Help me to type japanese characters from image, please

[Scorp]

Image is this one:

Why I want that - I heard that PS2 version of Cartagra have 9 additional endings. But it uses custom text encoding like on image, so to read it first have to create that japanese character table..

 

Edited December 1, 2015 by Scorp

[Deep Blue]

Well you have all the hiraganas (including dakuten and handakuten double mark and dot), all the katanaka same thing with hiragana, and 23xx- more or less kanji (50x47 if im not wrong) that means you can use any table of kanji available on the internet and it will work.

Do you need to create it in the same order?

[Scorp]

I know that it have all hiragana etc  Just my eye is not sharp for all these glyphs, so I would not distinct 千 from 干, 口 from ロ or 末 from 未. 

I need to match that image with actual characters, so all symbols need to be typed in exact same order as on image. While, as you understand, it is in very random order, so without knowledge of glyphs I would not be able to match them.

[Deep Blue]

well that's a very annoying and hard work I will see if they are arranged in some specific order because the kana are not just randomly organized or just mixed there so I guess there is some logic behind the kanji too

[Scorp]

There is no logic as you assume. Means, kana and all before are static ones (maybe copy-paste from Shift-JIS), but this is not really important, as these I can type myself. So lets say I need all after line 10.

KID have sort of dunno, optimization, where they started from first text in the game『上野連続猟奇殺人事件――』 and after that all other kanji, which were not previously used and so on till the end of the script. So you would not be able to copy it from some existing character table, it is unique for that exact game.

Only way seems typing manually. Well, not really hard, but annoying and tedious, that.s for sure. Maybe if 3-4 guys would agree to do that, divide table between and type - would be not too much hassle?

[Deep Blue]

I tried using ocr from vnr but I can't make it work(not even with simple text), did you try that already? 

also is not really that simple for example I don't know all the kanji myself so finding the kanji you don't know is simple but it takes time, I tried finding a list but like you said there is not logic behind it so i didnt find anything 上書破損中絶

Edited December 1, 2015 by Deep Blue

[Scorp]

Yes, also I tried like 3 different OCR services, results are really bad, lot of errors. Only human can help, it seems.

 

[Bolverk]

Tried out http://capture2text.sourceforge.net/ it's a OCR software. It works but mainly just on roughly 1 - 5 kanji in one go. So a lot of tedious work to do. Sometimes it really doesn't want to work. So the usual OCR there..

Worked out the first 3 lines of txt...

上書
破損中絶対差今失敗再開始読込空溶量足以必要野連続猟奇殺人事件ー新聞見出目入息秋五逗子行紙面眺無別死体
切刻残虐手ロ異常性示言葉並四被害者半煽立記戦争終実感数年琉前検閲消違国民士気名担当刑有島一磨文司顔思浮

[Scorp]

Looks good! Will try it then. Just I still doubt that any OCR could see difference in 千干口ロ末未 and similar stuff.

..

[tymmur]

I know that it have all hiragana etc  Just my eye is not sharp for all these glyphs, so I would not distinct 千 from 干, 口 from ロ or 末 from 未.

I viewed it at 300% and the problem was suddenly not that serious anymore. I know it sounds simple, but I was surprised at how clear and readable it looks when scaled up. Much better than text usually looks when upscaled.

 

I think the only way to go is to use OCR, but assume it to make mistakes, which mean you have to verify every single kanji it provides, as even minor mistakes in this step will be horrible later on. Looking at an upscaled png would be helpful for this task.

 

Omnipage's OCR engine would rule for a task like this, but sadly it cost $150 It doesn't look like they have "free for non-commercial use" or student discount or anything, making it out of reach. Still I mention it because it might be available at school or work (it was to me once). Scanning this image in public shouldn't be a big deal. It's not even close to being eroge or anything like that. For the unlucky of us, there is no real choice other than to use the free OCR, even if it isn't as good.

[Scorp]

I know that it have all hiragana etc  Just my eye is not sharp for all these glyphs, so I would not distinct 千 from 干, 口 from ロ or 末 from 未.

I viewed it at 300% and the problem was suddenly not that serious anymore. I know it sounds simple, but I was surprised at how clear and readable it looks when scaled up. Much better than text usually looks when upscaled.

Mind to check it for me after I do OCR? I believe I would not see the difference between correct and wrong one  

[Kelebek1]

EDIT: Oh you want to map it.

[tymmur]

Mind to check it for me after I do OCR? I believe I would not see the difference between correct and wrong one  

Post it in this thread I can try. If it's hard to see, the best option might be to make multiple people try to verify. One person can miss something, but the risk that 3 people miss the same mistake is not that great. This mean getting others to verify is not a replacement for checking yourself.