ワンクリックで
web-image-generator
为网站生成匹配的 AI 图片资源,支持 hero images、section backgrounds、illustrations 等多种类型
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
为网站生成匹配的 AI 图片资源,支持 hero images、section backgrounds、illustrations 等多种类型
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Design-system enforcement contract for generated website UI. Use when creating or refining sections, pages, shells, forms, cards, tables, and navigation.
Defines the end-to-end website generation workflow. Invoke when generating multi-section pages or full websites.
Website-only import adapted from HTML Anything template discipline for cinematic AI/product launch pages with staged demo proof and sharp CTA focus.
Website-only import adapted from HTML Anything template discipline for resource hubs and information-platform routes.
Website-only import adapted from HTML Anything template discipline for documentation/reference pages. Default primary template discipline for docs and reference routes.
Website-only import adapted from HTML Anything template discipline for industrial product catalogs, spec comparison, and buyer inquiry routes.
| name | web-image-generator |
| description | 为网站生成匹配的 AI 图片资源,支持 hero images、section backgrounds、illustrations 等多种类型 |
为网站生成项目创建优化的 AI 图片资源,包括 hero images、背景图、插图、图标等,确保视觉风格与设计系统一致。
特征:网站首屏的主视觉,需要强烈的视觉冲击力
| 关键点 | 说明 |
|---|---|
| 高分辨率 | 至少 1920x1080,推荐 2560x1440 |
| 视觉焦点 | 明确的主体,支持文字叠加 |
| 留白区域 | 为标题和 CTA 预留空间 |
| 色彩协调 | 与品牌色系匹配 |
Prompt 模板:
{subject description}, hero image style, {composition} composition with negative space for text overlay, {color palette}, cinematic lighting, high quality, 8K resolution, professional web design
Negative Prompt:
text, watermark, busy details, cluttered, low quality, blurry
特征:用于不同 section 的背景,需要低调支撑内容
| 关键点 | 说明 |
|---|---|
| 低对比度 | 不抢内容风头 |
| 可重复性 | 支持响应式布局 |
| 渐变友好 | 易于与纯色过渡 |
| 轻量化 | 文件大小控制 |
Prompt 模板:
Subtle {theme} background pattern, {style} style, {color scheme}, soft and understated, suitable for web section background, seamless, minimal details, professional
Negative Prompt:
high contrast, busy, distracting, text, watermark, photorealistic
特征:展示产品功能或概念的插图
| 关键点 | 说明 |
|---|---|
| 扁平化 | 现代 Web 风格 |
| 矢量感 | 清晰的线条和形状 |
| 统一风格 | 整站插图保持一致 |
| 透明背景 | 便于叠加 |
Prompt 模板:
{concept description}, flat design illustration, modern web style, {color palette}, clean lines, minimal details, transparent background, vector style, professional UI illustration
Negative Prompt:
realistic, photography, 3D render, complex textures, shadows, watermark
特征:展示产品界面或功能的真实截图风格
| 关键点 | 说明 |
|---|---|
| 真实感 | 模拟真实产品界面 |
| 设备框架 | 可选择添加设备外框 |
| 清晰度 | 高清晰度展示细节 |
| 阴影效果 | 增加立体感 |
Prompt 模板:
{product interface description}, modern UI design, clean interface, {color scheme}, professional software screenshot style, high resolution, sharp details, realistic mockup
Negative Prompt:
blurry, low quality, distorted, watermark, text overlay
特征:品牌 logo 或装饰性图标
| 关键点 | 说明 |
|---|---|
| 简洁性 | 高度简化的形状 |
| 可缩放 | 支持多种尺寸 |
| 单色/双色 | 易于主题切换 |
| 透明背景 | 必须 |
Prompt 模板:
{icon concept} icon, minimalist design, {color}, simple geometric shapes, flat style, transparent background, suitable for web logo, professional branding
Negative Prompt:
complex, detailed, realistic, 3D, shadows, gradients, watermark
为每张图片创建优化的 prompt:
## Image Resource List
| # | Filename | Type | Dimensions | Purpose | Status |
|---|----------|------|-----------|---------|--------|
| 1 | hero-main.png | Hero Image | 2560x1440 | Landing page hero | Pending |
| 2 | feature-1.png | Illustration | 800x600 | Feature section | Pending |
| 3 | bg-pattern.png | Background | 1920x1080 | Section background | Pending |
使用统一的图片生成工具:
# 使用 image_gen.py(如果可用)
python3 scripts/image_gen.py "your prompt" \
--aspect_ratio 16:9 \
--image_size 2K \
--output project/images \
--filename hero-main
# 或使用其他 AI 图片生成服务
# - Midjourney
# - DALL-E 3
# - Stable Diffusion
# - Gemini Imagen
配置要求:
IMAGE_BACKEND=gemini # 或 openai, qwen, zhipu, volcengine
GEMINI_API_KEY=your-api-key
GEMINI_MODEL=gemini-3.1-flash-image-preview
生成后的处理:
// React 示例
<img
src="/images/hero-main.webp"
srcSet="/images/hero-main@2x.webp 2x, /images/hero-main@3x.webp 3x"
alt="Modern workspace with collaborative team"
loading="lazy"
/>
// HTML 示例
<picture>
<source srcset="/images/hero-main.webp" type="image/webp">
<img src="/images/hero-main.png" alt="Modern workspace">
</picture>
| 用途 | 推荐尺寸 | 宽高比 |
|---|---|---|
| Hero Image (Desktop) | 2560x1440 | 16:9 |
| Hero Image (Mobile) | 1080x1920 | 9:16 |
| Feature Illustration | 800x600 | 4:3 |
| Section Background | 1920x1080 | 16:9 |
| Product Screenshot | 1440x900 | 16:10 |
| Icon/Logo | 512x512 | 1:1 |
| Social Share Image | 1200x630 | 1.91:1 |
/* Mobile First */
@media (min-width: 640px) { /* sm */ }
@media (min-width: 768px) { /* md */ }
@media (min-width: 1024px) { /* lg */ }
@media (min-width: 1280px) { /* xl */ }
@media (min-width: 1536px) { /* 2xl */ }
从设计系统提取颜色并转换为 prompt:
// 设计系统
const colors = {
primary: '#3B82F6', // Blue
secondary: '#8B5CF6', // Purple
accent: '#F59E0B', // Amber
neutral: '#6B7280' // Gray
}
// 转换为 prompt
"color palette: vibrant blue (#3B82F6), purple accent (#8B5CF6), warm amber highlights (#F59E0B)"
高质量:high quality, 8K resolution, sharp details, professional
构图:centered composition, rule of thirds, balanced layout
光效:soft lighting, natural light, cinematic lighting
风格:modern, clean, minimalist, professional
❌ realistic (除非需要照片风格)
❌ 3D render (除非特定需求)
❌ complex details (Web 图片需要简洁)
❌ dark, gloomy (除非品牌调性)
# 生成 hero image
python3 scripts/image_gen.py \
"Modern tech workspace with collaborative team, hero image style, centered composition with negative space for text overlay, color palette: blue (#3B82F6) and purple (#8B5CF6), cinematic lighting, high quality, 8K resolution" \
--aspect_ratio 16:9 \
--image_size 2K \
--output website-project/images \
--filename hero-main \
--negative_prompt "text, watermark, busy details, cluttered, low quality"
// 示例:调用 OpenAI DALL-E
const response = await openai.images.generate({
model: "dall-e-3",
prompt: "your optimized prompt",
size: "1792x1024",
quality: "hd",
n: 1,
});
images/image_prompts.mdimages/ 目录生成完成后检查:
images/ 目录在 website-generation-workflow 的 Phase 1.5 插入:
Phase 1: 规划与设计系统
↓
Phase 1.5: 图片资源生成 ← 使用 web-image-generator
↓
Phase 2: 分批生成 Section
创建 project/images/image_prompts.md:
# Website Image Generation Prompts
> Project: {project_name}
> Generated: {date}
> Design System: Primary {#HEX} | Secondary {#HEX} | Accent {#HEX}
---
## Image Resource List
| # | Filename | Type | Dimensions | Purpose | Status |
|---|----------|------|-----------|---------|--------|
| 1 | hero-main.png | Hero Image | 2560x1440 | Landing hero | Pending |
---
## Detailed Prompts
### Image 1: hero-main.png
**Type**: Hero Image
**Dimensions**: 2560x1440 (16:9)
**Purpose**: Landing page hero section
**Prompt**:
Modern tech workspace with collaborative team, hero image style, centered composition with negative space for text overlay, color palette: vibrant blue (#3B82F6) and purple accent (#8B5CF6), cinematic lighting, high quality, 8K resolution, professional web design
**Negative Prompt**:
text, watermark, busy details, cluttered, low quality, blurry
**Alt Text**:
> Modern collaborative workspace with team members working together on technology projects
**Generation Command**:
```bash
python3 scripts/image_gen.py "Modern tech workspace..." --aspect_ratio 16:9 --image_size 2K -o images -f hero-main
images/ 目录
## 常见问题
### Q: 图片风格不统一怎么办?
A: 在所有 prompts 中使用统一的风格关键词,例如:
- 统一添加 "flat design, modern web style"
- 统一色彩描述
- 统一质量关键词
### Q: 生成的图片有水印?
A:
- 使用 `--negative_prompt "watermark, signature, text"`
- 如果是 Gemini,使用 `gemini_watermark_remover.py` 移除
### Q: 文件太大影响加载速度?
A:
- 转换为 WebP 格式
- 使用图片压缩工具
- 实现懒加载(lazy loading)
- 使用 CDN
## 下一步
图片生成完成后:
1. 更新图片资源列表状态为 "Generated"
2. 通知用户图片已就绪
3. 继续 website-generation-workflow 的下一阶段