Privacy & About

The short version: your files stay on your device.

What happens to your files

Resizing happens inside your browser, on your own device.
Photos and signatures are not uploaded, stored or logged.
No accounts. Visits are counted with Google Analytics (see below), but your files and their names are never part of that. Close the tab and nothing of yours remains.
The site loads fonts and tools from itself only. The page blocks requests to other sites, except the Google Analytics visit counter.

Counting visits (Google Analytics)

We use Google Analytics to count visits and which pages are opened, so we know whether FormSizer is useful.
It records page views only. Your files, their names and anything you type never go to Google.
Google sets cookies and sees your approximate location, browser and device type. Advertising features and personalised ads are turned off.
If your browser sends "Do Not Track" or "Global Privacy Control", visits are not counted at all.
Ads from Google AdSense may appear on information pages, never on the pages where you work with files. They are requested as non-personalised and are skipped for the same privacy signals. The privacy policy in the footer has the details.

Resize Anything (images, PDFs and Word files)

The Resize Anything tools work on any image, PDF or Word file, not only exam forms. They run entirely on your device, just like the exam pages.
PDFs are read with pdf.js (Apache-2.0) and built with pdf-lib (MIT). Both are stored on this site, including the pdf.js worker, fonts and colour tables. Nothing is fetched from anywhere else.
The PDF tools only download when you open the PDF page. Large PDFs need a lot of memory: over 50 MB or 100 pages, a phone may struggle.
Passwords: if a PDF is protected, we say so and stop. We never try to break or bypass a password.
"Keep text sharp" only tidies the file, so savings are usually small. "Shrink harder" turns each page into a picture to reach a small size, so text can no longer be selected or searched, and a very small limit may be impossible.
Fillable form fields, bookmarks and links may not survive when pages are rebuilt. Always check the result before you use it.
PNG cannot be shrunk by lowering quality. To reach a KB limit use JPG or WebP, or let us make the picture smaller in pixels. WebP saving depends on your browser.
Word files (.docx): a .docx is a zip of parts. We open it on your device with JSZip (MIT licence, stored on this site and only downloaded when you open the Word page), make the pictures inside smaller, and zip it up again. Your original file is never changed; you get a new file named "-resized".
Word limits: only .docx (also .docm and .dotx) is supported. Old .doc files are not. Password-protected files are refused, and we never try to get around a password. We do not draw the pages, so there is no preview and we cannot count pages. Macros are kept but never run.
Shrinking only works on pictures, so a file that is mostly text may not reach a small limit. We say so and show the smallest size we could make. Changing the page size makes the text reflow, so the number of pages and the layout change; tables and pictures are not resized. Open the new file in Word and check it.

Converting between Word and PDF (approximate)

Word to PDF and PDF to Word run on your device, with no server. That is why they are approximate: we do not have Word or a PDF app inside the browser, so we draw or rebuild the pages ourselves. Each conversion lists what it simplified or left out. For a perfect result, use "Save as PDF" in Word, or "Save as Word" in your PDF app.
Word to PDF reads the .docx with JSZip (MIT), works out lines and pages itself, and draws a new PDF with pdf-lib (MIT), so the text stays selectable. It handles paragraphs, headings, bold, italic, underline, sizes, colours, lists, simple tables, JPEG and PNG pictures, page size and margins, headers, footers and links. It leaves out SmartArt, charts, text boxes, floating pictures, columns, footnotes, equations and fresh field values, and it says so for your file.
Fonts: by default the PDF uses its built-in fonts, which only cover Western European letters; other characters become "?" and we warn you. If you turn on "Embed fonts", a free font (Noto Sans and Noto Sans Devanagari, SIL Open Font Licence, stored on this site, about 0.4 MB, downloaded only then) is saved inside the PDF so Hindi and more letters show correctly. The font uses fontkit (MIT).
PDF to Word has two modes. "Editable text" reads the text and its positions with pdf.js and guesses paragraphs, headings and bullet points. It does not rebuild pictures, tables or columns. "Page pictures" saves each page as a picture inside a Word file, so it looks exactly like the PDF but cannot be edited.
Scanned PDFs have no real text. Reading them needs OCR (text recognition). It is not built in: an OCR engine such as tesseract.js (Apache-2.0) needs several MB of program and language data. It may come later as an optional download. For now, use "Page pictures".
We only write the Word file after checking that it is a sound package. We could not open the results in Word on your device for you, so open them and compare them with the original.

Signature clean-up

It lifts dark ink off the paper and puts it on a plain colour. Nothing is downloaded and nothing leaves your device.
It works best with flat lighting and a dark pen on white paper. Shadows, creases and very faint ink can still show up.

Photo background cleaning

It is optional and off until you turn it on.
It uses Google's MediaPipe image segmenter and its selfie model, both under the Apache-2.0 licence.
The tool and the model are stored on this site and run on your device. Your photo is never sent to Google or anyone else.
First use downloads about 15 MB, once, then it works offline.
It is not perfect. Check hair and shoulder edges. Plain portrait photos work best.

About the requirement data

Exam bodies change photo and signature rules often, and unofficial sites disagree with each other.
Each exam shows whether its numbers were checked against the official notification, with the link and date.
Anything not checked is marked "Specs not verified yet". Always compare with your notification.

Created by

Abhishek Kumar

A .NET full stack developer from India who built FormSizer to make exam-form files easier to prepare, without uploading them anywhere.

FormSizer is open source under the MIT licence. Not affiliated with any exam body. Exam names, Microsoft Word and other product names belong to their owners and are used only to say what the tool works with.