# 主图生成记录 v1.0

日期：2026-10-14。用途：替换GitHub中英首页的顶部主图，课程与PDF案例内容不变。

## 实际来源与边界

使用OpenAI内置image_gen图像生成工具，未使用API/CLI付费生成路径。根据文字提示新生成原创概念视觉，没有上传儿童照片、书页或迪士尼图像作为参考。人物为虚构卡通形象，不代表创作者家庭或实际试读者。工具未返回准确模型版本，不编造型号。

画面：温暖的森林阅读空间、打开的故事书、原创AI小助手、青蓝色全息漫画页面与节点光线。是AI生成的概念主图，不是两本PDF的实际截图，也不证明学习效果或已经实现图中功能。

保存原始PNG：course/assets/课程封面原图v1.1.png。首页优化版：course/assets/课程封面v1.1.webp。优化只转换格式和压缩，未裁切、重画或修改人物。旧course/assets/课程封面v1.0.svg保留为历史版本。

主图属于新增AI图像，沿用项目对AI图像单列的许可边界，不自动套用课程文字与代码的CC BY-NC 4.0。不能将本次已知来源推广为两本历史PDF的逐图来源已经确认。

## 实际生成提示词

```text
Use case: illustration-story. Asset type: a wide GitHub repository cover banner, approximately 2.4:1 landscape, for a free educational comic-reading course. Create a beautiful, premium original illustration combining child-loved comics with approachable AI technology. Scene: two entirely fictional children joyfully discovering an open storybook beside a small friendly original AI helper robot. The open book is the clear focal point; illustrated comic panels and tiny woodland story scenes rise from the pages as elegant translucent cyan/teal holograms, linked by a few glowing neural-network nodes. Communicate a book becoming comics, reading and creative learning. Original characters with expressive faces, readable silhouettes, rounded shapes, delicate ink outlines and polished painterly shading; sophisticated storybook cover quality, not clip art or a technical diagram. Warm cream paper background, forest-green accents, honey-yellow book light, restrained cyan technology glow. Welcoming, imaginative, joyful, uncluttered; attractive to children while credible for adult educators. Wide balanced composition with the main figures and book safely inside the central area; retain generous breathing room and avoid cropping faces, hands or book edges. Completely fictional people, no resemblance to any real child. No Disney characters or designs, no existing copyrighted character likenesses, no logos, no photographic faces. NO text, NO letters, NO numbers, NO watermark, NO title: the bilingual title is rendered as real text below the image. Opaque background.
```

## 核对

已实际查看生成结果：宽幅约2.4:1，书本、儿童与机器人主体清楚，无标题文字、Logo或已知IP角色；PNG保留原始生成结果，WebP用于网站加载。人物和故事场景为概念设计。

更新后检查中英文README主图引用、相对路径、图像尺寸、GitHub渲染与线上文件；结果随提交保存。

## 实际发布核对结果

- 本地主图及中英文入口等49个内部引用均存在；PNG/WebP为1942×809，优化版360,904B，图像无EXIF/XMP身份元数据。
- GitHub README API渲染包含新主图路径及可见AI生成概念图说明。
- 独立Chrome 155.0.8059.40访问公开GitHub页面：375px主图显示309×128.72px，1280px显示838×349.09px；自然尺寸1942×809，未裁切或造成横向溢出，英文首页加载同一图。
- Pages的PNG/WebP完整下载SHA-256均与本地一致。课程与PDF文件未修改，旧SVG保留。

[浏览器核对](%E8%AF%81%E6%8D%AE/%E4%B8%BB%E5%9B%BE%E6%B5%8F%E8%A7%88%E5%99%A8%E6%A0%B8%E5%AF%B9v1.0.json) · [线上图像核对](%E8%AF%81%E6%8D%AE/%E4%B8%BB%E5%9B%BE%E7%BA%BF%E4%B8%8A%E6%96%87%E4%BB%B6%E6%A0%B8%E5%AF%B9v1.0.json)。本次使用独立自动化Chrome检查公开页面；聊天的内嵌浏览器绑定超时，不记为内嵌浏览器人工检查通过。