ROM Hack [Release] GodHand - Japanese Game Translator

[cearp]

from just reading the title i thought this might be some ntr plugin that performs ocr and tries to translate the game... crazy right
(there would be lag i guess, we could have the function linked to a hotkey and only translated when we need it)

anyway thanks for sharing the tool you made

 

[pikatsu]

@Insight2k Sorry to ask again but can you add ocr ability from something like https://code.google.com/archive/p/nhocr/download

--------------------- MERGED ---------------------------

or http://capture2text.sourceforge.net/

 

[Insight2k]

@Insight2k Sorry to ask again but can you add ocr ability from something like https://code.google.com/archive/p/nhocr/download

--------------------- MERGED ---------------------------

or http://capture2text.sourceforge.net/

Hm, to do what exactly? I mean, I am reading files already and getting the output in the results table. Do you want extra support for images? Isn't that what Kuriimu does?

 

[pikatsu]

Kuriimu can only open game archives and extract images from them. I meant to ocr the extracted png or jpeg images from Kuriimu in order to take or translate the text from them or ocr text from images taken by ntr. But to do this you have to merge your tool with the best free japanese ocr source

 

[ruineka]

@Insight2k It looks like you have made a pretty decent tool here. I am unable to test it right now, but I do have a question. Does this currently support custom tables that the user can import to replace the characters? If not would it be possible to implement custom tables to extend it's functionality?

For example:

Default
E38182 =あ

Custom
6F = あ

South Cape is right by the way. Kanji gives detail to the text, without it and going strictly by romaji you can get words or phrases that have multiple meanings. I always give the あめ example to explain this to people. If you were to convert this to romaji you get ame which can mean candy or rain. Kanji gives us the indicator of which is which.

I do, however, think people who are looking for a rough translation of menus and other small things will find this very useful. So there is definitely a place for this tool.

 

[pikatsu]

Dragon Quest games use this table http://www.mediafire.com/file/cw9fuup3xt3y75l/dqm_3ds_table.txt

 

[Insight2k]

@pikatsu It is possible without getting into detail. I haven't had the time to take a detailed look, but I can do that and evaluate what is possible or not.

@ruineka I have to check this. I will come back to you.

-Update-
New version has been released.

 

[ruineka]

Just so you know the Google Translate API changes pretty frequently. If you ever find your application crashing in the future it is most likely be do to the api changes causing it to malfunction. I saw this happen recently with another project.

 

[Insight2k]

Just so you know the Google Translate API changes pretty frequently. If you ever find your application crashing in the future it is most likely be do to the api changes causing it to malfunction. I saw this happen recently with another project.

Hopefully, I will not have the same issue ^^
I checked your idea for custom encoding tables. It should be possible. I have to think about creating an algorithm for that.
That way, you could handle the output via a table.

@pikatsu
I have checked Capture2Text. I used two test images and one screenshot from an ingame screen. the results were not that good imo. It had difficulties to get the right japanese words.
Executing via command-line did also not work as expected.
Can you provide me with up to five test images from your desired game? I would like to run additionally tests.

 

[pikatsu]

I have read somewhere that this is one of the best OCR engines https://github.com/tesseract-ocr/tesseract

 

[Insight2k]

I have read somewhere that this is one of the best OCR engines https://github.com/tesseract-ocr/tesseract

Tested two and it does look good, I guess. I give it a try ^^

 

[Insight2k]

I have read somewhere that this is one of the best OCR engines https://github.com/tesseract-ocr/tesseract

I have released a new version with the basic use of the Capture2Text OCR engine.

 

[pikatsu]

OCR function gives me only strange characters like ³Æ’žÎ. Have i to install something like japanese font or anything.

 

[Insight2k]

OCR function gives me only strange characters like ³Æ’žÎ. Have i to install something like japanese font or anything.

Yes, true. You need to set your "region" in system settings under management "Language for non-compatible unicode programs" to japan.
Hm thats not cool, but I don't mind. I have a workaround already in mind.

@pikatsu I made a fix. Please try again.

After I have done the encoding tables, I am thinking about creating the possibility for creating translation patches if there aren't existing any (good?) tools. I haven't made up my mind yet.

 

[pikatsu]

I tried the last version it recognises kana almost 100%. Kanji results were nearly perfect when i tried images with clear text but had some error with a kind of text (it was not because of the tool just the image was difficult to scan). Very good start

 

[Insight2k]

@Insight2k It looks like you have made a pretty decent tool here. I am unable to test it right now, but I do have a question. Does this currently support custom tables that the user can import to replace the characters? If not would it be possible to implement custom tables to extend it's functionality?

For example:

Default
E38182 =あ

Custom
6F = あ

South Cape is right by the way. Kanji gives detail to the text, without it and going strictly by romaji you can get words or phrases that have multiple meanings. I always give the あめ example to explain this to people. If you were to convert this to romaji you get ame which can mean candy or rain. Kanji gives us the indicator of which is which.

I do, however, think people who are looking for a rough translation of menus and other small things will find this very useful. So there is definitely a place for this tool.

I have basic functionality for custom tables. Could you give me some additional examples? I used the mentioned table for dq.

Those are my results:
HEX: 835F837E815B
Original: ダミー
Custom: たeたさせa
HEX: E38182
Original: 縺・
Custom: せそ

---------------------
Edit
---------------------

Updated initial post with the current ReadMe of the project.
New version released with custom encoder support. It is an alpha version.
Please test and give me feedback if it works.

 

[Insight2k]

New version is online. See initial post for changes.

Thanks to @ruineka for the feedback for the custom encoder tables.

Next will be providing a store for the changes and a release management function for deploying patches to the community.

 

[pikatsu]

I have basic functionality for custom tables. Could you give me some additional examples? I used the mentioned table for dq.

Those are my results:
HEX: 835F837E815B
Original: ダミー
Custom: たeたさせa
HEX: E38182
Original: 縺・
Custom: せそ

---------------------
Edit
---------------------

Updated initial post with the current ReadMe of the project.
New version released with custom encoder support. It is an alpha version.
Please test and give me feedback if it works.

Dragon Quest uses
4 byte = numTables
4 byte * numTables = numEntries;
4 Byte * numEntries[n] = offsets;
There are no sizes, a text always ends with "E31B" ("\e")

 

[Insight2k]

The custom table, dq, which you have provided where only up to a hexlength of 6. that's what I have implented.
I fixed in the current version the \e and \n

edit:
now i get what you have said.
not really imo. ruineka provided me a sample.bin where i could test and look some things up and it did work out quite good. a text end was declared by the \e value. not every line ends automatically with \e or \n. as long it is declared via the values, everything is fine.

 

[Insight2k]

Final release so far.
Check initial post and have fun.

Projects and Patching are finally done.
If there are further translators I can embed via API, I will do so.