for a few years now i wanted to build a naomi game menu which would load the netboot file from the network. no pi no extra pc no nothing else besides a headless system that would initialize the menu game only. after doing a few tests and looking at the netboot binaries i gave up for now. there is alot of ground work before this goal of mine
- Thread starter twistedsymphony
- Start date
So I decided to poke around the hex of the net bootable typingofthedead.bin. I haven't made any real progress but here's what I've found.
There's a file structure of some kind in here. right around 000F1800 there's a bunch of .bin files listed that match pretty closely to the bin files I saw on the Dreamcast version, the word file is absent from this list however.
Searching for English Text in ASCII I can find just about everything that shows up in the service menu but I cannot find any of the in-game text at all. writing down a few words that appear in game (both the romanji and katakana) I can't find the romanji in the hex anywhere.
katakana falls outside the ASCII protocol so it requires more 2 bytes per character instead of 1. the Unicode hex values for Japanese katakana can be found here: http://www.unicode.org/charts/PDF/U30A0.pdf
Manually translating some of the in-game katakana into the unicode hex and searching for that in the file turns up nothing as well. It seems that the first byte for most (all?) Japanese characters is a 30 so if there is japanese text in unicode you'd expect to see something like
30 D0 30 EB 30 FC 30 E9 where every other byte is a 30. (by the way that's バルハラ or BARUHARA which is a gameplay word in the NAOMI version) I can't even find the text that appears in the game menu, though it's possible that text is actually an image file. I think it'd be rather hopeless to convert any image files without being able to extract the content to a normal file structure.
Does anyone know if there is another text format that is used for Japanese outside of unicode? I suppose it's possible that the words aren't even stored in a way that we can easily find them in hex either.
There's a file structure of some kind in here. right around 000F1800 there's a bunch of .bin files listed that match pretty closely to the bin files I saw on the Dreamcast version, the word file is absent from this list however.
Searching for English Text in ASCII I can find just about everything that shows up in the service menu but I cannot find any of the in-game text at all. writing down a few words that appear in game (both the romanji and katakana) I can't find the romanji in the hex anywhere.
katakana falls outside the ASCII protocol so it requires more 2 bytes per character instead of 1. the Unicode hex values for Japanese katakana can be found here: http://www.unicode.org/charts/PDF/U30A0.pdf
Manually translating some of the in-game katakana into the unicode hex and searching for that in the file turns up nothing as well. It seems that the first byte for most (all?) Japanese characters is a 30 so if there is japanese text in unicode you'd expect to see something like
30 D0 30 EB 30 FC 30 E9 where every other byte is a 30. (by the way that's バルハラ or BARUHARA which is a gameplay word in the NAOMI version) I can't even find the text that appears in the game menu, though it's possible that text is actually an image file. I think it'd be rather hopeless to convert any image files without being able to extract the content to a normal file structure.
Does anyone know if there is another text format that is used for Japanese outside of unicode? I suppose it's possible that the words aren't even stored in a way that we can easily find them in hex either.
Last edited:
So I decided to take a look at Lupin the 3rd The Typing, it's a GD-ROM game so we can use GD-ROM explorer right? Maybe it would provide some insights into Typing of the Dead... Maybe we could translate that by just throwing all the words into a translator (I mean they're just single words they don't have to make sense inside the context of a phrase). wrong 
converted the mame .chd to a BIN and TOC file that GD-ROM explorer could open
guess what... inside the GD-ROM is pretty much just one big 1.2GB BIN file 
so yeah exactly in the same place we are with Typing of the Dead. 
converted the mame .chd to a BIN and TOC file that GD-ROM explorer could open
Code:
chdman extractcd -i gds-0021a.chd -o lupetype_gd.toc -ob lupetype_gd.binI've been working with the netboot version this whole time I've just been occasionally calling it cf as it's essentially the same image before matching it to the card. I'll stop doing that to avoid confusion.
No real progress on this since the failure to crack open the .bin file... however I did come across something else.
When I was searching the hex before I was using the UTF-8 hex characters to try to find the japanese words used in the game and I couldn't find them which made be believe that they were compressed, encrypted or encoded somehow.
While I was trying to translate a control panel on a different game i came across this Wiki entry for the KA kana: https://en.wikipedia.org/wiki/Ka_(kana)
...and i noticed that the Unicode uses different hex codes than UTF-8, not only that there is another stadard called "Shift JIS" (Japanese Industrial Standard https://en.wikipedia.org/wiki/Shift_JIS) which is also completely different.
So I may have to revisit the basic hex searching to see if maybe TOTD used Unicode or Shift JIS to make the dictionary file instead of UTF-8.
When I was searching the hex before I was using the UTF-8 hex characters to try to find the japanese words used in the game and I couldn't find them which made be believe that they were compressed, encrypted or encoded somehow.
While I was trying to translate a control panel on a different game i came across this Wiki entry for the KA kana: https://en.wikipedia.org/wiki/Ka_(kana)
...and i noticed that the Unicode uses different hex codes than UTF-8, not only that there is another stadard called "Shift JIS" (Japanese Industrial Standard https://en.wikipedia.org/wiki/Shift_JIS) which is also completely different.
So I may have to revisit the basic hex searching to see if maybe TOTD used Unicode or Shift JIS to make the dictionary file instead of UTF-8.
that's a great idea... unfortunately don't have nor have I been able to find a ripped copy of the JP Dreamcast version.
twistedsymphony, check your PMs. 
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
Awesome work!twistedsymphony, check your PMs.
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
Have you done this kind of thing before? if we're doing string replacement in the hex I'm curious how we deal with situations where the word we're replacing is shorter or longer.
I would suspect there some flag on each word that identifies when it could be used. When you play the same level multiple times it uses different words but they words do get more difficult as the game progresses so I'd guess that there is a "level 1" word bank and a "level 2" word bank that it pulls from. It's possible that the US word banks are simply larger than the ones used in the NAOMI release. It wouldn't be the first time Sega gave the US release a bit more content than the Japanese release having had a few extra months to work on the game during localization.
There are some other odd-ball sections of the game too, sometimes during gameplay you can activate a "dictionary" bonus which means that the next string of words are similar in spelling, or similar in topic.
There is also at least one boss fight where they present you with a question and give you 4 options for answers. so there must be some way that they flag the question and answer text as related and identify which is the correct answer and which is not.
being able to identify this "meta data" I think will be important to converting it successfully.
Thanks!Awesome work!Have you done this kind of thing before? if we're doing string replacement in the hex I'm curious how we deal with situations where the word we're replacing is shorter or longer.twistedsymphony, check your PMs.
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
I would suspect there some flag on each word that identifies when it could be used. When you play the same level multiple times it uses different words but they words do get more difficult as the game progresses so I'd guess that there is a "level 1" word bank and a "level 2" word bank that it pulls from. It's possible that the US word banks are simply larger than the ones used in the NAOMI release. It wouldn't be the first time Sega gave the US release a bit more content than the Japanese release having had a few extra months to work on the game during localization.
There are some other odd-ball sections of the game too, sometimes during gameplay you can activate a "dictionary" bonus which means that the next string of words are similar in spelling, or similar in topic.
There is also at least one boss fight where they present you with a question and give you 4 options for answers. so there must be some way that they flag the question and answer text as related and identify which is the correct answer and which is not.
being able to identify this "meta data" I think will be important to converting it successfully.
Nope, not really. There is a lot of extra space / padding in the Naomi BIN, so there is definitely more than enough room to get all the English text translations in, though it may require moving some stuff around and possible re-addressing. Also, I'm not sure whether it would be easy to add in the extra stuff from the English dictionary; you might be limited to the number of entries in the existing Naomi release depending on how the game reads from the dictionary and what it expects there.
Those multiple choice questions seem to be towards the end of the dictionary, though I'm not certain how they are structured.
I think there are two sets of questions.Thanks!Awesome work!Have you done this kind of thing before? if we're doing string replacement in the hex I'm curious how we deal with situations where the word we're replacing is shorter or longer.twistedsymphony, check your PMs.
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
I would suspect there some flag on each word that identifies when it could be used. When you play the same level multiple times it uses different words but they words do get more difficult as the game progresses so I'd guess that there is a "level 1" word bank and a "level 2" word bank that it pulls from. It's possible that the US word banks are simply larger than the ones used in the NAOMI release. It wouldn't be the first time Sega gave the US release a bit more content than the Japanese release having had a few extra months to work on the game during localization.
There are some other odd-ball sections of the game too, sometimes during gameplay you can activate a "dictionary" bonus which means that the next string of words are similar in spelling, or similar in topic.
There is also at least one boss fight where they present you with a question and give you 4 options for answers. so there must be some way that they flag the question and answer text as related and identify which is the correct answer and which is not.
being able to identify this "meta data" I think will be important to converting it successfully.
Nope, not really. There is a lot of extra space / padding in the Naomi BIN, so there is definitely more than enough room to get all the English text translations in, though it may require moving some stuff around and possible re-addressing. Also, I'm not sure whether it would be easy to add in the extra stuff from the English dictionary; you might be limited to the number of entries in the existing Naomi release depending on how the game reads from the dictionary and what it expects there.
Those multiple choice questions seem to be towards the end of the dictionary, though I'm not certain how they are structured.
the 3rd boss ("Tower") is a hydra who does it, and then the final boss is a combination of all previous bosses and one of his attacks is similar to the 3rd boss and prompts a question. which is why you might see it at the end of the file.
I think the rest of the bosses just have timing windows attached to them or otherwise obscure the typing window so you only get brief opportunities to type, which is most likely not controlled by the WORD file.
Yeah something like that. I seem to remember at least several multiple choice questions the last time I played through it on the Naomi in Japanese.I think there are two sets of questions.Thanks!Nope, not really. There is a lot of extra space / padding in the Naomi BIN, so there is definitely more than enough room to get all the English text translations in, though it may require moving some stuff around and possible re-addressing. Also, I'm not sure whether it would be easy to add in the extra stuff from the English dictionary; you might be limited to the number of entries in the existing Naomi release depending on how the game reads from the dictionary and what it expects there.Awesome work!Have you done this kind of thing before? if we're doing string replacement in the hex I'm curious how we deal with situations where the word we're replacing is shorter or longer.I would suspect there some flag on each word that identifies when it could be used. When you play the same level multiple times it uses different words but they words do get more difficult as the game progresses so I'd guess that there is a "level 1" word bank and a "level 2" word bank that it pulls from. It's possible that the US word banks are simply larger than the ones used in the NAOMI release. It wouldn't be the first time Sega gave the US release a bit more content than the Japanese release having had a few extra months to work on the game during localization.twistedsymphony, check your PMs.
Compare 0A5A0168 in Naomi bin to 000E88C0 in JP WORD.AFS for example for a string match. It looks like there is also some kind of data before each string, which may just be from the AFS file structure, or may be used for formatting text on screen.
Table of contents locations (there is a list of BIN files, which seem to make up the AFS file and may refer to certain dialogue lines, or may be random)
0A40B000 - Naomi BIN
001A1000 - JP WORD.AFS
001D3000 - US WORD.AFS
It looks like there are more strings in the English version than the JP version, so a straight replacement of the full WORD.AFS file probably won't work. The BIN files are named differently across the JP/ENG versions too. In fact, the dictionary area of the Naomi binary does not start with "AFS" like the WORD.AFS files do.
I guess a starting point would be to replace strings from the US WORD.AFS directly into the Naomi BIN once you've determined the location of a desired string by cross referencing the JP WORD.AFS.
Most text is prepended by 'FFFFFFFF0000000000000000', so that should help you in determining the locations of strings.
There are some other odd-ball sections of the game too, sometimes during gameplay you can activate a "dictionary" bonus which means that the next string of words are similar in spelling, or similar in topic.
There is also at least one boss fight where they present you with a question and give you 4 options for answers. so there must be some way that they flag the question and answer text as related and identify which is the correct answer and which is not.
being able to identify this "meta data" I think will be important to converting it successfully.
Those multiple choice questions seem to be towards the end of the dictionary, though I'm not certain how they are structured.
the 3rd boss ("Tower") is a hydra who does it, and then the final boss is a combination of all previous bosses and one of his attacks is similar to the 3rd boss and prompts a question. which is why you might see it at the end of the file.
I think the rest of the bosses just have timing windows attached to them or otherwise obscure the typing window so you only get brief opportunities to type, which is most likely not controlled by the WORD file.
I did this and it's actually playable. This looks good for a Naomi translation being viable.' said:
Okay so the Naomi binary definitely doesn't use an embedded AFS file structure for its dictionary. Instead, it has the BIN files from WORD.AFS extracted and inserted directly into the filesystem for the Naomi image. See offset F298C in the Naomi ROM for where the dictionary files are listed. E.g. S000L010.bin.
So basically, a tool would need to be written to extract the files from the Naomi ROM, as well as a tool to extract the BIN files from the US WORD.AFS, then you would replace the dictionary binaries, then a tool would need to be written to repack the Naomi ROM with the modified files.
Alternatively, it should be possible to manually edit the strings in the Naomi ROM by hand, but that would take ages and a lot of trial and error I'd imagine.
So basically, a tool would need to be written to extract the files from the Naomi ROM, as well as a tool to extract the BIN files from the US WORD.AFS, then you would replace the dictionary binaries, then a tool would need to be written to repack the Naomi ROM with the modified files.
Alternatively, it should be possible to manually edit the strings in the Naomi ROM by hand, but that would take ages and a lot of trial and error I'd imagine.
back before this site when I first toyed around with this idea I spent some time over on Romhacking.net it's basically a site dedicated to this kind of thing, albeit for old 8 and 16-bit games. I had asked for some help over there and came up empty but there are a lot of tools and information about manual translation.
I spent a bit of time poking around the JP WORD file and I still don't know what encoding is being used. It's probably Shit JIS as everything I've been reading points to that being the closest thing to a standard in Japan but I haven't quite figured out how it works. that table on wikipeida is woefully incomplete.
I found a more complete table here: http://www.rikai.com/library/kanjitables/kanji_codes.sjis.shtml though this still doesn't fully explain how it works.
-------------
From what I've read of other translations people talk about translation tables
I think if we can identify the regions for each word bin file in the NAOMI bin and then identify the location and length of each word, we might be able to cross reference that with each bin file inside the US WORD file and swap out words of similar length. or maybe to start just identify that there are the same number of words in each file and swap the first word with the first word and the second word with the second word and so on.
I spent a bit of time poking around the JP WORD file and I still don't know what encoding is being used. It's probably Shit JIS as everything I've been reading points to that being the closest thing to a standard in Japan but I haven't quite figured out how it works. that table on wikipeida is woefully incomplete.
I found a more complete table here: http://www.rikai.com/library/kanjitables/kanji_codes.sjis.shtml though this still doesn't fully explain how it works.
-------------
From what I've read of other translations people talk about translation tables
I think if we can identify the regions for each word bin file in the NAOMI bin and then identify the location and length of each word, we might be able to cross reference that with each bin file inside the US WORD file and swap out words of similar length. or maybe to start just identify that there are the same number of words in each file and swap the first word with the first word and the second word with the second word and so on.
It means nothing really but when I used to use japanese on japanese chat forums in the late 90's the standard used was shift-jis.
Considering this game was made around that time, it would make sense that the developers would use a common standard for the time.
I found this tool called WindHex which looks promising: http://www.romhacking.net/utilities/291/
I found this tool called WindHex which looks promising: http://www.romhacking.net/utilities/291/
