What OCR does
OCR turns text inside a screen region, screenshot, game, video subtitle, image, slide, or PDF into selectable text. Transifyr then sends that text into the translation engine you selected.
OCR Engine priority
Transifyr supports three OCR engine modes. The best default is Auto because it balances speed and reliability.
Auto: tries Native OCR first. If the result is empty, noisy, or unreliable, it falls back to Tesseract.js.
Native OCR: uses Windows OCR only. Choose it when you want maximum speed on clean UI text or white backgrounds.
Tesseract.js: skips Native OCR and uses Tesseract only. Choose it when Native OCR struggles with colorful video backgrounds or stylized subtitles.
OCR language follows Screenshot Source Language
There is no separate OCR Language setting. OCR uses the Screenshot Source Language as its recognition hint, so the OCR and translation source stay aligned.
- Screenshot Source = English → Tesseract uses eng.
- Screenshot Source = Japanese → Tesseract uses jpn+eng.
- Screenshot Source = Korean → Tesseract uses kor+eng.
- Screenshot Source = Chinese → Tesseract uses chi_sim+eng.
- Screenshot Source = Auto → Tesseract uses a lightweight default, currently jpn+eng.
What does jpn+eng mean?
jpn+eng means Tesseract loads Japanese and English recognition packs together. This helps with Japanese subtitles that include English names, UI labels, numbers, or game terms.
Recommended setup
Set OCR Engine to Auto.
Set Screenshot Source Language to the language shown in the image or subtitle.
Set Target Language to the language you want to read.
Use Screenshot OCR or Live OCR to translate the selected screen region.
When OCR looks inaccurate
- Use a smaller region that contains only the subtitle or text area.
- Set Screenshot Source Language manually instead of Auto.
- Try Tesseract.js mode for colorful, animated, or noisy video backgrounds.
- Try Native OCR mode for clean app UI, documents, and white backgrounds.
- Increase the selected region slightly if letters are cut off.