YomiNinja 0.6 - Download
Added 2024-01-27 19:33:12 +0000 UTCI apologize for the delay, but unexpected personal issues required my immediate attention, leading to the postponement of the detailed v0.6 release post. The full post will be published this weekend. However, I'm excited to let you know that YomiNinja v0.6 is ready for use right now!
Here's how to achieve the same results as the previous sneak peek:
- Ensure Cloud Vision API credentials are set (Settings > OCR Engines > Google Cloud Vision). Watch this helpful video (1:14 - 3:24). If you prefer not to set up an account, use the "demo" mode as described in the UI.
- Lower the opacity of "Active BG" and "Inactive BG" colors of the bounding box.
- Lower the text color opacity.
- Make sure that "Individual character positioning" is enabled.
- Disable JPDB Reader from the extensions screen.
- Use the Cloud Vision OCR hotkey.
Please note that when using Cloud Vision, your images are sent directly to the Cloud Vision API without intermediaries, alleviating privacy concerns. Currently, individual character coordinates are only provided by Cloud Vision, and I'll be exploring possibilities to bring this level of precision to offline engines as well.
For a better experience with Yomitan, it's recommended to enable "Hide popup on cursor exit":
- Right-click the Yomitan icon.
- Click on "Options".
- Go to the "Scanning" section.
- Enable "Hide popup on cursor exit".
It is recommended to set the JPDB Reader API key or disable this extension if you do not need it.
Otherwise, it may start throwing errors on the right side of the screen, such as "invalid API key
while parsing...".
To open JPDB Reader settings set its API key:
- Make sure JPDB Reader is enabled.
- Click on its icon.
- Right-click its settings link.
- Click on "Open link in a new window".
- Paste your JPDB.io API key into the "API Token" field and click on "Save"
Thank you for your understanding, and I appreciate your patience. Feel free to start testing YomiNinja v0.6 now! Please leave a comment if you have any doubts or issues.
Thank you so much for your support!
Comments
Well, you are a wizard to make such a tool, thatβs for sure! And how to address a magician other than Sensei?. Count on us for the testing and feedback!
Yury Katkov
2024-01-30 18:43:16 +0000 UTCThanks for your detailed feedback! 1. Offline engines will be always the default option. I'll make the JPDBReader disabled by default, this probably caused many misunderstands as it's not so popular. 2. An automated approach with notifications could be good too. I'll explore the options. 3. I agree. Coming up with the right name for this feature was tricky. I imagined it including different functionalities down the line. Each region might have its own set of OCR configs and appearance settings, overriding the global settings. 4. Adding pop-ups has crossed my mind several times. However, I hesitated due to potential UI changes. Now that the UI is stabilizing, I'll consider implementing them soon. 5. If you have the time and wouldn't mind assisting, your help would be greatly appreciated. 6. This would be pretty cool. 7. Yes, similar to point 4, I suppose? 8. I'll look into it further. It might require additional configuration on the cloud console side.
Matheus M.
2024-01-30 15:56:40 +0000 UTCI don't think I can call myself a sensei π . Yes, mostly that's the plan. I'm planning to start a Discord for gathering feedback from early access members. This should help address these things more efficiently, I believe.
Matheus M.
2024-01-30 13:45:32 +0000 UTCOk, got it working, yep, it is much better in Chrono trigger steam!
Andrew
2024-01-30 04:22:54 +0000 UTCWhat's next, sensei? MangaOCR as another engine? UX, quality of life and manuals? Support for even more extensions? Integration with Textractor or Agent? Anything else?
Yury Katkov
2024-01-29 16:41:14 +0000 UTCI got everything working! I can say that it is epic, seriously, nothing is getting even close to the features we have in Yomininja. Some quality of life requests: 1. Google Vision API is hard to set up, but hey, δ»ζΉγγͺγ. However, it means that it should not be a default OCR engine. The default one should be the zero-setup Paddle. Similarly, everything that needs to be set up, needs to be disabled by default, like JPDB reader is not functional without the API key. 2. When the Vision API is set up, it would be nice to have a button that checks if the connection works and there's no problems with authentication or billing or anything. 3. I think that OCR Template is a complicated word. Maybe OCR Regions? Or maybe avoiding introducing the term altogether, saying smth like "Only scan part of the screen (much faster)". 4. The options are hard to understand. Not sure how it can be helped because they describe the nuanced behavior. I guess ideally they should have a tiny video fragment on hover, like it's done in modern versions of Photoshop and Illustrator. There videos as hard to make and direct. As a first step, the help button with the popup with the text will be enough. 5. Overall, a good video tutorial and textual help would have been great. If the UI is more or less fixed, I can occupy myself with it, I have a lot of experience writing help for Semantic MediaWiki 6. Maybe a step-by-step contextual tutorial would have been nice. 7. Overlay appearance set up can benefit from a picture that shows what's what. 8. For Cloud API, it's probably possible to ask it, how much money is already spent.
Yury Katkov
2024-01-29 16:18:36 +0000 UTCyes, that fixes it! Thanks so much!!! OMG, the tool is getting better and better and better!
Yury Katkov
2024-01-29 16:09:59 +0000 UTCThe error is from the JPDB Reader extension. It happens when the jpdb.io API key is empty. To fix this, open JPDB Reader settings (as explained in the post above) to set a key or disable it from the extensions tab. Also, be aware that this extension might not work correctly when "Individual character positioning" is enabled.
Matheus M.
2024-01-29 00:11:19 +0000 UTCI activated the account, added the JSON api, set up billing
Yury Katkov
2024-01-28 23:06:46 +0000 UTCI have interesting results with google ocr on windows 11. The text is recognized and everything works but it produces a lot of error messages that look like this: invalid API key while parsing γZ L B a t t l e 3 ε¦β¦γ
Yury Katkov
2024-01-28 23:06:24 +0000 UTCbeen using it today and it's been great. loving yomichan support. i'm trying to avoid using google ocr too much. fortunately paddleocr is surprisingly good by itself. seems to be like 95+ percent of the time paddleocr works. maybe 1 in 20 panels i need to use google ocr. haven't run into a panel that hasn't been correctly ocr'ed with this combination yet.
risho
2024-01-28 22:42:35 +0000 UTCI went back in to my account and saw I had an issue with my billing. I was using an older card that no longer worked but resolving that fixed my issue right away. Thank you for all your work you've done in bringing this to Yomininja.
Tenck
2024-01-28 16:19:46 +0000 UTCyeah
Andrew
2024-01-28 15:06:50 +0000 UTCThank you! Is this one? https://www.youtube.com/live/Z70VnZWY9qU?si=i8-zJjIaXd1IyyEH&t=2016
Matheus M.
2024-01-28 14:50:16 +0000 UTCOh, really glad to hear. Your app is a gem anyway, even with only Paddle :D. Really glad I have stumbled upon it! Steam version to be precise. I have no idea if they have a different font.
Andrew
2024-01-28 14:23:48 +0000 UTCOh, sorry for that. Is it the PS1 version of Chrono Trigger? The next OCR engine will be MangaOCR. It's offline and seems to work well with it.
Matheus M.
2024-01-28 14:07:33 +0000 UTCYeah, it seems to work much better with this font now! I'm impressed by the variety of fonts it can recognize. It even handles handwritten text, though the precision of the coordinates is not great.
Matheus M.
2024-01-28 13:09:19 +0000 UTCOh, this is probably billing. But I live in Mordor, I can't directly set up billing :( Maybe I'll get help from the guy who helps me pay for the patreon. But just in case, in some future version is it possible to have other offline OCR which won't require any kind of billing? Paddle is terrbile at some games, namely Chrono Trigger.
Andrew
2024-01-28 13:04:40 +0000 UTCYes, it's the same engine. However, it appears that Lens doesn't provide individual character coordinates.
Matheus M.
2024-01-28 12:24:22 +0000 UTCDid the issue occur while in "main" or "demo" mode? Note that demo credentials don't last very long, causing recognition to fail after a few requests. You can try loading new credentials again, but it's not guaranteed.
Matheus M.
2024-01-28 12:14:38 +0000 UTCDoes this issue only occur with the main API? Does the demo mode work? Please confirm these points: - Check if billing is enabled for the project in Google's console. - Ensure that the API mode is set to "Main". - Import the credentials using the JSON file.
Matheus M.
2024-01-28 12:04:03 +0000 UTCwill try the new ninja as soon as I have access to me windows machine, can't wait to feed it the battle screens of 13 sentinelsγit uses weird squarish font: https://youtu.be/PDkIegH5s8k?t=202
Yury Katkov
2024-01-28 10:30:47 +0000 UTCis it the engine that google lense uses??? If so, I guess it will be a next level of accuracy of recognition!
Yury Katkov
2024-01-28 10:25:53 +0000 UTCI came across an issue where the cloud vision wont recognize text. Do you know how I might be able to resolve that? If not it's fine cause I'm playing Phoenix Wright series and the default OCR works just fine with it. Edit: I just noticed that if I do the demo mode it'll recognize text, so I'm not entirely sure what's going on.
Tenck
2024-01-28 09:57:34 +0000 UTCI get a pop up "No text Recognized! please try again". And looking at the traffic in google's console, it has some requests. I tried both in game and test thing inside yomininja.
Andrew
2024-01-28 09:40:46 +0000 UTCgame changer
Seff Mcc
2024-01-28 01:05:13 +0000 UTC
