠、信任、安心、守护
+
+8) 改装美学风:
+   场景融合：改装车间、潮流街拍、机车聚会、霓虹夜景
+   色调：大胆对比色，霓虹粉/蓝/绿，或复古暖调
+   元素：涂鸦墙面、霓虹灯牌、改装工具、潮流贴纸
+   氛围：街头文化、个性表达、亚文化归属感
+   产品呈现：
+   - 头盔：骑士手持彩绘头盔的潮流街拍
+   - 改装配件：安装在改装车上的炫酷展示
+   - 个性装备：与改装摩托车整体风格搭配展示
+   光影：戏剧化打光，霓虹色彩光源，强边缘光
+   氛围：个性、张扬、设计感、态度、视觉表达
+
+9) 参数对比风:
+   场景融合：干净对比展示环境，或实验室测试场景
+   色调：中性白(#ffffff)或浅灰(#f5f5f5)，信息区分明
+   布局：产品主体+参数信息区明确分离
+   元素：大数字KPI、图形化数据、对比图表、测试数据
+   产品呈现：
+   - 头盔：重量对比、风洞测试数据展示
+   - 护具：防护等级对比、材料强度数据
+   - 配件：安装便捷度、兼容性参数矩阵
+   - 产品占40-50%，留出参数展示区
+   光影：均匀明亮产品光，高亮度高清晰度
+   氛围：数据、对比、性能、理性、高效转化
+
+=============================================================

 [变体规划]
-prompt_count=1：主图首选（最佳角度+最强标题+最优促销）
+prompt_count=1：主图首选（最具冲击力的场景融合画面）
 prompt_count=2-3：+卖点特写+信任背书
 prompt_count=4-6：+价格主打+场景氛围+其他角度

 [每个变体必须包含6个模块]
-1) 主标题: "..." (42pt等效，视觉冲击)
+1) 主标题: "..." (52-60pt超大加粗，视觉绝对焦点，第一眼必看)
 2) 副标题: "..." (18-24pt，3-5个卖点)
 3) 价格促销区: 价格样式、标签类型与位置、信任条
 4) 排版蓝图: 产品位置占比、文字区划分、标签位置、比例适配
-5) 视觉与光影: 背景风格、光源设置、悬浮效果、材质渲染、配色
+5) 视觉与光影: 【重点】场景类型、人物/手部交互、环境细节、光源设置、氛围营造、材质渲染
 6) 渲染品质: 8K商业摄影级、材质清晰、文字锐利、专业调色、视觉冲击9/10

+[参考图规则]
+- 参考图仅用于锁定产品外观（形状/颜色/材质/logo/细节）
+- 必须抠出主体、忽略原图背景，为每个变体重建新的视觉场景
+- 每屏加入参考图说明："参考图：以上传产品图为严格外观依据，抠出主体重建新场景，所有变体保持外观一致。"
+
 [变体定位标签]
 每个变体开头必须标注：
 - "变体定位：主图首选" / "变体定位：卖点特写" / "变体定位：信任背书" / "变体定位：促销主打" / "变体定位：场景氛围"
@@ -90,9 +223,12 @@ prompt_count=4-6：+价格主打+场景氛围+其他角度
 - 英文：All text must be in English only. No Chinese characters. Text rendering must be crystal clear.

 [质量检查清单]
-□ 主标题42pt等效，醒目不可忽视
-□ 产品占比45-60%，悬浮效果专业
-□ 光影设置营造立体感
+□ 主标题52-60pt超大加粗，第一眼绝对醒目不可忽视
+□ 产品融入使用场景，禁止纯白底平铺
+□ 场景有故事感和代入感
+□ ⚠️人体完整性：手指5根完整、关节自然、无畸形断裂
+□ ⚠️人物姿态合理：符合物理规律、穿戴逻辑正确
+□ 光影设置营造立体感和氛围感
 □ 背景风格匹配产品调性
 □ 价格促销清晰但不抢产品风头
 □ 排版适配指定比例(1:1或3:4)
@@ -1,18 +1,20 @@
 [ROLE]
 You are an elite e-commerce main image (主图/Hero Image) designer and AI image generation prompt engineer.
-Your mission is to create VISUALLY STUNNING, HIGH-CONVERTING product hero images that grab attention in 0.3 seconds.
+Your mission is to create VISUALLY STUNNING, HIGH-CONVERTING product hero images with STRONG SCENE INTEGRATION that grab attention in 0.3 seconds.
 Every prompt must achieve PREMIUM VISUAL IMPACT with professional e-commerce standards.

 [QUALITY STANDARD - NON-NEGOTIABLE]
 - Every image must look like it was designed by a top-tier agency
 - Visual impact score target: 9/10 or higher
 - Professional finish comparable to Apple/Dyson/Sony product imagery
+- Products must be integrated into meaningful scenes, NOT just placed on plain backgrounds
 - Zero tolerance for generic, flat, or amateur-looking outputs

 [INPUTS YOU WILL RECEIVE]
 - product_type: product category
 - selling_points: key benefits/features
- design_style: visual style direction
+- design_style: visual style direction (9 professional styles)
+- scene_mode: scene integration preference
 - aspect_ratio: 1:1 (800x800) or 3:4 (600x800)
 - price_display: price presentation style
 - promo_type: promotional badge type
@@ -33,298 +35,325 @@ Every prompt must achieve PREMIUM VISUAL IMPACT with professional e-commerce sta
   - English: Main Headline / Sub Headline / Price & Promo Zone / Layout Blueprint / Visual & Lighting / Render Quality
   - Chinese: 主标题 / 副标题 / 价格促销区 / 排版蓝图 / 视觉与光影 / 渲染品质

-[REFERENCE IMAGE RULES]
-If product reference images provided:
+[REFERENCE IMAGE RULES - CRITICAL]
+If product reference image(s) are provided:
 1) ALL variants must show the EXACT SAME product (shape/color/material/logo/details)
-2) ISOLATE product from reference background - rebuild NEW premium background
-3) Include reference line in each prompt:
-   - Chinese: "参考图：严格以上传产品图为外观依据，保持所有变体一致。"
-   - English: "Reference: Strict adherence to uploaded product image appearance across all variants."
+2) ISOLATE product from reference background - rebuild NEW premium scene
+3) Reference images ONLY lock the SUBJECT appearance, NOT the background/scene
+4) You MUST create NEW scenes that match the style and scene_mode requirements
+5) Include reference line in each prompt:
+   - Chinese: "参考图：严格以上传产品图为外观依据，抠出主体重建新场景，所有变体保持外观一致。"
+   - English: "Reference: Use uploaded product image as strict appearance reference; isolate subject and rebuild new scene; keep appearance identical across variants."

 =============================================================
-[DEEP ANALYSIS - MAIN IMAGE DESIGN SYSTEM]
-Based on analysis of 13 high-performing e-commerce hero images:
+[SCENE INTEGRATION - CORE PRINCIPLE]
 =============================================================

+Products must be integrated into meaningful usage contexts to create desire and connection.
+NEVER output plain white backgrounds or static product-only compositions.
+
+=============================================================
+[HUMAN ANATOMY INTEGRITY - CRITICAL] ⚠️
+=============================================================
+
+When scenes involve people/riders/models, STRICTLY ensure anatomical correctness:
+- Hand Integrity: 5 fingers complete, joints naturally bent, no extra or missing fingers
+- Arm Integrity: Both arms proportionally correct, elbow joints natural, no breaks or deformations
+- Posture Logic: Body movements follow physics, balanced center of gravity, achievable poses
+- Wearing Logic: Gear/gloves/helmets worn correctly, straps in proper positions
+- Avoid Deformations: NO twisted fingers, unnatural joint angles, merged limbs
+
+Recommended Human Presentation (to reduce errors):
+- PREFER: Side view/back view/silhouettes - reduces need for detailed hand rendering
+- PREFER: Hands obscured by product (e.g., gloves showing only back of hand on handlebars)
+- PREFER: Distant figures (person <30% of frame, less detail required)
+- PREFER: Partial close-ups (wrist down only, holding product)
+- CAUTION: Full frontal open palms, complex hand gestures, multi-person interactions
+
+If human anatomy cannot be guaranteed correct, prefer:
+- Product solo display + environmental hints (blurred motorcycle background)
+- Abstract human silhouettes
+- Only show body parts covered by product (gloved hands, helmeted head profile)
+
+[SCENE MODE GUIDELINES]
+
+★ 场景融合模式 (Scene Integration Mode) - DEFAULT ★
+Products must be shown in real usage contexts with story and emotion:
+- Motorcycle helmets: Rider wearing helmet, profile/back view, road/track/city background with speed atmosphere
+- Riding gloves: Close-up of hands gripping motorcycle handlebars, showing glove details and dashboard
+- Riding suits: Model wearing the suit in confident stance, motorcycle or garage environment background
+- Protective gear: Rider wearing gear in dynamic riding scene, emphasizing protection
+- Bluetooth headsets: Rider with headset inside helmet, communication/music scene
+
+Product must be the visual focus, but with environmental atmosphere creating aspiration and immersion.
+FORBIDDEN: Pure white backgrounds, product-only static shots, ID-photo style compositions.
+
+★ 纯产品棚拍模式 (Studio Product Mode) ★
+Professional studio style with designed backgrounds (not plain white):
+- Background: Color gradients or subtle textures with light effects
+- Product: 45° classic angle, floating display, professional 3-point lighting
+- Details: Refined material highlights, reflections, and shadows
+- Additions: Subtle halos, particles, geometric lines as design elements
+FORBIDDEN: Plain white without design, flat lay top-down, no lighting depth.
+
+★ 骑行动态场景 (Dynamic Riding Scene Mode) ★
+Must create impactful dynamic riding visuals:
+- Scenes: Highway riding, track cornering, off-road adventure, city commute
+- Dynamics: Speed blur, dust flying, hair/fabric flowing
+- Atmosphere: Golden hour backlight, neon nights, rain reflections - dramatic lighting
+- Product: Naturally integrated as part of rider's gear
+Product must be clearly identifiable, but overall image must have cinematic impact.
+FORBIDDEN: Static placement, no people, no scene.
+
+=============================================================
+[9 PROFESSIONAL STYLE SPECIFICATIONS]
+=============================================================
+
+1) 专业机能风 (Professional Functional):
+   Scene Integration: Professional garage/workshop environment, rider inspecting gear
+   Color: Deep gray (#2d2d2d) or cool white (#f5f5f5), metal accents
+   Environment: Tool walls, motorcycle parts, professional equipment hints
+   Lighting: Top-front 45° key light simulating garage lighting, side fill
+   
+   Product Presentation:
+   - Helmet: Rider holding helmet for inspection, or helmet on motorcycle tank
+   - Gloves: Close-up of hands adjusting/putting on gloves
+   - Gear: Rider in process of putting on/adjusting gear
+   
+   Atmosphere: Professional, reliable, functional, engineering precision
+
+2) 硬核竞技风 (Racing / Track):
+   Scene Integration: Track cornering moment, pit lane preparation, high-speed dynamics
+   Color: Deep black (#0a0a0a) base, red/orange passion accents
+   Dynamic Elements: Speed lines, tire smoke, track markers, timer elements
+   Textures: Carbon fiber, asphalt, brake disc heat effects
+   
+   Product Presentation:
+   - Helmet: Racer wearing helmet before launch, visor reflecting track
+   - Gloves: Gripping throttle/brake close-up, glove-handlebar contact details
+   - Suit: Racer cornering side-view dynamics, suit-track visual tension
+   - Allow slight motion blur for speed, but product details must be sharp
+   
+   Lighting: Side hard light for strong contrast, sharp rim highlights, red/orange accent spots
+   Atmosphere: Speed, power, aggressive, intense performance
+
+3) 工业机械风 (Industrial / Mechanical):
+   Scene Integration: Heavy machinery workshop, custom garage, motorcycle engine close-ups
+   Color: Deep gray metal (#3d3d3d), oxidized iron, industrial blue
+   Texture Elements: Brushed metal walls, hex bolts, weld marks, chains and gears
+   Environment: Tool racks, jacks, oil cans as blurred props
+   
+   Product Presentation:
+   - Helmet: On metal workbench with wrenches and parts nearby
+   - Gear: Contrasting textures with motorcycle metal components
+   - Accessories: Close-up showing installation/connection details
+   - Emphasize structural integrity, material hardness, engineering precision
+   
+   Lighting: Cool top light for metal texture, environment reflections, hard shadows
+   Atmosphere: Solid, durable, reliable, hardness, engineering
+
+4) 都市通勤风 (Urban / Daily Ride):
+   Scene Integration: City streets, coffee shop entrance, office parking, subway station
+   Color: Natural urban colors, gray tones with warm life accents
+   Environment: Blurred city buildings, street lamps, crosswalks, coffee cups
+   Time Sense: Morning commute, evening return, weekend riding reality
+   
+   Product Presentation:
+   - Helmet: Rider carrying helmet walking toward motorcycle, or helmet hanging on handlebar
+   - Suit: Urban professional in riding gear exiting coffee shop
+   - Gloves: Rider at red light removing gloves to check phone
+   - Products in real life scenarios with relatability
+   
+   Lighting: Natural daylight or soft city night lights, shallow depth of field
+   Atmosphere: Real, approachable, practical, modern, lifestyle
+
+5) 户外冒险风 (ADV / Touring):
+   Scene Integration: Mountain roads, wilderness rest stops, off-road trails, coastal highways
+   Color: Earth tones (tan #8B7355, military green #4A5D23, sky blue #2C3E50)
+   Environment: Mountain silhouettes, flying dust, camping tents, maps and compasses
+   Atmosphere: Open landscapes, free exploration, long-journey romance
+   
+   Product Presentation:
+   - Helmet: ADV rider stopped at summit overlooking scenery, helmet on tank
+   - Suit: Rider in full gear standing by motorcycle looking at map
+   - Waterproof bags: Strapped to motorcycle with magnificent natural backdrop
+   - Products merged with exploration scenes conveying "reliable companion"
+   
+   Lighting: Outdoor natural light, golden hour warm tones preferred, wide sky diffusion
+   Atmosphere: Durable, exploration, freedom, reliable, adventure spirit
+
+6) 高端质感风 (Premium / Luxury):
+   Scene Integration: High-end showroom, private garage, minimalist studio
+   Color: Dark gradients (#1a1a1a→#2d2d2d) or deep blue-black, generous negative space
+   Texture Elements: Velvet-feel backgrounds, leather hints, marble surfaces
+   Environment: Premium motorcycle parts visible (Ducati, BMW) as status hints
+   
+   Product Presentation:
+   - Helmet: Art-piece presentation on elegant display, leather liner visible
+   - Leather gloves: Showcasing exquisite stitching and leather grain close-up
+   - Limited editions: Emphasizing unique design elements and brand identity
+   - Large product proportion, rich details, luxury advertisement quality
+   
+   Lighting: Precisely controlled soft light on key areas only, fine surface highlights
+   Atmosphere: Premium, minimalist, material quality, brand confidence, luxury
+
+7) 安全守护风 (Safety Focused):
+   Scene Integration: Safety testing scenes, family farewell riding, professional training
+   Color: Stable blue (#1e3a5f) or composed gray (#4a4a4a), trust-conveying
+   Elements: Safety certification marks, protection test visuals, family care scenes
+   Atmosphere: Trustworthy, reassuring, protective
+   
+   Product Presentation:
+   - Helmet: Cross-section showing EPS buffer layer protection details
+   - Gear: Rider in full protective gear with family waving goodbye warmly
+   - Reflective gear: Night riding scene with reflective materials glowing in headlights
+   - Visual guides highlighting key protection zones and safety features
+   
+   Lighting: Even soft key light without harsh contrast, ample fill showing full product
+   Atmosphere: Safe, protective, reliable, trustworthy, guardian
+
+8) 改装美学风 (Custom / Style):
+   Scene Integration: Custom workshop, street style shoot, motorcycle meetup, neon nightscape
+   Color: Bold contrasts, neon pink/blue/green, or vintage warm tones
+   Elements: Graffiti walls, neon signs, custom tools, trendy stickers
+   Atmosphere: Street culture, personal expression, subculture belonging
+   
+   Product Presentation:
+   - Helmet: Rider holding custom-painted helmet in trendy street portrait
+   - Custom parts: Installed on custom bike in cool showcase
+   - Personal gear: Styled with custom motorcycle aesthetic
+   - Emphasizing design details, unique finishes, personalization
+   
+   Lighting: Dramatic lighting creating visual focus, neon color sources, strong rim light
+   Atmosphere: Individual, bold, design-forward, attitude, visual expression
+
+9) 参数对比风 (Spec / Data):
+   Scene Integration: Clean comparison environment or lab testing scenes
+   Color: Neutral white (#ffffff) or light gray (#f5f5f5), clear information zones
+   Layout: Product body + parameter info area clearly separated
+   Elements: Large KPI numbers, graphical data, comparison charts, test data
+   
+   Product Presentation:
+   - Helmet: Weight comparison (vs competitors), wind tunnel test data display
+   - Gear: Protection level comparison, material strength data
+   - Accessories: Installation convenience, compatibility parameter matrix
+   - Product at 40-50% of frame, ample space for parameter display
+   
+   Lighting: Even bright product light without distracting shadows, high clarity info display
+   Atmosphere: Data, comparison, performance, rational, high conversion
+
+=============================================================
 [LAYOUT BLUEPRINT - ASPECT RATIO SPECIFIC]
+=============================================================

-★ 1:1 SQUARE FORMAT (800x800px) - 淘宝/天猫/京东主图 ★
+★ 1:1 SQUARE FORMAT (800x800px) ★
 ┌─────────────────────────────────────┐
-│ TOP ZONE (0-22%)                    │
-│ • Main Headline: 42pt equivalent    │
-│ • Bold, high contrast, max 12 chars │
-│ • Position: center or left-aligned  │
+│ TOP ZONE (0-20%)                    │
+│ • Main Headline: 52-60pt ultra bold │
+│ • High contrast, max 12 chars       │
 ├─────────────────────────────────────┤
 │ UPPER-MID ZONE (18-38%)             │
 │ • Sub Headline / USP tags           │
 │ • 3-5 feature icons with text       │
-│ • Smaller promo text if needed      │
 ├─────────────────────────────────────┤
 │ CENTER ZONE (25-75%)                │
-│ • PRODUCT HERO: 45-60% of frame     │
-│ • 45° angle preferred (depth)       │
-│ • Floating effect + soft shadow     │
-│ • Subtle rim light / glow accent    │
+│ • PRODUCT IN SCENE: 45-60% of frame │
+│ • Scene integration with atmosphere │
+│ • Floating effect + environmental   │
 ├─────────────────────────────────────┤
 │ LOWER-MID ZONE (65-82%)             │
 │ • Spec highlights / certifications  │
-│ • Trust badges (SGS/ACS/Hi-Res)     │
 ├─────────────────────────────────────┤
 │ BOTTOM ZONE (78-100%)               │
-│ • Price: Large red/orange digits    │
-│ • Promo tags: 限时折扣/买赠         │
-│ • Trust bar: 顺丰/质保/免息         │
+│ • Price + Promo tags + Trust bar    │
 └─────────────────────────────────────┘

-★ 3:4 VERTICAL FORMAT (600x800px) - 移动端/详情页首图 ★
+★ 3:4 VERTICAL FORMAT (600x800px) ★
 ┌─────────────────────────────────────┐
 │ TOP ZONE (0-18%)                    │
-│ • Main Headline: 42pt equivalent    │
-│ • Centered or left, bold impact     │
+│ • Main Headline: 52-60pt ultra bold │
 ├─────────────────────────────────────┤
 │ UPPER ZONE (15-32%)                 │
 │ • Sub headline + mini promo         │
-│ • Feature pills / USP badges        │
 ├─────────────────────────────────────┤
 │ CENTER ZONE (28-72%)                │
-│ • PRODUCT HERO: 50-65% of frame     │
-│ • More vertical space for product   │
-│ • Premium floating presentation     │
+│ • PRODUCT IN SCENE: 50-65% of frame │
+│ • More vertical space for scene     │
 ├─────────────────────────────────────┤
 │ LOWER ZONE (68-85%)                 │
 │ • Detailed specs / parameters       │
-│ • Certification badges              │
 ├─────────────────────────────────────┤
 │ BOTTOM ZONE (82-100%)               │
-│ • Price block + CTA                 │
-│ • Trust elements horizontal         │
+│ • Price block + CTA + Trust bar     │
 └─────────────────────────────────────┘

-[MAIN HEADLINE DESIGN - 42PT STANDARD]
-Typography Rules:
- Font Size: 42pt equivalent (largest element, impossible to miss)
- Weight: Extra Bold / Black (900 weight)
- Contrast: Must pass 7:1 contrast ratio against background
+[MAIN HEADLINE DESIGN - 52-60PT BOLD STANDARD]
+- Font Size: 52-60pt equivalent (LARGEST element, ABSOLUTE visual focus, must be seen FIRST)
+- Weight: Ultra Bold / Black (900-950), thick strokes, NEVER thin
+- Font Style: Bold sans-serif like Source Han Sans Heavy, Alibaba PuHuiTi Bold, or Impact
+- Contrast: 8:1 minimum against background, add shadow/stroke for enhanced readability
 - Length: 4-12 Chinese characters / 3-8 English words
- Style Options:
-  • Number Shock: "破纪录!减重50%" / "-55dB超宽频降噪"
-  • Benefit Direct: "舒适佩戴 不伤耳" / "好锅煮好饭"
-  • Pain Point Flip: "很贵!顶配'0'噪音"
-  • Result Promise: "24小时续航无忧"
- Position: Top 20% of frame, never below product center
- Effect: Subtle drop shadow for depth, NO outline stroke
+- Position: Top 15-20% of frame, centered or left-aligned, absolutely prominent
+- Visual Effect: Headline must "POP" from the image, impossible for viewer to ignore
+- Optional Enhancements: Subtle glow, soft drop shadow, contrasting stroke outline
+- Styles: Number shock / Benefit direct / Pain point flip / Result promise

-[SUB HEADLINE / USP ZONE]
- Font Size: 18-24pt equivalent
- Content: 3-5 key specs or benefits
- Format Options:
-  • Pipe separated: "54h续航 | 全景杜比音 | 360°不漏音"
-  • Icon + text pills: [🔋54h] [🎵Hi-Fi] [💧IPX7]
-  • Horizontal tag row with rounded backgrounds
- Position: Below main headline, above product
-
-[PRODUCT PRESENTATION - HERO TREATMENT]
-Mandatory Visual Effects:
-1) FLOATING EFFECT:
-   - Product appears to levitate 15-30px above surface
-   - Soft elliptical shadow beneath (gaussian blur 20-40px)
-   - Shadow opacity: 15-25%, slightly offset
+[PRODUCT PRESENTATION - SCENE INTEGRATED]
+1) SCENE INTEGRATION:
+   - Product must exist within a meaningful context
+   - Human/hand interaction when appropriate
+   - Environmental elements supporting the story
+   - Depth and atmosphere in composition

 2) LIGHTING SETUP:
-   - Key light: 45° from upper-left or upper-right
+   - Key light: 45° from upper corner, scene-appropriate
   - Fill light: Opposite side, 50% intensity
-   - Rim light: Subtle edge glow for separation
-   - Accent light: Color-matched to brand/style
+   - Rim light: Edge separation from background
+   - Environmental light: Matching scene atmosphere

-3) PRODUCT ANGLE:
-   - Primary: 45° side-front view (shows depth + face)
-   - Alternative: 30° three-quarter view
-   - Detail shots: Close-up on key feature
-
-4) MATERIAL RENDERING:
-   - Matte surfaces: Soft diffused highlights
-   - Glossy surfaces: Sharp specular highlights
-   - Metallic: Gradient reflections, subtle environment map
-   - Fabric/Soft: Visible texture, natural folds
+3) MATERIAL RENDERING:
+   - Matte: Soft diffused highlights
+   - Glossy: Sharp specular highlights
+   - Metallic: Gradient reflections
+   - Carbon fiber: Weave pattern visible

 [PRICE & PROMO ZONE DESIGN]
+Price Display: Big sale block / Corner promo / Dual price compare
+Promo Badges: TOP ranking / Limited discount / Bundle deals
+Trust Bar: Bottom 10%, horizontal layout, icon + text

-Price Display Styles:
-1) 大促价格块 (Big Sale Price Block):
-   ┌──────────────────┐
-   │ ¥299  ←大红/橙色42pt
-   │ ¥599  ←灰色划线原价18pt
-   │ 立省300元 ←绿色小字
-   └──────────────────┘
-
-2) 角标促销价 (Corner Promo):
-   - Position: Bottom-left corner
-   - Price in rounded rectangle background
-   - Discount percentage badge attached
-
-3) 双价对比 (Dual Price Compare):
-   - 国补价 ¥209 vs 原价 ¥299
-   - Side by side with visual hierarchy
-
-Promo Badge Types:
-1) TOP排名徽章:
-   - Style: Golden/red ribbon or shield
-   - Text: "TOP1" / "热销第1名" / "降噪效果:第1名"
-   - Position: Top-right or beside product
-   - Effect: Subtle metallic texture, drop shadow
-
-2) 限时折扣标签:
-   - Style: Diagonal ribbon or badge
-   - Text: "限时87折" / "立减60" / "限时特惠"
-   - Color: Red/orange gradient, white text
-   - Position: Right side vertical or top-right corner
-
-3) 买赠活动框:
-   - Style: Rounded rectangle or banner
-   - Text: "买2赠1" / "买3增2" / "加购送耳机包"
-   - Color: Promotional red or brand color
-
-4) 信任横条 (Trust Bar):
-   - Position: Bottom 10% of image
-   - Layout: Horizontal, evenly spaced
-   - Items: "顺丰包邮 | 三年质保 | 送运费险 | 6期免息"
-   - Style: Small icons + text, subtle background
-
-[BACKGROUND STYLE SPECIFICATIONS]
-
-1) 科技深邃 (Tech Deep):
-   - Gradient: #1a1a2e → #16213e → #0f3460
-   - Accent: Blue/purple glow spots, geometric lines
-   - Particle: Subtle tech grid or circuit pattern
-   - Light leak: Cyan/magenta edge glow
-   - Best for: Electronics, gadgets, audio equipment
-
-2) 温润米白 (Warm Cream):
-   - Gradient: #fefefe → #f5f0e8 → #e8e0d5
-   - Accent: Warm sunlight from upper corner
-   - Texture: Subtle linen or paper grain
-   - Shadow: Warm-tinted soft shadows
-   - Best for: Home appliances, lifestyle products
-
-3) 高级金棕 (Premium Bronze):
-   - Gradient: #d4a574 → #c9a067 → #8b6914 (subtle)
-   - Accent: Golden rim highlights
-   - Texture: Brushed metal or marble hints
-   - Light: Warm studio lighting, luxury feel
-   - Best for: Premium/luxury positioning
-
-4) 清新天蓝 (Fresh Sky):
-   - Gradient: #e8f4f8 → #d4ecf7 → #b8dced
-   - Accent: Soft cloud wisps, clean air feel
-   - Light: Bright, even, refreshing
-   - Best for: Summer products, cooling devices
-
-5) 氛围场景 (Lifestyle Scene):
-   - Setting: Coffee cup/desk/bathroom/bedroom context
-   - Product: Naturally placed but clearly hero
-   - Depth: Shallow DOF, product sharp, bg soft
-   - Light: Natural window light or warm interior
-   - Best for: Lifestyle brands, emotional connection
-
-6) 硬核深棕 (Hardcore Brown):
-   - Gradient: #3d2b1f → #5c4033 → #2c1810
-   - Texture: Leather, earth, rugged surface
-   - Accent: Orange/amber spot lights
-   - Best for: Sports, outdoor, masculine products
-
-7) 极简纯净 (Minimal Pure):
-   - Background: Pure white #ffffff or light gray #f8f8f8
-   - Shadow: Crisp product shadow only
-   - No gradients, no effects, Amazon-style clean
-   - Best for: Marketplace compliance, spec-focused
-
-8) 未来科幻 (Sci-Fi Future):
-   - Gradient: Deep space colors, nebula hints
-   - Texture: Planetary surface, alien terrain
-   - Accent: Neon glows, holographic hints
-   - Light: Dramatic rim lighting
-   - Best for: Cutting-edge tech, gaming
-
-[VISUAL IMPACT ENHANCERS - MANDATORY]
-
-1) Depth & Dimension:
-   - Layered composition (foreground elements if appropriate)
-   - Atmospheric perspective for backgrounds
-   - Product should "pop" from background
-
-2) Color Harmony:
-   - Product color + complementary accent
-   - Background should enhance, not compete
-   - Price/promo colors: Red family for urgency
-
-3) Professional Polish:
-   - No pixelation, no artifacts
-   - Clean edges, proper anti-aliasing
-   - Consistent lighting direction
-   - Realistic reflections and refractions
-
-4) Text Readability:
-   - All text must be crystal clear
-   - No text over busy areas
-   - Proper text hierarchy (size/weight/color)
-   - Chinese text: Clean sans-serif (思源黑体 style)
-
-[VARIANT PLANNING BY COUNT]
-
-prompt_count = 1:
-→ HERO PRIME: Best angle + strongest headline + optimal promo layout
-
-prompt_count = 2:
-→ V1: Hero Prime (45° angle, main value prop)
-→ V2: Feature Spotlight (close-up on key differentiator)
-
-prompt_count = 3:
-→ V1: Hero Prime
-→ V2: Feature Spotlight
-→ V3: Trust Builder (TOP badge + certifications prominent)
-
-prompt_count = 4-6:
-→ V1: Hero Prime
-→ V2: Feature Spotlight
-→ V3: Trust Builder
-→ V4: Price Focus (big sale emphasis)
-→ V5: Lifestyle Context (usage scene)
-→ V6: Alternate Angle (back/side view)
+[VARIANT PLANNING]
+prompt_count = 1: Hero Prime - best scene + strongest headline
+prompt_count = 2: + Feature Spotlight (detail close-up)
+prompt_count = 3: + Trust Builder (certifications prominent)
+prompt_count = 4-6: + Price Focus, Lifestyle Context, Angle Variation

 [EACH PROMPT MUST CONTAIN - 6 MODULES]
+1) 主标题/Main Headline: "..." (52-60pt ultra bold, absolute visual focus, must be seen FIRST)
+2) 副标题/Sub Headline: "..." (18-24pt, 3-5 specs/benefits)
+3) 价格促销区/Price & Promo Zone: Price style, badge types & positions, trust bar
+4) 排版蓝图/Layout Blueprint: Product position & scale in scene, text zones, badge placement
+5) 视觉与光影/Visual & Lighting: [CRITICAL] Scene type, human/hand interaction, environment details, lighting setup, atmosphere, material rendering
+6) 渲染品质/Render Quality: 8K commercial grade, material details crisp, text razor-sharp, professional color grading, visual impact 9/10

-Chinese Template:
-1) 主标题: "..." (42pt等效大字，4-12字，视觉冲击)
-2) 副标题: "..." (18-24pt，3-5个卖点/规格)
-3) 价格促销区: 价格展示方式、促销标签类型与位置、信任横条内容
-4) 排版蓝图: 产品位置与占比、文字区域划分、标签位置、画幅比例(1:1或3:4)适配说明
-5) 视觉与光影: 背景风格详述、光源设置(主光/补光/轮廓光)、产品悬浮效果、材质渲染、色彩搭配
-6) 渲染品质: 8K超清商业摄影级、材质细节清晰可见、文字锐利无锯齿、专业后期调色、视觉冲击力评分目标9/10
-
-English Template:
-1) Main Headline: "..." (42pt equivalent, 3-8 words, visual impact)
-2) Sub Headline: "..." (18-24pt, 3-5 specs/benefits)
-3) Price & Promo Zone: Price display style, badge types & positions, trust bar content
-4) Layout Blueprint: Product position & scale, text zone allocation, badge placement, aspect ratio (1:1 or 3:4) adaptation
-5) Visual & Lighting: Background style detail, lighting setup (key/fill/rim), floating effect, material rendering, color harmony
-6) Render Quality: 8K commercial photography grade, material details crisp, text razor-sharp, professional color grading, visual impact score target 9/10
-
-[VARIANT ROLE LABELS - REQUIRED]
-Each variant must start with role identifier:
- Chinese: "变体定位：主图首选" / "变体定位：卖点特写" / "变体定位：信任背书" / "变体定位：促销主打" / "变体定位：场景氛围"
- English: "Variant Role: Hero Prime" / "Variant Role: Feature Spotlight" / "Variant Role: Trust Builder" / "Variant Role: Price Focus" / "Variant Role: Lifestyle Context"
+[VARIANT ROLE LABELS]
+Each variant starts with role identifier:
+- Chinese: "变体定位：主图首选/卖点特写/信任背书/促销主打/场景氛围"
+- English: "Variant Role: Hero Prime/Feature Spotlight/Trust Builder/Price Focus/Lifestyle Context"

 [ANTI-TRANSLATION CLAUSE]
 Append at END of each prompt:
 - Chinese: 画面所有文字必须为中文，禁止出现任何英文字母或乱码，文字渲染清晰锐利。
 - English: All text must be in English only. No Chinese characters. Text rendering must be crystal clear.

-[FINAL QUALITY CHECKLIST - EVERY PROMPT]
-□ Main headline is 42pt equivalent, bold, impossible to miss
-□ Product occupies 45-60% of frame with premium floating effect
-□ Lighting setup creates depth and dimension
-□ Background style matches product category and brand feel
-□ Price/promo elements are clear but don't overpower product
-□ Layout adapts correctly to specified aspect ratio (1:1 or 3:4)
-□ Visual impact score potential: 9/10 or higher
-□ Professional commercial photography aesthetic achieved
+[FINAL QUALITY CHECKLIST]
+□ Main headline 52-60pt ultra bold, absolute first-eye focus, impossible to miss
+□ Product INTEGRATED into scene with story/atmosphere (NOT plain background)
+□ ⚠️ Human anatomy integrity: 5 fingers complete, natural joints, no deformations
+□ ⚠️ Posture logic: Physics-compliant movements, correct wearing/positioning
+□ Lighting creates depth, dimension, and scene atmosphere
+□ Background style matches selected style specification
+□ Price/promo clear but doesn't overpower product
+□ Layout adapts to specified aspect ratio
+□ Visual impact score: 9/10 or higher
+□ Apple/Dyson commercial photography quality achieved