I managed to procure a Japanese copy of the game. I mostly wanted it for comparison, and so I could work out the text encoding. With a little time, I should be able to dump a delimited file of any large block of untranslated text, that we can work from.
Code: Select all
Yeah, it's supposed to have music. But out of curiosity, couldn't you just compare it to the PAL version to see how they fixed it?
Edit: Urgh. The character encoding appears to be completely custom, and it's context-specific, apparently. A particular shift encoding gets 火の紋章 in a rune name, but it's 火の followed by different Kanji in the description field. The untranslated stuff is mostly dialog, so one set of shifts, but working out their encoding will take some time.
Edit 2: I think I've exhausted my supply of CJK and Japanese text-encodings. I don't see one that remotely matches up to this, so I have to assume it's something they worked out in-house, based on the characters used in their script. I can still decode better than half of the Japanese text, since I know enough to be able to sort out the Latin characters, and the Hiragana and Katakana. Fully decoding is going to take some time. Especially if I have to do it by myself. I used a Kanji dictionary to look up the first ten or so characters. Only took about half an hour.
Edit 3: Proving once again that I often get my best ideas when I get up and do something else for a while. I cleaned up the image a bit, converted it to gray-scale, carved it up into a few pieces, and removed some of the garbage. Then I ran the image through a Japanese OCR program. It wasn't 100% effective, but with a few more slices and passes, I should get well over 90% of the character set without having to resort to looking them up by radicals or trying to get an identification tool to recognize them one-by-one.
Edit 4: Hastily assembled web page of the conversion results for about 5/16ths of the Kanji and character set. I already have the simple stuff mapped to encoding, but there are a couple of characters I can't recognize. A couple of them I suspect aren't actual characters, and are just menu type symbols like the "Lv." and "Exp." symbols. The "suspect" characters are yellow. The one I outright could not determine is red. The columns are rows are taken from the image I attached.
Character Matrix
Eventually I want to describe all of this in a document that ties it to the encoding, but for now, I just want to get the character set itself completed.
Edit 5: I finished the character set, mapped it to the encoding, and translated the first few lines of dialogue in the game as a test. So far, I've been able to locate the text strings in the game files reliably, but I haven't been able to tie them to any kind of tree or control data. It may just be hard-coded in the event code. In any case, I should be able to start pulling text later in the week.