OCR issues
Poor recognition, wrong language, missing text layer, and edits lost after re-OCR.
When recognition comes back garbled, in the wrong language, won't search, or quietly overwrites a value you typed in, the fix is usually one setting away. Work through the problem that matches what you're seeing — each one is a quick recipe you can run from the open document.
Screenshot
The Viewer with a document open: the OCR button at the top of the right sidebar, with the right sidebar's Enhance tab alongside it. — shot troubleshooting-ocr-issues-01
Recognition is inaccurate or full of garbage characters
Bad accuracy almost always comes from one of three things: the wrong language is set, the engine isn't a good fit for the document, or the image is too noisy to read. Try them in this order — language first, because it's the most common cause and the fastest to change.
Set the right languages. If the text is in a language the recognizer isn't expecting, it guesses — and guesses badly. Set the languages that actually appear on the page so the engine stops guessing. See OCR languages for where to set them per template and how the Default OCR languages fallback works.
Try the other engine. Scanix ships more than one OCR engine, and they have different strengths. If one engine struggles with your document, switch to the other on the OCR Engine tab in Settings and run OCR again. The engine choice saves instantly — no Apply step — so you can switch and re-run in seconds.
Enhance the image first, then re-run. Skew, speckle, faint print, and busy backgrounds all drag recognition down. Clean the page up with the enhancement tools, then recognize the cleaned image. When the page has already been OCR'd, the apply button reads Apply & Re-OCR and refreshes recognition against the new image in one step; on a page that hasn't been recognized yet it reads Apply and leaves OCR to the top button. See Image enhancement.
Set the language before you re-run, not after
Changing languages doesn't re-read the page on its own — it only affects the next OCR run. After you adjust languages or the engine, run OCR again so the new settings take effect.
The exported PDF isn't searchable
If you can't find text inside an exported PDF, the document has no OCR text layer yet. Searchable output is built from OCR results — no recognition, no searchable layer.
Run OCR on the document. Open it in the Viewer and run OCR from the top of the right sidebar. See Run OCR for the full walkthrough.
Export as Searchable PDF. In the Viewer's Export Document dialog, choose Searchable PDF — a PDF with an invisible text layer, so the whole document is full-text searchable.
Export formats stay greyed out until OCR runs
Every format in the Export Document dialog needs OCR results first. Until the document has been recognized, the formats are greyed out and read Run OCR first to enable this format. Run OCR, and they unlock.
My edits disappeared after re-running OCR
This is expected behaviour, not a bug: re-running OCR produces a fresh result. When you recognize a page again, the engine reads it from scratch and writes a brand-new recognition result for the page — it doesn't merge the new reading with corrections you typed in. So a value you fixed by hand can be replaced by what the new pass found.
The fix is to keep your corrections before you re-run. Save the text edits you want to keep first, then click Re-OCR. If a re-run does overwrite something, just retype it once the run finishes. See Edit & correct text for how saving works.
Save edits before you click Re-OCR
Running Re-OCR replaces the page's recognized text with a clean reading. Save any text edits you want to keep before you re-run — there's no merge, so an unsaved correction can be overwritten by the new pass.
Auto-save doesn't protect against re-OCR
Auto-save OCR corrections (on by default) saves your text and region edits as you make them — but a fresh Re-OCR still rewrites that page's recognized text. Auto-save isn't a substitute for finishing your edits before you re-run. See Edit & correct text.
Troubleshooting
Next steps
Run OCR
Recognize a document in the Viewer, page by page or all at once.
OCR languages
Set the languages on the page so the engine stops guessing.
Image enhancement
Deskew, clean, and sharpen a page so recognition has a clear image to read.