DjVu to PDF: Convert Old Scanned Books and Papers

We took a 363-page scanned book from the Internet Archive. As a DjVu file it was 4.3 MB. Converted to PDF at 300 DPI, it was 70 MB.

Nothing went wrong. That's what happens when a format built in a research lab to compress scans meets a format built to open everywhere. DjVu really is better at storing scanned pages. It just lost the war for being opened, so if you have one, you probably need it as a PDF. The trick is converting without ending up with a 70 MB file you didn't need.

Quick answer: To convert DjVu to PDF, render each page to an image and place the images in a PDF. DjVu to PDF does it in your browser: drop in the .djvu file, pick 150 DPI for reading on a screen (72 to keep it small, 300 only for printing), and download the PDF. The result has no searchable text, even if the DjVu had some, so run it through OCR PDF afterwards if you need to search or copy.

Why DjVu Files Are So Small

DjVu came out of AT&T Labs in the late 1990s, from a team that included Yann LeCun and Léon Bottou, later better known for their work on neural networks. Their idea was to stop treating a scanned page as one photograph.

A DjVu page is split into layers (Wikipedia has the details). A sharp black-and-white mask, around 300 dpi, holds the text and line art. Behind it, a background layer at around 100 dpi holds paper texture, photos and colour, where nobody notices the lower resolution. Each layer gets a compressor suited to it: wavelets for the pictures, and for the mask a scheme called JB2 that notices the letter "e" appears hundreds of times on a page and stores its shape once.

The developers' own figures: a black-and-white technical paper in 15 to 40 KB a page, a colour magazine page in 40 to 70 KB, where a decent JPEG of the same page needs around 500 KB. That's why digital libraries, Wikisource and university archives built whole collections in DjVu in the 2000s, and why so many old journals, sheet music and out-of-print books still circulate in it.

Why PDF Won Anyway

Not by being better at scans. By being already installed.

DjVu needed a plugin in the browser and a separate viewer on the desktop. When browsers dropped NPAPI plugins around 2015, the DjVu plugins went with them. In February 2016 the Internet Archive stopped creating DjVu files for new uploads, citing the format's declining use and the trouble of keeping its viewer alive. Phones never had a default DjVu reader. Today a few dedicated apps (SumatraPDF and Okular on the desktop, KOReader on e-readers) still open it, and almost nothing else does.

So you convert, and you accept the trade.

The Size Trade, Measured

Here's what that 363-page book turned into, using the same pipeline as our converter: each page rendered at the chosen DPI and saved as a JPEG at 85% quality.

The Internet Archive's PDF gets close to DjVu because it's built the way DjVu is, with text and background stored as separate layers, by a long server-side process. A converter that renders each page to one JPEG can't match that. What it gets you is a plain PDF that opens anywhere, sized by the DPI you choose.

That book's pages are small, about 4.25 by 5.5 inches, so at 150 DPI each page is 638 by 825 pixels. That's crisp on a laptop and a phone. 300 DPI gives you 1,275 by 1,650, which only pays off on paper. The rule of thumb:

• 72 DPI: smallest file. Readable on a screen for plain text, soft on footnotes and small type.
• 150 DPI (the default): the right answer for reading on a screen and sharing.
• 300 DPI: for printing, or for a scan with tiny print or detailed figures you'll zoom into.

What Happens to the Text Layer

Many DjVu files carry a hidden OCR text layer. In our sample, every one of the 363 pages had one. It's what lets you search a DjVu in a viewer that supports it.

DjVu to PDF does not carry that layer over. Each page becomes a picture, and the PDF it produces can't be searched or copied from. We'd rather say that up front than have you find out with Ctrl+F.

The fix takes one extra step. Send the converted PDF to OCR PDF, which recognises the text and lays an invisible, searchable layer over each page, the same way the DjVu did. It runs in your browser too. Old typefaces, long s and faded print reduce accuracy, so spot-check a page before trusting a search; our guide to OCR for scanned PDFs covers what affects quality.

How to Convert DjVu to PDF With OxygenPDF

1. Open DjVu to PDF and choose your .djvu or .djv file. The decoder (a WebAssembly build of a DjVu engine) loads on demand and shows you page one.
2. Pick the resolution: 72, 150 or 300 DPI.
3. Set JPEG quality. 85% is the default; drop toward 70% for a smaller file of a plain text scan, raise it for photos and colour plates.
4. Convert. The tool renders the pages one by one and shows its progress, then gives you a PDF with the same name.
5. If you need search, open OCR PDF and drop the converted file in.

One limit to know: the tool reads bundled DjVu files, the usual kind where every page lives inside one file. Some archives publish indirect DjVu, an index file pointing to a folder of separate page files. A browser can't reach those companion files, so the tool rejects them with a message instead of producing half a book. Bundle them first with DjVuLibre's djvmcvt tool, or look for a single-file download of the same item.

Making the PDF Smaller

If the 150 DPI file is still bigger than you'd like, work through these in order:

1. Lower the JPEG quality and convert again. Text scans hold up well down to about 70%.
2. Drop to 72 DPI if it's for reading on a phone or a laptop and the type isn't tiny.
3. Remove pages you don't need, like blank versos and library stamps, with Delete Pages.
4. Run OCR last. Compress PDF can shrink the file further, but it re-renders every page as an image, which would throw away a text layer you'd already added. Compress first, then OCR.

And if the size matters more than anything else, keep the DjVu as your archive copy and share the PDF. Nothing says you have to throw the original away.

Your Scans Stay on Your Machine

Archive scans are often someone's research, or a family's papers, or a book you scanned yourself. DjVu to PDF decodes and renders everything in your browser tab, so the file is never uploaded and there's no size limit except your device's memory. For why every OxygenPDF tool works this way, see why local-first PDF tools.

Frequently Asked Questions

What is a DjVu file?

DjVu is a document format for scanned pages, created at AT&T Labs in the late 1990s. It splits each page into a high-resolution text layer and a lower-resolution background and compresses each separately, so scanned books come out far smaller than as JPEGs or most PDFs. It's common in digital libraries and academic archives.

Will the PDF be searchable?

Not straight out of DjVu to PDF. The pages are converted as images and the DjVu's text layer is not copied over. Run the PDF through OCR PDF to add a searchable text layer.

Why is my PDF so much bigger than the DjVu file?

DjVu stores text and background separately and compresses each with a specialised method. Our converter stores each page as one JPEG image, which is far less compact. A 4.3 MB DjVu book came out at 7.2 MB at 72 DPI and 24.8 MB at 150 DPI in our test.

What DPI should I use to convert DjVu to PDF?

150 DPI for reading on a screen and sharing, 72 DPI when file size matters most, and 300 DPI only for printing or for pages with very small print.

Why won't my DjVu file open in the converter?

It's probably an indirect DjVu: a small index file that refers to separate page files. Bundle the pages into one file with DjVuLibre's djvmcvt, or download the single-file version of the document, then try again.

Got an old scan only one app on your computer can read? Convert your DjVu to PDF in your browser. Nothing is uploaded.