Updated:
Recognize text from a scan or photo with on-device OCR, correct important recognition errors, and use the Translate action for the extracted text. The original page remains available for comparison.
Scan or import the page in ScanLens, run OCR in the correct source language, review the recognized text against the image, correct consequential errors, and then use the Translate action for the extracted text. Translate page by page or in smaller sections when the document is long or has columns, tables, stamps, or mixed languages.
This workflow translates recognized text; it does not promise to replace the original PDF layout with a translated version. The scan remains the visual source of truth. Names, dates, totals, legal terms, and technical language should be checked by a fluent reviewer or qualified translator when accuracy matters.
OCR converts pixels into characters. Blur, glare, small type, handwriting, complex columns, and the wrong source language can introduce errors before translation begins.
The Translate action works from the OCR result. A perfectly translated OCR error is still wrong, which is why the recognized source text must be checked first.
Tables, signatures, stamps, spacing, and page design remain in the original scan. Use the page image to understand where each extracted passage belongs.
Machine translation is useful for understanding and drafting. It is not a substitute for a qualified translator when a document controls legal rights, medical care, immigration, finance, or safety.
ScanLens recognizes text from a document and exposes actions to copy, translate, export, and edit that recognized text. This is a strong fit for understanding a letter, translating selected paragraphs, preparing a draft, or moving checked wording into another app.
The page does not promise automatic replacement of every line inside the original PDF, preservation of fonts and column geometry, certified translation, or reliable interpretation of handwriting and seals. For full document production, use the recognized text as input to an appropriate translation and layout workflow, then compare the finished result with the source.
Shadows across a line can change letters and punctuation. Diffuse daylight or balanced room light usually produces cleaner recognition than a single hard lamp.
More pixels per character improve recognition. Move closer while keeping every page edge visible and the camera parallel to the document.
Translate columns, tables, captions, and side notes in smaller logical blocks when a single reading order would mix unrelated text.
Amounts, dates, measurements, model numbers, account details, and addresses are high-impact fields. Compare every one with the scan before relying on the translation.
ScanLens can recognize text from scans and provide a Translate action for that recognized text. Review and correct the OCR result before translating important content.
This page does not claim automatic translated-layout replacement. The workflow translates recognized text while the original scanned page remains the visual reference.
The current public product fact is exactly 14 OCR languages. Translation language availability is a separate matter and can depend on the language pair and current system services.
Handwriting recognition is narrower and more variable than printed-text OCR. Test representative writing, review the recognized source text closely, and do not rely on it for consequential content without human verification.
It can help with initial understanding, but consequential documents should be reviewed by a qualified person. OCR and translation can each introduce errors.
Core ScanLens OCR runs on the iPhone. A translation action may use the capabilities and network behavior of the translation service or system workflow involved, so check the current device and service settings for sensitive text.
ScanLens keeps the source page beside editable OCR text so you can correct important fields and translate with context.