How to process a sensitive file without uploading it
Use a tool that runs inside your browser instead of on someone else's server. The file is read into browser memory, changed by JavaScript on your own device, and saved back to your disk. It never travels. Every PixMidas tool works this way; this page covers how, which jobs, and where it stops working.
Why most online tools upload your file
The usual design for a web tool is straightforward: your browser sends the file to a server, the server does the work with software installed there, and it sends the result back. That is how most online PDF and image services operate, and it is why they can offer heavy features like OCR. The cost is that your document sits, however briefly, on a machine you do not control, covered by a retention policy you have to take on trust.
Browsers no longer need that arrangement for most everyday file jobs. The APIs required to read, decode, re-encode and write files are built into the browser itself.
What actually happens when you drop a file in
The sequence is the same across the site. You choose a file, and the browser's File API hands the page a reference to it in local memory. Nothing has been sent anywhere at this point, and nothing will be.
For images, the file is decoded and redrawn through the HTML5 Canvas API, then re-encoded at the settings you picked. For PDFs, pdf.js reads and renders the document and pdf-lib writes the new one. HEIC photos from an iPhone are decoded by a WebAssembly build of libheif, again in the browser. The finished result is created as a blob in memory and offered to you as a download.
Those libraries are served from pixmidas.com itself rather than from a public CDN, so opening a tool and using it does not produce a request to any third party. Close the tab and the file, the result and everything in between are gone from memory.
Which sensitive jobs this covers
Real situations, and the tool that handles each one locally.
A signed contract is too large to attach to an email or upload to a client portal.
PDF Compressor pdf.js reads the document and pdf-lib rewrites it at a lower image quality, both inside the browser.
Names, account numbers or addresses in a document have to be blacked out before it is shared.
Document Redactor Each page is rendered to a canvas, the redaction is painted on, and the page is rebuilt as an image inside a new PDF. The original text underneath is gone, not hidden, so it cannot be selected or recovered from the exported file.
A photo about to be posted publicly still carries GPS coordinates, a camera serial and a timestamp.
EXIF Remover The image is redrawn through an HTML5 canvas and re-encoded. Canvas output carries no metadata, so every EXIF field is dropped.
A confidential report needs a visible ownership mark before it goes to a third party.
PDF Watermark pdf-lib stamps the text onto each page and writes a new file in browser memory.
Only pages 4 to 9 of a medical or legal file should be sent, not the whole document.
Split PDF pdf-lib copies the selected pages into a new document. The pages you leave out are never written to the output.
Several scanned statements have to arrive as one file.
Merge PDFs pdf-lib loads each file into memory and writes a single combined document locally.
Photographed or scanned identity documents need to become one PDF for an application.
Images to PDF Each image is embedded into a new PDF by pdf-lib, in the browser.
A single page of a sensitive PDF has to be sent as an image rather than as a document.
PDF to JPG pdf.js renders the page to a canvas and the canvas is exported as a JPG. No page you did not export leaves the file.
Images follow the same pattern: Image Compressor to meet a size limit, Image Resizer for exact dimensions, HEIC to JPG for iPhone photos, and Image Watermark before sharing a photo you own.
Redaction is the case where the mechanism matters most
Drawing a black rectangle over text in a PDF viewer hides nothing. The text is still in the file and anyone can select it, copy it, or extract it with a script. This is how redaction failures reach the news.
The Document Redactor works differently. It renders each page to a canvas, applies your redaction to the actual pixels, and then builds a new PDF in which every page is that flattened image. The text underneath is not covered, it is absent from the output. The trade-off is worth knowing before you rely on it: the exported PDF is image-based, so its text is no longer selectable or searchable by anyone, including you.
Who this matters to
Lawyers preparing documents for disclosure. Journalists handling material from a source. HR and recruitment staff moving CVs and payslips around. Accountants sending client statements. Healthcare administrators handling patient letters. Anyone submitting identity documents to an application portal. In each case the file is routine, but sending it through a third-party server to shrink it by two megabytes is a risk taken for no reason.
Where this approach stops working
Being straight about the limits is part of the point.
- Size. Image tools accept files up to 50 MB and PDF tools up to 100 MB. Beyond that, and sometimes below it on a modest phone, you are limited by the memory the browser has available.
- No OCR. Text inside a scanned page stays an image. Nothing here makes a scan searchable.
- No deep PDF editing. You can compress, merge, split, sign, watermark, redact and convert. You cannot rewrite a paragraph inside an existing PDF.
- Browser required. JavaScript must be enabled, and you need a current version of Chrome, Edge, Firefox or Safari.
- The page request is still a web request. Loading the site reaches a server and a CDN, and your IP address is part of that, as with any website. The page also loads Cloudflare Web Analytics, which counts visits without setting cookies. None of it carries file names or file contents.
If your task needs OCR or content-level PDF editing, a server-based service is the right tool and it is worth choosing one whose retention policy you have actually read.
Frequently asked questions
How can I edit a PDF without uploading it to a website?
Use a tool that runs in the browser rather than on a server. On PixMidas the file is read into browser memory through the File API, processed by JavaScript on your own device using pdf.js and pdf-lib, and written back out as a download. No network request carries the document, and nothing is stored once you close the tab.
How do I process a sensitive document locally?
Open the tool you need, drop the file in, and do the work before the file ever reaches a network. On PixMidas that covers compressing, merging, splitting, watermarking, signing, redacting and converting PDFs, and compressing, resizing, converting and stripping metadata from images. All of it runs client-side. The practical limits are your device memory, a 100 MB cap on PDF tools and a 50 MB cap on image tools.
Is a browser-based tool actually more private than an upload-based one?
For the file itself, yes, because the file never travels. What still travels is the ordinary web request for the page: your IP address reaches the server and the CDN, as with any website, and the page loads Cloudflare Web Analytics, which counts visits without setting cookies. Nothing about your file is recorded, because nothing about your file is ever sent.
Can redactions be reversed in the exported file?
No. The Document Redactor does not draw a black box over live text. It renders each page to a canvas, applies the redaction to the pixels, and builds a new PDF from those images. The text underneath no longer exists in the output. The trade-off is that the exported PDF is image-based, so its text is no longer selectable or searchable.
What are the limits of processing files in the browser?
Three things. Size: image tools accept up to 50 MB and PDF tools up to 100 MB, and very large files are limited by the memory your device has free. Capability: there is no OCR, so text inside a scanned page stays an image. Environment: JavaScript must be enabled and you need a current version of Chrome, Edge, Firefox or Safari.
Does anything at all get sent to a server?
The page itself is downloaded from the server the first time you open it, like any website. After that, the file you choose is handled entirely by code already running in your browser. PixMidas serves every library it uses from its own domain, so processing a file does not trigger a request to any third party either.