MakeACopy 2.4.0 (Release Candidate)
- Arabic and Persian OCR support: Full recognition with AUTO page segmentation and disabled Latin whitelist for proper Arabic script detection.
- New dictionaries: Added Arabic and Persian frequency dictionaries for improved OCR post-processing.
- 21 languages now supported: Extended dictionary download to cover all supported OCR languages.
- OCR quality improvements: Dynamic Best/Fast model detection, enhanced adaptive preprocessing with Wolf and NICK thresholds, and dictionary-based error correction.
- PDF text layer improvements: Dynamic line tolerance based on median word height, proximity-based word clustering, and improved line sorting by average Y-position.
- BiDi support: RTL detection via Unicode directionality analysis, correct right-to-left sorting for Arabic and Persian scripts.
- Noto Naskh Arabic font: Added for proper Arabic and Persian script rendering in PDFs.
- Fallback dictionary loading: Supports both compressed (.txt.gz) and uncompressed (.txt) dictionaries.
- PDF import: Import PDF files via the image picker button. Single-page PDFs are rendered directly; multi-page PDFs show a page selection dialog with thumbnail previews.
- Bug fix: Fixed crash when closing PDF resources (IllegalStateException: Document already closed).
- Bug fix: Fixed OpenCV assertion error in classic B/W export mode when adaptive threshold was skipped.
- Bug fix: Improved readability of B/W PDFs at low DPI settings. The despeckle algorithm now scales its aggressiveness based on target DPI, preserving fine text details at lower resolutions.
- Bug fix: Fixed issue where old OCR results were displayed after switching from PDF import to camera capture when cropping was skipped.
Notes
- This is a Release Candidate build; feedback on camera stability and preview behavior is especially welcome.