brand-consistent-visuals

品牌一致性视觉内容

生成专业的品牌视觉内容，确保所有创意资产都保持一致的品牌形象。这项技能通过多步骤 AI 分析来理解品牌定位、制定战略性创意简报，并生成真实体现品牌形象、价值观和视觉语言的视觉内容。

概述

这项技能使 AI Agent 能够创建与品牌一致的视觉内容，具体方式包括：

• 分析全面的品牌输入（网站、徽标、颜色、字体、指南）
• 了解品牌定位、个性和目标受众
• 创建或完善创意简报，提供专业的创意指导
• 生成多样、高品质且保持视觉一致性的品牌视觉内容
• 确保所有输出符合品牌形象和战略定位

何时使用这项技能

在以下情况下使用这项技能：

• 生成保持品牌形象一致的品牌视觉内容
• 创建符合品牌指南的营销材料
• 制定体现品牌定位的创意概念
• 制作多样的品牌资产（社交媒体、广告、模型图、编辑设计）
• 确保多个创意输出之间的视觉一致性
• 将品牌战略转化为视觉呈现
• 创建专业的品牌演示文稿和情绪板

先决条件

必需输入：

• website_url（字符串，必需）：品牌网站 URL，是品牌分析的主要来源
• 图像生成能力（Gemini 3 Pro 或同等能力）

可选输入：

• custom_description（字符串）：描述创作需求的创意简报
• brand_logo（字符串）：单个品牌徽标 URL 或 base64 数据 URI
• brand_logos（数组）：多个品牌徽标 URL 或数据 URI
• brand_info（字符串）：品牌信息（颜色、字体、指南）
• brand_memory（对象）：包含名称、概述、调色板和字体的结构化品牌数据
• num_variations（整数）：要生成的变体数量（1-12，默认值：3）
• aspect_ratio（字符串）：输出内容的宽高比（默认值："1:1"）
• output_format（字符串）：输出格式——png、jpg、webp（默认值："png"）

分步说明

第 1 步：全面分析品牌

分析所有品牌输入，了解品牌定位并制定战略方向：

要分析的输入：

• 网站 URL（主要事实来源）
• 品牌徽标和视觉资产
• 品牌信息（颜色、字体、指南）
• 品牌记忆数据（如有）
• 自定义创意描述（如有）

分析任务：

1. 确定品牌所属行业：根据品牌定位确定其主要行业类别，而非表面上的产品类别
2. 提取品牌洞察：
   • 品牌成熟度（初创、成长阶段、成熟、高端）
   • 目标受众的成熟度
   • 品牌个性（大胆、极简、活泼、严肃、奢华、实用）
   • 视觉风格（现代、永恒、实验性、保守）
3. 创建／完善创意简报：
   • 如果没有自定义描述：从头创建全面的创意简报
   • 如果提供了自定义描述：通过更明确的创意指导加以完善
   • 定义设计内容（品牌套件、广告、社交媒体、包装）
   • 说明设计目的（品牌目标、定位、受众）
   • 定义设计的外观和感觉（风格、调性、视觉语言）
   • 参考品牌颜色、字体和设计原则

输出：

{
  brand_industry: string;           // e.g., "Fashion & Apparel"
  enhanced_creative_brief: string;  // Detailed professional creative brief
  brand_insights: string;           // 2-3 sharp sentences on positioning
}

第 2 步：对创意需求分类

识别所需创意资产的完整范围，避免将其局限于单一格式：

分类规则：

• 确定主要资产（主要交付内容）
• 确定次要资产（辅助格式）
• 定义使用场景（使用位置和方式）
• 考虑完整的品牌形象体系

输出：

{
  primary_assets: string[];    // e.g., ["brand mockups", "logo applications"]
  secondary_assets: string[];  // e.g., ["social", "web", "editorial"]
  usage_context: string;       // e.g., "brand identity system"
}

第 3 步：生成专业图像提示词

为每种品牌视觉效果创建详细且独特的提示词：

提示词要求：

1. 详细描述（每条 150-300 个英文单词）：

   • 具体的视觉细节：纹理、材料、表面处理、表面
   • 光线细节：方向、质感、色温、阴影
   • 相机细节：角度、距离、焦距、景深
   • 构图：取景、留白、视觉层级
   • 色彩：具体色调、饱和度、对比关系
   • 情绪与氛围：能量、情感、品牌个性

2. 独特性要求（至关重要）：

   • 每条提示词都必须完全独特
   • 不得在不同提示词中重复背景
   • 不得在不同提示词中重复文案或描述
   • 每种背景都必须各不相同（不能使用混凝土墙面或地面）
   • 每条提示词都要使用不同的描述语言
   • 提示词之间不得存在顺序或层级关系

3. 专业标准：

   • 采用符合真实物理规律的真实摄影效果
   • 模特的物理表现（如包含模特）：姿势真实、站立稳固
   • 文本/设计的布局逻辑：对齐、间距和融合方式得当
   • 不得出现幻觉：不得有悬浮物体或不可能的物理效果
   • 材质真实可信：阴影、深度、比例和透视准确
   • 设计克制：构图简洁，空间运用有意为之

4. 宽高比规范：

   • 每条提示词都必须以 --ar X:X 格式结尾
   • 根据创意类型选择合适的宽高比：
     • 产品特写、方形社交媒体素材：--ar 1:1
     • 肖像/竖版：--ar 9:16 或 --ar 4:5
     • 横版/宽幅：--ar 16:9 或 --ar 21:9
     • 标准印刷品：--ar 4:3 或 --ar 3:4
     • 海报：--ar 3:2 或 --ar 2:3

输出：

{
  prompts: string[];  // Array of detailed prompts, each ending with --ar X:X
}

第 4 步：生成品牌视觉效果

使用详细提示词和品牌素材生成图像：

生成参数：

• 使用 Gemini 3 Pro Image Preview 或同等工具
• 将品牌 Logo 作为参考图像（转换为 S3 URL）
• 提取每条提示词中的宽高比（如果使用“default”设置）
• 应用经过验证的输出格式
• 并行生成所有变体以提高效率

质量检查：

• 确认每张生成的图像都有有效 URL
• 检查元数据中是否有生成错误
• 筛除生成失败的结果
• 确保所有输出的品牌一致性

输出：

{
  success: boolean;
  outputAssets: Array<{
    type: "image";
    url: string;
    metadata: {
      prompt: string;
      aspect_ratio: string;
      model_used: string;
      generation_time: number;
    };
  }>;
}

示例

示例 1：时尚品牌社交媒体素材

输入：

{
  website_url: "https://urbanvault.fashion",
  custom_description: "Create social media content showcasing our minimalist aesthetic",
  brand_logos: ["https://example.com/logo.png"],
  brand_info: "name: Urban Vault, palette: { primary: [{ hex: #000000, label: black }], secondary: [{ hex: #FFFFFF, label: white }] }",
  num_variations: 4,
  aspect_ratio: "1:1",
  output_format: "png"
}

第 1 步输出（品牌分析）：

{
  brand_industry: "Fashion & Apparel - Contemporary Minimalist",
  enhanced_creative_brief: "Create a series of social media visuals for Urban Vault, a contemporary minimalist fashion brand targeting sophisticated urban professionals aged 25-40. The brand embodies clean lines, monochromatic palettes, and architectural silhouettes. Visual direction: High-contrast black and white photography with strong geometric compositions. Each image should feature the brand's signature minimalist aesthetic through carefully curated lifestyle moments - morning coffee rituals, urban commutes, workspace details. Emphasize negative space, clean typography, and subtle brand integration. The mood is calm, intentional, and refined. Avoid busy backgrounds or excessive styling. Each visual should feel like a moment of clarity in urban chaos.",
  brand_insights: "Urban Vault positions itself as the antidote to fast fashion chaos, offering timeless minimalist pieces for the intentional consumer. Visual identity emphasizes restraint, quality, and architectural precision."
}

第 2 步输出（素材分类）：

{
  primary_assets: ["social media posts", "Instagram grid content"],
  secondary_assets: ["stories", "carousel posts", "brand mood boards"],
  usage_context: "social media brand presence and community engagement"
}

第 3 步输出（图像提示词——4 条中的第 1 条示例）：

Professional lifestyle photography of a minimalist morning scene. A pristine white marble countertop occupies the lower third of the frame, featuring a single black ceramic coffee cup positioned using the rule of thirds. Steam rises delicately from the cup, backlit by soft morning light streaming through a nearby window. The Urban Vault logo appears subtly embossed on a small black business card leaning against the cup. Background: floor-to-ceiling windows revealing a blurred cityscape, creating depth through atmospheric perspective. The composition uses extreme negative space - 70% of the frame is clean white and soft gray tones. Lighting: natural diffused morning light from camera left, creating soft shadows that define the cup's form. Camera: 50mm lens at f/2.8, shallow depth of field keeping the cup sharp while the background gently blurs. Color palette: monochromatic with subtle warm undertones in the marble. Mood: serene, intentional, refined. The scene feels like a quiet moment of clarity before the day begins. Typography: "Start Intentionally" in thin sans-serif font, positioned in the upper right negative space, size 24pt, perfectly aligned with the grid. Material details: the marble shows subtle natural veining, the ceramic has a matte finish with soft light reflections, the business card has a slight texture. Spatial relationships: the cup sits 3 inches from the card, both properly grounded on the surface with realistic contact shadows. No floating elements, no impossible physics. --ar 1:1

第 4 步输出（生成的图像）：

{
  success: true,
  outputAssets: [
    {
      type: "image",
      url: "https://s3.amazonaws.com/shopos-assets/urban-vault-social-1.png",
      metadata: {
        prompt: "[Full prompt from Step 3]",
        aspect_ratio: "1:1",
        model_used: "gemini-3.0-pro-image-preview",
        generation_time: 12.4
      }
    },
    // ... 3 more variations with unique backgrounds and compositions
  ]
}

示例 2：科技初创公司品牌套件

输入：

{
  website_url: "https://nexusai.tech",
  brand_memory: {
    name: "Nexus AI",
    overview: "AI-powered workflow automation for modern teams",
    tagline: { text: "Work smarter, not harder", tones: ["innovative", "approachable"] },
    palette: {
      primary: [{ hex: "#6366F1", label: "indigo" }],
      secondary: [{ hex: "#8B5CF6", label: "purple" }, { hex: "#EC4899", label: "pink" }]
    }
  },
  num_variations: 3,
  aspect_ratio: "16:9",
  output_format: "png"
}

第 1 步输出（品牌分析）：

{
  brand_industry: "Technology - SaaS & AI",
  enhanced_creative_brief: "Generate professional brand visualizations for Nexus AI, an innovative AI-powered workflow automation platform targeting modern tech-forward teams. The brand balances cutting-edge technology with human approachability. Visual direction: Create dynamic, futuristic compositions that showcase the brand's AI capabilities while maintaining warmth and accessibility. Use the brand's vibrant gradient palette (indigo to purple to pink) as a signature element. Each visual should demonstrate the concept of 'intelligent automation' through abstract representations of data flow, neural networks, or seamless integrations. Incorporate the tagline 'Work smarter, not harder' in modern sans-serif typography. Mood: innovative yet approachable, powerful yet friendly, technical yet human. Avoid cold corporate aesthetics or overly complex technical imagery.",
  brand_insights: "Nexus AI differentiates through approachable innovation - making powerful AI accessible to everyday teams. Visual identity combines technical sophistication with human warmth through vibrant gradients and clean design."
}

第 2 步输出（素材分类）：

{
  primary_assets: ["brand visualizations", "product marketing assets"],
  secondary_assets: ["web banners", "presentation slides", "social media"],
  usage_context: "brand identity system and product marketing"
}

第 3 步输出（图像提示词——3 条中的第 1 条示例）：

Abstract 3D visualization representing AI-powered workflow automation. The composition features a flowing network of luminous nodes and connections forming an organic neural pathway through three-dimensional space. Primary color: vibrant indigo (#6366F1) nodes pulsing with soft inner glow, connected by gradient lines transitioning from purple (#8B5CF6) to pink (#EC4899). The network flows from bottom-left to top-right, creating dynamic diagonal movement. Background: deep navy gradient fading to black, with subtle particle effects suggesting data streams. Lighting: each node emits soft volumetric light, creating atmospheric depth and dimension. The largest central node features the Nexus AI logo subtly integrated as a holographic projection. Camera: wide-angle perspective at 35mm equivalent, positioned slightly below the network looking up, creating an impressive sense of scale. Depth of field: foreground nodes sharp, background nodes softly blurred with bokeh effects. Typography: "Work smarter, not harder" positioned in the lower third, using modern geometric sans-serif (Outfit or similar), 48pt, white with 80% opacity, perfectly aligned with the grid. The text has a subtle glow effect matching the indigo brand color. Spatial composition: nodes distributed using the golden ratio, creating natural visual flow. Material properties: nodes have glass-like translucency with internal light diffusion, connections have fiber-optic luminescence. Physics: all elements properly positioned in 3D space with accurate perspective, realistic light falloff, and proper depth relationships. No floating elements without context. The overall effect is sophisticated, futuristic, and inviting - technical excellence made approachable. --ar 16:9

第 4 步输出（生成的图像）：

{
  success: true,
  outputAssets: [
    {
      type: "image",
      url: "https://s3.amazonaws.com/shopos-assets/nexus-ai-brand-1.png",
      metadata: {
        prompt: "[Full prompt from Step 3]",
        aspect_ratio: "16:9",
        model_used: "gemini-3.0-pro-image-preview",
        generation_time: 15.2
      }
    },
    // ... 2 more variations with unique compositions
  ]
}

示例 3：奢华美妆品牌专题内容

输入：

{
  website_url: "https://lumiere-beauty.com",
  custom_description: "Create editorial-style imagery for our new skincare line launch",
  brand_logos: ["data:image/png;base64,iVBORw0KGgoAAAANS..."],
  brand_info: "name: Lumière Beauty, palette: { primary: [{ hex: #D4AF37, label: gold }], secondary: [{ hex: #FFFFFF, label: white }, { hex: #F5F5DC, label: cream }] }, fonts: [{ family: Playfair Display, weight: 400 }]",
  num_variations: 4,
  aspect_ratio: "2:3",
  output_format: "jpg"
}

第 1 步输出（品牌分析）：

{
  brand_industry: "Beauty & Cosmetics - Luxury Skincare",
  enhanced_creative_brief: "Create sophisticated editorial imagery for Lumière Beauty's new skincare line, targeting affluent consumers aged 30-55 who value luxury, efficacy, and timeless elegance. The brand embodies French-inspired sophistication with a focus on radiant, luminous skin. Visual direction: Soft, ethereal photography with emphasis on natural light, delicate textures, and golden hour warmth. Each image should evoke the feeling of 'inner radiance' through careful use of the brand's signature gold accent against cream and white backgrounds. Feature the skincare products in elevated, art-directed still life compositions that feel more like fine art than traditional product photography. Incorporate natural elements (silk, marble, fresh botanicals) to suggest luxury and purity. Typography should use Playfair Display for elegance. Mood: luminous, sophisticated, timeless, aspirational. Avoid clinical or overly retouched aesthetics - embrace natural beauty and authentic luxury.",
  brand_insights: "Lumière Beauty positions itself as the intersection of French luxury and modern skincare science, promising transformative radiance through premium formulations. Visual identity emphasizes ethereal beauty, golden light, and timeless sophistication."
}

输出将继续包含素材分类、提示词和生成的图像……

集成模式

TypeScript 后端集成

Claude 的工具定义

import Anthropic from "@anthropic-ai/sdk";

const brandConsistentVisualsTool: Anthropic.Tool = {
  name: "generate_brand_consistent_visuals",
  description: "Generate professional brand visualizations that maintain consistent brand identity across all creative assets using multi-step AI analysis and brand guidelines. Analyzes brand positioning, creates strategic creative briefs, and generates visuals aligned with brand identity.",
  input_schema: {
    type: "object",
    properties: {
      website_url: {
        type: "string",
        description: "Brand's website URL (required) - primary source for brand analysis"
      },
      custom_description: {
        type: "string",
        description: "Creative brief describing what to create (optional). If not provided, the system will analyze the brand and generate suitable creative content."
      },
      brand_logo: {
        type: "string",
        description: "Single brand logo URL or base64 data URI (optional)"
      },
      brand_logos: {
        type: "array",
        items: { type: "string" },
        description: "Multiple brand logo URLs or data URIs (optional)"
      },
      brand_info: {
        type: "string",
        description: "Brand information including colors, fonts, guidelines (optional). Format: 'name: Brand Name, palette: { primary: [{ hex: #000000, label: black }] }'"
      },
      brand_memory: {
        type: "object",
        description: "Structured brand data with name, overview, palette, fonts, tagline (optional)",
        properties: {
          name: { type: "string" },
          overview: { type: "string" },
          tagline: {
            type: "object",
            properties: {
              text: { type: "string" },
              tones: { type: "array", items: { type: "string" } }
            }
          },
          palette: {
            type: "object",
            properties: {
              primary: {
                type: "array",
                items: {
                  type: "object",
                  properties: {
                    hex: { type: "string" },
                    label: { type: "string" }
                  }
                }
              },
              secondary: {
                type: "array",
                items: {
                  type: "object",
                  properties: {
                    hex: { type: "string" },
                    label: { type: "string" }
                  }
                }
              }
            }
          },
          fonts: {
            type: "array",
            items: {
              type: "object",
              properties: {
                family: { type: "string" },
                weight: { type: "string" },
                size: { type: "string" },
                color: { type: "string" }
              }
            }
          }
        }
      },
      num_variations: {
        type: "integer",
        description: "Number of variations to generate (1-12, default: 3)",
        minimum: 1,
        maximum: 12
      },
      aspect_ratio: {
        type: "string",
        description: "Aspect ratio for outputs. Options: '1:1', '2:3', '3:2', '16:9', '9:16', '4:3', '3:4', '4:5', '5:4', '21:9', 'default' (extracts from prompts)",
        enum: ["1:1", "2:3", "3:2", "16:9", "9:16", "4:3", "3:4", "4:5", "5:4", "21:9", "default"]
      },
      output_format: {
        type: "string",
        description: "Output format: png, jpg, jpeg, webp (default: png)",
        enum: ["png", "jpg", "jpeg", "webp"]
      }
    },
    required: ["website_url"]
  }
};

工具实现

interface BrandConsistentVisualsInput {
  website_url: string;
  custom_description?: string;
  brand_logo?: string;
  brand_logos?: string[];
  brand_info?: string;
  brand_memory?: {
    name?: string;
    overview?: string;
    tagline?: {
      text?: string;
      tones?: string[];
    };
    palette?: {
      primary?: Array<{ hex: string; label: string }>;
      secondary?: Array<{ hex: string; label: string }>;
    };
    fonts?: Array<{
      family?: string;
      weight?: string;
      size?: string;
      color?: string;
    }>;
  };
  num_variations?: number;
  aspect_ratio?: string;
  output_format?: string;
}

interface BrandConsistentVisualsOutput {
  success: boolean;
  outputAssets: Array<{
    type: "image";
    url: string;
    metadata: {
      prompt: string;
      aspect_ratio: string;
      model_used: string;
      generation_time: number;
    };
  }>;
  error?: string;
}

async function generateBrandConsistentVisuals(
  input: BrandConsistentVisualsInput
): Promise<BrandConsistentVisualsOutput> {
  try {
    // Call your ShopOS workflow API endpoint
    const response = await fetch("https://api.shopos.com/workflow/brand-kit/execute", {
      method: "POST",
      headers: {
        "Content-Type": "application/json",
        "Authorization": `Bearer ${process.env.SHOPOS_API_KEY}`
      },
      body: JSON.stringify({
        website_url: input.website_url,
        custom_description: input.custom_description || "",
        brand_logo: input.brand_logo || "",
        brand_logos: input.brand_logos || [],
        brand_info: input.brand_info || "",
        brand_memory: input.brand_memory || null,
        num_variations: input.num_variations || 3,
        aspect_ratio: input.aspect_ratio || "1:1",
        output_format: input.output_format || "png"
      })
    });

    if (!response.ok) {
      throw new Error(`API request failed: ${response.statusText}`);
    }

    const result = await response.json();
    return result;
  } catch (error) {
    console.error("Error generating brand-consistent visuals:", error);
    return {
      success: false,
      outputAssets: [],
      error: error instanceof Error ? error.message : "Unknown error occurred"
    };
  }
}

Agent 用法示例

import Anthropic from "@anthropic-ai/sdk";

const anthropic = new Anthropic({
  apiKey: process.env.ANTHROPIC_API_KEY
});

async function runBrandVisualsAgent(userMessage: string) {
  const messages: Anthropic.MessageParam[] = [
    {
      role: "user",
      content: userMessage
    }
  ];

  let continueLoop = true;

  while (continueLoop) {
    const response = await anthropic.messages.create({
      model: "claude-3-5-sonnet-20241022",
      max_tokens: 4096,
      tools: [brandConsistentVisualsTool],
      messages
    });

    console.log("Assistant:", response.content);

    // Check if Claude wants to use the tool
    if (response.stop_reason === "tool_use") {
      const toolUse = response.content.find(
        (block): block is Anthropic.ToolUseBlock => block.type === "tool_use"
      );

      if (toolUse && toolUse.name === "generate_brand_consistent_visuals") {
        console.log("Generating brand-consistent visuals...");
        
        const result = await generateBrandConsistentVisuals(
          toolUse.input as BrandConsistentVisualsInput
        );

        // Add assistant's response and tool result to messages
        messages.push({
          role: "assistant",
          content: response.content
        });

        messages.push({
          role: "user",
          content: [
            {
              type: "tool_result",
              tool_use_id: toolUse.id,
              content: JSON.stringify(result)
            }
          ]
        });

        // Continue the loop to get Claude's final response
        continue;
      }
    }

    // If we get here, Claude has finished
    continueLoop = false;
  }
}

// Example usage
runBrandVisualsAgent(
  "Create social media content for Urban Vault fashion brand. Their website is https://urbanvault.fashion. Generate 4 variations in 1:1 format."
);

最佳实践

品牌分析

• 始终将 website_url 作为了解品牌的首要事实来源
• 分析品牌定位，而不只是表面上的产品
• 考虑品牌成熟度、目标受众和视觉文化
• 提取战略洞见，不要只总结输入内容

创意简报质量

• 编写符合代理机构水准的简报，清晰说明做什么、为什么以及如何做
• 提供具体的视觉方向和设计原则
• 参考品牌颜色、字体排印和规范
• 避免泛泛而谈或含糊其辞——内容要具体且可执行

提示词生成

• 为每条提示词撰写 150-300 字的详细视觉描述
• 确保内容完全独特——不重复背景或文案
• 使用专业的摄影和设计术语
• 指定符合现实的物理效果和恰当的空间关系
• 包含技术规格：灯光、相机、构图、色彩

视觉一致性

• 在所有变体中保持品牌个性
• 始终如一地使用品牌色彩、字体和视觉语言
• 确保每个输出都真实体现「品牌风格」
• 在保持一致性的同时兼顾创意多样性

质量标准

• 验证所有生成图片均具有有效 URL
• 检查元数据中是否存在生成错误
• 确保宽高比和输出格式正确
• 返回结果前滤除生成失败的图片

常见问题

1. 创意简报过于笼统：避免使用「现代」或「动感」等含糊词语；要明确具体的视觉方向
2. 背景重复：每条提示词都必须使用独特环境——不能使用混凝土墙面或地面
3. 文案重复：每条提示词都必须使用不同的描述语言和视觉叙事
4. 忽视品牌定位：不要只使用品牌色彩；要理解其战略定位
5. 幻觉：指定符合现实的物理效果和恰当的支撑关系，避免物体悬浮
6. 文字位置不佳：仔细设计文字的融合方式——确保对齐、间距和层级恰当
7. 品牌调性不一致：在所有变体中保持品牌个性
8. 缺少网站分析：务必彻底分析网站——它是主要信息来源

相关技能

• product-background-generation：用于产品专属的品牌视觉呈现
• fashion-model-photography：用于时尚品牌的编辑内容
• garment-lifestyle-photography：用于生活方式品牌摄影
• product-analysis-styling：用于以产品为核心的品牌素材

技术说明

• 使用 Gemini 3 Pro Image Preview 生成（或使用同等模型）
• 支持并发生成多个变体，提高效率
• 将品牌徽标转换为 S3 URL，确保图像生成可靠
• 使用「default」设置时，从提示词中提取宽高比
• 生成前验证所有输入（宽高比、输出格式）
• 实施全面的错误处理和重试逻辑
• 发布进度事件，以实时更新状态

版本历史

• 1.0.0（2024-01）：初始版本，包含多步骤品牌分析和专业提示词生成