Nekocrow

Okay, I've uploaded the new version with the "Background Eater" filter, that should work for outlined text. It's pretty late here, so hopefully I didn't break anything in the process ^__^ If anyone has text recognition problems with some unruly novel, let me know, and I'll cook up a filter. (I have checked the recognition on the top 30 or so novels at VNDB so far, and everything seems to be in order, but there's always outliers) -------- Also, working on a prettier version of the reader. It's still raw, but it should be ready in a day or two. ^__^

Oh, my bad, you are right -- seems OCR has trouble with chonky outlines like that. I wrote a de-outline filter, just need some more time to optimize it. It should be ready by tomorrow, I think.

Hmm, something ain't right, chief. It should work even without any config. Did you turn on vertical text recognition, maybe? It works for manga, but it does indeed garble horizontal text. (Now that I think about it, I should make it more clear in the FAQ) Could you send me a screenshot of the settings and the image selection box? I just tried testing it with the Higurashi intro and the recognition is stellar throughout, even without any configuration. The reader runs on a full Tesseract OCR engine, the very same engine that downloadable readers use, shouldn't be any problems there. Can someone test it, just to make sure? There could be a super sneaky teeny-tiny bug somewhere, you never know, haha ^__^ That is exactly what all other Visual Novel readers are as well - a translation box. Normally, VN readers only include the translation. The end-goal is to make them as unnoticeable as possible, so one can focus on the novel - and that is why we love them. ^__^ The dictionaries, sentence analysis, verb deconjugation, grammar explanation, and kanji breakdown are more for the learners of Japanese than the readers. The goal of this project is a bit different -- it is a learning tool for those who want to learn Japanese while having fun reading novels, and then, eventually, move on to reading novels in full Japanese, without relying on tools, and with no need to wait for the English version. A stepping stone, so to say. Sorry for the ramble, haha. I'm open to suggestions! If you would like extra features, let me know!

Edit Edit: Pretty Mode beta is now ready. Reworked dark design, bigger kanji, better Japanese/English fonts and (optional) text animations + automatic text flicker reduction when translating the page via browser. Edit: The recognition issue with outlined text, mentioned below, has been improved. Link YouTube 1-minute micro guide Heeey there, folks! So this here is a fully featured, totally free, visual novel and manga reader that I proudly cooked up. It is meant as a streamlined Japanese learning tool, but you can use it with the browser's in-built translation too. The main TL;DR features are: Works right from the browser, nothing to install, nothing to configure, ready to be used in 1 minute! Uses OCR (text recognition), works on every novel / image / video Works with browser extensions (translation / flashcards/ AI etc.) Comes with a full text analyzer engine, kanji info database, full dictionary, 200 000 example sentences, smart automatic verb inflector/definflector, auto kana-to-romaji converter, word audio examples, and many other goodies! Most importantly, if you have any feature requests, let me know! How to: Press the 'Auto' button and select the novel window Select the textbox area Translate the page using the browser or any translation extension Aaand that's basically it -- the app is configured to automatically scan and re-analyze the selected window every 3 seconds. The full FAQ is available on the website. Here's the short version of it: What is this? This is a fully featured reader for Japanese visual novels (and manga, and other stuff) with built-in OCR text-from-image recognition, full dictionary, kanji data, text analyzer, auto-furigana, verb deinflector, example sentences and more! Can I get full translation of the text / other functionality Since this reader is a standard web app, it supports all the usual browser extensions, including the ones that offer translation / flashcards and other goodies! If your translator works on text selection events, you can enable the auto-select option to automatically select the raw text output. Likewise, if your translator works on clipboard copy events, you can enable the auto-copy option to insta-copy the raw text output into the clipboard. Note: most browsers require a user gesture event (like a click on a page) each time before the auto-copy can be performed, therefore this option is best used with the manual mode. Note: the processed text output does not work well with extensions, so make sure to select the "Show Raw" option to display the raw text. How to improve character recognition? Only select the textbox area. Selecting too much empty space might lead to inaccuracies. Use the image invert, contrast and saturation options, until you get black text on a pure white background. Use opaque textbox if the novel you're reading has that option. Try using the "Best" recognition mode. It is slower, but (sometimes) better. Usually, the "Fast" option works well enough, though. Can I use it with fullscreen enabled? Certainly! Resize the browser window to a small size, put it in a corner, and use any app that enables it to stay always-on-top. Optionally, you may consider using an app that gives transparency to the browser window as well. Is my data safe? The part of the app that does image recognition is local -- none of your data / screenshots ever leave your PC, it works just like an offline app. Oh, and the reader is only a small part of the main website, which is a complete dictionary / kanji reference / sentence database and many other things (also totally free)! nekocrow.com/about Happy Reading!