History
YomiNinja Changelog — Version History
All releases from the initial launch in September 2023 to the current stable version.
All versions
v0.9.3
Latest stable
August 4, 2024
- Fixed Google Cloud Vision integration breaking after token refresh
- Fixed custom keyboard shortcuts not persisting after app restart
- Fixed overlay positioning on macOS when display scaling is set above 100%
- Fixed overlay misalignment on macOS for windows on secondary monitors
v0.9.2
August 3, 2024
- Fixed Google Translate bubble rendering off-screen on multi-monitor setups
- Fixed overlay not appearing correctly when upgrading from v0.7.x without a clean install
v0.9.1
Major release
August 2, 2024
- VOICEVOX TTS integration — native Japanese text-to-speech across all platforms. Reads OCR-captured text aloud using VOICEVOX voice characters.
- MangaOCR on-demand mode — load the model only when capturing, reducing idle memory usage. Recommended for machines with less than 8GB RAM.
- GPU acceleration via CUDA and ROCm — NVIDIA (CUDA) and AMD (ROCm, Linux only) GPU support for faster OCR inference.
- Furigana threshold control — slider to filter small text (furigana) from OCR output, preventing it from being captured alongside kanji.
- Faster screenshot capture — improved screenshotting pipeline reduces capture latency on all platforms.
- Overlay hotkeys — new keyboard shortcuts for moving and resizing the overlay without using the settings panel.
- Wayland experimental support (Linux) — initial Wayland support added; not yet stable.
v0.8.3
Major release
Mid-2024
- Web browser overlay mode (PWA) — alternative overlay rendered inside a browser window for better extension compatibility in some configurations.
- Apple Vision vertical text — Apple Vision Framework on macOS now correctly handles tategumi (vertical Japanese text layout).
- Comic Text Detector + MangaOCR for macOS — combined pipeline for stylized font recognition on Mac.
- Text positioning options — control over where overlay text is anchored relative to the original text position.
- Image preprocessing settings — adjustable contrast and brightness applied before OCR.
v0.7.2
Major release
July 22, 2024
- MangaOCR engine (Windows + Linux) — specialized manga font OCR model with significantly higher accuracy on stylized text.
- macOS support (initial) — first macOS build. Intel and Apple Silicon packages.
- Apple Vision Framework OCR — native macOS OCR engine integration.
- Auto OCR — continuous detection mode that fires automatically on screen content changes.
- OCR Templates — saved capture regions for faster and more accurate targeted OCR.
- TTS basics — initial text-to-speech support (VOICEVOX full integration came later in v0.9.1).
v0.6.4
Early 2024
- Google Lens cloud OCR integration — no API key required
- Google Cloud Vision integration — highest-accuracy cloud OCR with API key
- Character-level text positioning — individual character bounding boxes enable precise hover-to-lookup
v0.5.1
Early 2024
- Chrome Extensions Manager — install additional browser extensions into YomiNinja's Chromium context
- OCR Templates (initial implementation)
- Google Translate bubble integration
v0.4.1
Early 2024
- Linux support — AppImage, .deb, .rpm, .pacman packages
- Popup dictionary improvements
v0.3.1
Late 2023
- Built-in popup dictionaries (Yomitan pre-install)
- 10ten Reader integration
- Extensions toolbar
v0.2.1
Late 2023
- WebSocket output — real-time text streaming for texthooker pages and Anki mining
- Image inversion options for dark-text-on-light-background games
v0.1 / initial
Origin
September 8, 2023
- Initial public release
- Windows-only OCR screen capture overlay
- PaddleOCR engine (local, offline)
- Transparent interactive text overlay
- Basic Yomitan integration
For the complete commit history and pre-release builds, see the GitHub Releases page. Patreon supporters get access to pre-release builds before they are publicly available.