YomiNinja v0.8.1
Added 2024-09-11 18:04:18 +0000 UTCEnhancements:
The overlay can now be viewed in web browsers, allowing the use of any browser extension with YomiNinja (experimental).
Added a PWA (Progressive Web App) mode, enabling the web version of the overlay to function similarly to the standard overlay.
Added support for vertical text recognition in Apple Vision OCR engine.
Added an option to disable GPU Compositing, potentially fixing the issue where the overlay turns black.
Added a search field to the capture source selection screen.
Updated the built-in Yomitan version to 24.9.2.0
Bug Fixes:
Fixed an issue where the overlay did not work with fullscreen applications on macOS.
Fixed misplaced recognized text on macOS.
Fixed an issue where Migaku's login button was unresponsive.
Fixed an issue where Yomitan's action popup buttons were unresponsive.
Fixed an issue where OCR results would disappear when enabling/disabling an extension.
Fixed an issue where the recognized text was misplaced on Windows after using the PrintScreen key.
Fixed a misleading "No text recognized" message when multiple concurrent OCR requests were made.
Fixed an issue where the app failed to launch when the PaddleOCR service could not start.
Fixed an issue where the TTS (text-to-speech) audio would unexpectedly pause between lines, causing interruptions in the playback flow.
Known Issues:
Before updating the app, you need to manually delete the app permissions (accessibility and screen recording) in the macOS privacy settings.
The app might fail to fully load on the first launch. However, it should start without issues on subsequent launches.
The ARM64 version might show the message "YomiNinja is damaged and can't be opened". You can work around this issue by following step 4 of this article.
Problems with multi-monitor systems on Linux and macOS.
YomiNinja Web App
With the web app, you can now view the overlay in a browser, enabling any browser extension to work with YomiNinja. This is an experimental feature with some limitations, but it will be helpful for users who need to use extensions that are not currently supported.
How to View the Overlay in a Browser
1. Click the "View Overlay in Your Browser" button, and the overlay will load in your default browser.
2. If you're using a Chromium-based browser, click the "Install" button; otherwise, click the "Pop Out" button.
3. Ensure the overlay window is in the same state as your capture source (fullscreen, maximized, or floating).
Limitations
Web browsers do not support windows with transparent backgrounds.
Click-through may not work as expected with certain applications.
The web app can only be installed on Chromium-based browsers.
Notes:
After installing the web app, you can launch it like a regular app, but make sure YomiNinja is already running.
The overlay will automatically move and resize only if it's in floating mode.
If an extension doesn't work, disable strict mode.
Click on the /\ button to toggle the title bar (PWA mode).
Future Updates
Version 0.9 will be released in a few days, introducing image filters, initial support for Wayland (Linux), a faster screen capture mode, VOICEVOX TTS voices, and more.
I'm also working on improving support for built-in extensions, with an experimental implementation in v0.9 to address Yomitan screenshot issues. As this implementation is refined, extensions like Migaku may be supported in future versions. Soon, a poll will also be posted to update the priorities for upcoming features.
Any feedback is appreciated, so please let me know your suggestions!
Thank you so much for your patience and support!
Getting Started With YomiNinja
Comments
Hey mate just wanted to report that on Windows 11 the program doesn't really work at all with Genshin Impact without running it as administrator, once you do that though the OCR takes likes 7-10s instead of the usual 2-3s which is quite painful for every line. There's also issues with Sword Art Online Re:Hollow Fragment where the program is not giving me a mouse to hover over the text and read it despite having the setting on. More dangerously if you use the window capture and choose a portion of the screen, the game starts flickering a white light on and off extremely rapidly. Thanks for reading, I am excited to join as a paying patron when these things get ironed out as it's the only things I really would use it for!
squirrel_weak12
2024-10-03 08:53:14 +0000 UTCyes! works fine
Yury Katkov
2024-09-24 23:18:43 +0000 UTCHey Yury, do you use it on an ARM Mac?
Mikail Freitas
2024-09-15 23:21:59 +0000 UTCThanks you for this update, also the browser web app seems really useful, I'm really excited for 9.0, especially for TTS. there is just one problem I have with the zhongwen extension, that is the latest 6.1.0 doesn't work when I install it in yomininja, although the older 5.15.0 one works fine.
kaine nier
2024-09-14 20:12:27 +0000 UTCThank you very much! I'm eager for this.
Sonto
2024-09-14 19:22:58 +0000 UTCMatt, you are unstoppable!!! Thanks for this update! I will try it now, but I suspect that the browser support is huge. Ability to run any extension opens up endless possibilities. I use Yominija both on Mac and on my Windows machine and I'm discovering new use cases every week. For example, I figured that I can now read pdf and epub books 10 times more easily. Watching videos with the slides is another big one: I use yominija to scan the content of the slides for my AWS course.
Yury Katkov
2024-09-12 10:20:53 +0000 UTCThere are two options to speed up MangaOCR. One is GPU acceleration, and the other is implementing a "Selective Recognition Mode", where MangaOCR will only process the areas you interact with (hovered) instead of processing everything before showing the results. The second option might be useful for users without a compatible GPU. I'll include both options in the next poll to decide which should be prioritized. I'm planning to implement at least one of these options in v0.9.1 or v0.10.
Matheus M.
2024-09-12 01:34:07 +0000 UTCWill the speed of MangaOCR be fixed any time soon? it's kinda unusable with how slow it is.
Sonto
2024-09-11 22:13:09 +0000 UTCElectron provides methods to detect the display where the mouse cursor is located, but these methods seem to be broken on macOS and Linux. It also doesn't return the correct display coordinates. I’ve recently set up macOS on a machine with multiple displays to investigate this further.
Matheus M.
2024-09-11 18:29:05 +0000 UTCi see you mention the multi monitor issues on macos and linux. do you know what the problem is or how to fix it? i've noticed that issue too
risho
2024-09-11 18:08:41 +0000 UTC
