How to Translate a Scanned PDF: OCR, Full Translation, Layout Intact

How to Translate a Scanned PDF: OCR, Full-Document Translation and Bilingual Output with the Layout Intact Every page of a scanned PDF is really a picture: you cannot select or copy the text, so browser translation extensions and pop-up dictionaries have nothing to work with. The usual workaround, screenshot, recognise, paste into a translator, is slow and scrambles tables, columns and captions. OmniDoc does OCR, full-document translation and layout reconstruction in a single job : upload the scan and get back a translated PDF that looks like the original. Why scans are hard to translate No text layer : the words and their positions on the page must be recognised before anything can be translated. Complex pages : columns, tables, headers, footers, footnotes and captions sit side by side; read them in the wrong order and the translation comes out jumbled. Text length changes : a translation is longer or shorter than the source and still has to fit back in place without overflowing. Text inside images : chart axes and diagram labels need handling too. How OmniDoc handles it Automatic OCR : scanned PDFs and JPG/PNG images are recognised automatically; no conversion step is needed first. Layout kept : the translation goes back into the original paragraphs, tables, images and formulas. Bilingual or translation only : the bilingual PDF places each source page beside its translated page for line-by-line checking, or you can keep just the translation. Text in embedded images : for text-based PDFs and Word files, turn on image parsing to translate the text inside pictures as well. 90+ languages : English, Chinese, Japanese, Korean, German, French, Spanish, Russian and many more, in any direction. Step by step Sign in at omnidoc.top (or open Document Translation from the OmniDoc desktop home screen). Upload the scanned PDF or image and choose the target language. Pick options to match the document: the Academic style for papers, Legal for contracts, Technical for manuals; turn on deep thinking for difficult text and context-aware translation for long documents so terms stay consistent. Choose bilingual or translation only, start the job, and download the result when it finishes. Tips for better recognition Scan pages flat, the right way up, and at a reasonable resolution. Upload a whole book or a stack of documents in parts so you can review them batch by batch. When names, organisations and technical terms matter, add their fixed translations to a glossary first. For important documents, check the bilingual version page by page, especially numbers, amounts and dates. Pricing OCR and translation run in the same job and are charged by what is delivered; failed jobs cost nothing. Before a PDF starts, the balance must be above page count × 2 credits. Credit packs start at 3,000 credits for ¥9.9, and purchased credits do not expire. Let an AI assistant do it If OmniDoc is connected to Codex, Claude Code or Claude Desktop, just ask: “Use OmniDoc to translate D:\Docs\scanned-contract.pdf into English, bilingual, legal style.” The assistant uploads the file, waits for the job and saves the translation next to the original. Setup guide: https://omnidoc.top/blog/en/ai-word-excel-ppt-omnidoc/ . Start translating at https://omnidoc.top/en/ or download the desktop app at https://download.omnidoc.top/ . For how translation extensions differ from full-document translation, see https://omnidoc.top/blog/en/translation-plugin-pdf-word-layout/ . - 1 -