企评社

Interface labels are shown in English. Company reviews, forum posts and tool descriptions are currently available in Chinese only.

PDF compression

PDF too big? Two ways to compress: lossless keeps the text layer, lossy shrinks the file a lot

文件不上传服务器:压缩全程在你的浏览器里完成

把 PDF 拖到这里,或点击选择

一次处理一个文件;原文件只被读取,导出的是一份新 PDF,原始文件不会被修改

文件不会上传到服务器:解析、渲染与生成新 PDF 全部在你的浏览器里完成

PDF 为什么会很大,该怎么压小

体积几乎都花在图片、字体和重复对象上。PDF 里的文字本身极小,一份上百页的纯文字文档常常不到 1MB;真正撑大文件的是三样东西:内嵌图片(扫描件一整页就是一张高分辨率大图,动辄几 MB)、嵌入的字体文件,以及大量细碎对象与重复资源(同一张 Logo 被贴了几十次却各存一份)。所以「PDF 太大」通常等价于「里面有图片或冗余结构」。

无损压缩解决结构冗余,有损压缩解决图片体积。无损压缩只重排文件的内部结构(对象流、去掉冗余对象),一个像素都不动,因此文字层、书签、链接、表单全部保留,压缩后依旧可以搜索、复制、选中文字——但收益有限,多数文件只有 0~15%,本来就被其他工具优化过的文件甚至可能略微变大。有损压缩是把每一页按目标清晰度重新渲染成 JPEG 图片,再拼成一份新 PDF:体积通常能降到原来的几分之一甚至十分之一,代价是文字层被彻底丢弃,压缩后的 PDF 不能再搜索、复制或选中文字。

扫描件该选哪一档。扫描件本身就是图片,用有损压缩最常见:110 DPI + 质量 65% 一般能把体积压到原来的三分之一到十分之一,屏幕阅读完全够用;如果还要打印或给 OCR 识别,选 150 DPI 更稳妥。反过来,合同、论文、说明书这类以文字为主的 PDF 请不要用有损压缩——它们本来就不大,压不出多少体积,却会白丢文字层,优先用无损模式。

为什么我们不做「上传到服务器压缩」。上传意味着把合同、发票、证件、工资条交给一台你无法验证的服务器,压缩完再传回来;很多同类站点还顺手留下副本。本页的做法是:用 pdf.js 在你的浏览器里把页面渲染到画布,用 pdf-lib 在你的本机生成新的 PDF,全程没有任何文件上传请求(可打开开发者工具 Network 面板核验),原始文件也不会被修改。代价是慢一点、吃一点本机内存,换来的是这些材料不必离开你的电脑。

FAQ

Lossless compression barely shrank the file - did it fail?

That is a normal outcome, not a failure. Lossless compression can only rearrange structure; on an already-optimized PDF there is often nothing to spare, and some files even grow slightly. When that happens we tell you straight - 'this file is already well-structured, lossless compression gains little' - instead of pretending success. For real shrinkage, switch to lossy compression.

Can I still search or copy text after lossy compression?

No. Lossy compression renders every page into an image: the text layer no longer exists, pages cannot be selected, copied or searched. To keep text, use lossless compression, or run an OCR tool to extract text first.

Which quality tier suits scanned documents?

The default 110 DPI at 65% quality fits most on-screen reading with the smallest size; pick 150 DPI for printing or OCR feeding; use 200 DPI only when you truly need to zoom into details - size grows noticeably. Raising clarity only makes the file bigger; it never makes a blurry scan sharper.

Can it squeeze a PDF under 1MB?

There is no 'one click to 1MB' guarantee - the final size depends on page count and content. Image-heavy scans usually can get under 1MB with lossy compression at lower quality / lower DPI; text-only documents are small to begin with and need no force. Best practice: run the default tier once, read the size in the report, then decide whether to push harder.

Will my original be modified? Can password-protected PDFs be processed?

Your original is neither modified nor uploaded: the tool reads a copy in browser memory only, exports a brand-new file, and loading another file clears the copy. PDFs with an open password cannot be parsed by the browser - first open one in a reader and 'save as' a copy without the password, then compress that.

Related tools