THE SHORTEST RELIABLE PATH
Follow the state in order.
Choose the smallest useful region
Frame the text closely enough to exclude unrelated controls, translucent overlays, and decorative material.
Match the mode to the source
Use Single for a menu or scan, Continuous for subtitles, and Batch for several independent blocks.
Confirm language and reading order
Use an expected language hint for small or mixed text, then review the extracted order before translation.
Review the result in context
OCR and machine translation can both be wrong, so check names, numbers, prices, and safety-critical text.
OCR solves visible but inaccessible text
A PDF, image, video subtitle, remote desktop, game interface, or scanned document may display words without exposing a selectable text range. Lemvia uses the macOS screen-capture permission for the region you draw and performs text recognition on the Mac.
After recognition, Lemvia sends only the extracted words to the translation route you chose. The route can be Lemvia Cloud or a compatible API account you control. Local OCR does not make translation itself local when a network route is selected.
Accuracy depends on the captured evidence
Small fonts, compression, gradients, motion blur, vertical text, overlapping captions, and mixed scripts all reduce certainty. A smaller frame often helps more than increasing the number of translation retries because OCR quality determines the source text the translator receives.
For consequential material, compare the extracted source with the screen before relying on the translation. Lemvia is a productivity tool, not a substitute for professional review of legal, medical, financial, or safety-critical text.
Direct answers
When should I use Continuous OCR?
Use it for subtitles, dashboards, or another region whose contents change while its screen position stays stable.
Why did OCR miss small characters?
Tighten the frame, remove translucent overlays if possible, choose Accurate mode, and provide the expected language hint.
Is the screenshot stored in translation history?
Translation history stores the text result locally. Lemvia uses the framed image for OCR processing rather than treating the entire screen as a history item.
PRIMARY SOURCES