Skip to content
OmniConvert

PDF files: why the text is so hard to get back out

What a PDF actually stores, why converting one to Word loses the layout, and how a scanned PDF differs from a digital one. Convert PDF in your browser, no upload.

At a glance

Type
Documents
Extensions
.pdf
MIME type
application/pdf
Also known as
Portable Document Format

A PDF describes a finished page rather than a document. Its content stream is a sequence of drawing operators — set this font, move to these coordinates, show this string of glyphs, stroke this path — inherited almost directly from PostScript, which Adobe designed to drive printers. That lineage explains both of the format's defining properties: it looks identical everywhere, because nothing is left for the viewer to decide, and it is unusually hard to extract from, because the structure a human sees was never written down.

Read more

Concretely, a PDF has no paragraphs, no lines and often no words. It has positioned glyph runs. The sentence you are reading might be stored as a dozen separate runs at computed coordinates, and the space between two words may be an absence of ink rather than a space character. Every PDF-to-text and PDF-to-Word converter, including this one, is therefore reconstructing: grouping runs by baseline to recover lines, guessing paragraph boundaries, inferring headings from type size. It works well for ordinary prose and it flattens columns and tables, because the information needed to keep them was never in the file.

The second thing to know is that "PDF" covers two entirely different situations. A PDF exported from Word or a browser contains real text you can search and copy. A PDF produced by a scanner or a phone camera contains photographs of pages and no text at all — the words are pixels, and nothing can extract them without character recognition. This site detects that case and runs an AI recognition model on your own machine rather than uploading the file, which is unusual: OCR is normally the point where a converter sends your document to a server.

PDFs are also where privacy stops being abstract. The format is what contracts, payslips, bank statements, medical letters, invoices and scanned identity documents travel as. Every conversion here runs inside the page — the PDF is opened, rendered and read by code in your browser, and no request carries the file. That is the reason this converter is built the way it is, and PDF is the category where it matters most.

PDF tools

Page-level edits — no conversion involved. All 14 run on your device.

Convert PDF to other formats

Convert other formats to PDF

FAQ

Why does PDF to Word never look right?

Because a PDF stores positioned glyphs, not structure. Converting means inferring paragraphs, headings and reading order from coordinates, and there is no way to recover a table or a two-column layout that was never recorded as such. A converter that keeps the layout better is doing it by rendering the page on a server with a full office suite — which means receiving your file.

How can I tell if my PDF is scanned?

Try selecting text in a PDF reader. If you can select and copy words, the file has a text layer. If your cursor draws a box over a picture instead, it is a scan and any text extraction needs character recognition.

Does the OCR here upload my document?

No. When a PDF has no text layer, an AI recognition model — a five-million-parameter neural network — is downloaded to your browser and runs there. About 21 MB the first time, then cached. Every competing converter that offers OCR sends the file to a server instead.

How accurate is the recognition?

Around 98% of characters on clean printed documents in our own testing, Chinese and English alike. Real scans are harder — stamps, handwriting, low resolution and a page fed in crooked all reduce it. Treat the output as a draft to check against the original.

Why is the file size limit for PDFs so much lower than for video?

80 MB, because pages are rendered into memory as full-resolution bitmaps and that cost is measured in pixels rather than in how well the PDF compressed. A video conversion streams through the file; a PDF page has to exist as a raster all at once.

Which PDF conversion should I choose?

TXT if you want the words to search or paste. DOCX if you want to edit and can accept losing the layout. JPG or PNG if you want the page to look exactly as it does now — JPG for photographic scans, PNG for text and diagrams.

Can I turn images into a PDF?

Yes, and losslessly: PDF carries JPEG and PNG data natively, so the image is embedded exactly as it is rather than being re-compressed. The resulting page matches the image's pixel dimensions rather than being fitted to A4.

Are password-protected PDFs supported?

No. An encrypted PDF cannot be opened for reading or rendering, and the job fails with a message saying so rather than producing blank pages. Remove the password in a PDF reader first.