Martin Delophyandhaixin.yang
a6b7a9e51b
* fix editor media caption and audio workflows * improve editor timeline and caption voice workflows --------- Co-authored-by: haixin.yang <haixin.yang@weimob.com>
2 lines
1.9 KiB
HTML
2 lines
1.9 KiB
HTML
<!doctype html><html lang="zh-CN"><head><meta charset="UTF-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>技术文章 | Timeline Studio</title><meta name="description" content="Timeline Studio 的浏览器 AI、WebGPU、ONNX 与本地优先视频编辑工程实践。"><link rel="stylesheet" href="/articles/article.css"></head><body><header class="site-header"><nav class="nav"><a class="brand" href="/"><span class="brand-mark">TS</span>Timeline Studio</a><div class="nav-links"><a href="/">打开编辑器</a><a href="https://github.com/MartinDelophy/ai-video-editor">GitHub ↗</a></div></nav></header><main class="index-wrap"><div class="index-title"><p class="eyebrow">Engineering Notes · 2026</p><h1>把浏览器推到创作工具的边界</h1><p class="deck">两篇来自 Timeline Studio 真实代码与验证数据的工程手记：一篇谈完整系统，一篇深入神经网络算子。</p></div><section class="article-grid"><a class="article-card" href="/articles/local-first-browser-ai-video-editor/"><img src="/articles/assets/local-first-architecture.webp" alt="浏览器端本地优先 AI 视频编辑器架构概念图"><div class="article-card-body"><span class="tag">系统架构 · 12 分钟</span><h2>把 AI 视频编辑器装进浏览器：本地优先架构的六个关键决策</h2><p>从 Worker、模型缓存到多轨时间线与导出，拆解一个真正可用的浏览器 AI 工作流。</p></div></a><a class="article-card" href="/articles/liveportrait-webgpu-port/"><img src="/articles/assets/liveportrait-webgpu.webp" alt="LivePortrait 5D 张量到 WebGPU 的转换概念图"><div class="article-card-body"><span class="tag">WebGPU 深潜 · 15 分钟</span><h2>5D GridSample 进不了浏览器之后：LivePortrait 的 WebGPU 移植实战</h2><p>如何改写 ONNX 图、验证数值等价，并让大型数字人管线在浏览器里完成闭环。</p></div></a></section></main></body></html>
|