- Jul 24, 2017
- 2,536
- 28,611
It can but you would need to go into the code and modify the regexes and instances of 'English' and 'Japanese' to whatever language you want
would that be hard to do? are the regexes all over the code or are we talking a single code file? I assume its the modules files?
Would I just swap out your japanese regex
r'[一-龠]+|[ぁ-ゔ]+|[ァ-ヴー]+'
to chinese(simplified) as an example:
r'[\u4E00-\u9FFF]+'
or does it not support unicode range?
Last edited:
- Jul 24, 2017
- 2,536
- 28,611
It does you can use unicode range. It shouldn't be too hard you can Ctrl+f
r' to find them all. modules are the only ones that need changing (And maybe the prompt.txt). Each file is based on an engine and there are really only 2 atm, mv and ace.Sorry for asking, this is from the mv module:
Line 214 and multiple others
does all these need to be edited too?
if I read the regex right it should be something like this?
r'^[^\u4E00-\u9FFF\<\>【】]+'
Line 429
Line 435
How common are these? albeit i only dabbled a bit with translation++ before, I don't think I ever encountered an <SG . . .> event.
What's the english equivalent for the <特徴 . . .>?
Line 1148
Is this game specific?
Line 214 and multiple others
Code:
# Need to remove outside code
jaString = re.sub(r'^[^一-龠ぁ-ゔァ-ヴー\<\>【】]+', '', jaString)if I read the regex right it should be something like this?
r'^[^\u4E00-\u9FFF\<\>【】]+'
Line 429
Code:
tokens += translateNote(name, r'<特徴1:([^>]*)>')
Code:
tokens += translateNote(name, r'<SG説明:([^>]*)>')What's the english equivalent for the <特徴 . . .>?
Line 1148
Code:
context = 'Character Context: 莉音 == Rio | Female, 結衣 == Yui | Female, 美雪 == Miyuki | Female, あかり == Akari | Female, カガミ == Kagami | Female, ミズキ == Mizuki | Female, スズカ == Suzuka | Female, シズク == Shizuku | Female, 太郎 == Taro | Male'- Jul 24, 2017
- 2,536
- 28,611
Yup, it just matches any non-japanese characters and removes it. Basically `一-龠ぁ-ゔァ-ヴー` is just matching Japanese characters so you can probably replace that with`\u4E00-\u9FFF\` and it will probably work.
Game specific, doubt any non-japanese devs use those. They are usually extra text for skills, items, weapons, etc. Not all games use this but I've run into a few.
Yes, that's for Winter Memories. It gives GPT the names and genders of the characters to improve the consistency of the translation.
- Jul 24, 2017
- 2,536
- 28,611
Prompt example in the repo is outdated. I just do it in the module now.
Updated it just now.
Last edited:
I have encountered an error at the
When I select python file it returns the error:
Code:
Exception has occurred: SyntaxError
invalid syntax (main.py, line 24)
File "F:\Gamez\VN\_Tools - translation\DazedMTLTool-master\start.py", line 1, in <module> from modules.main import main SyntaxError: invalid syntax (main.py, line 24)- Jul 24, 2017
- 2,536
- 28,611
- Jul 24, 2017
- 2,536
- 28,611
I haven't but Claude 2 is more expensive than GPT4 and I hear more sensitive with the filter.
Nope. Its more for people who want control over each step of the MTL process and plan to make edits to it.
- Jul 17, 2020
- 312
- 262
- Nov 6, 2018
- 446
- 323
GPT-3.5-turbo-16k should be cheaper, right? It seems sufficient for translation tasks.
- Jul 24, 2017
- 2,536
- 28,611
We need some brain fuel to use this thing, but the only thing I'm getting is brain farts.
OP Gurio Umino seems to have no problem using this. Probably launching it as we speak, on his X68000.
- Nov 5, 2018
- 611
- 898
Unity games often are translated by extracting text to a .txt file.
Would you be able to adapt the tool for this use-case?
If you do decide to do this, here is the formatting that the text files normally use (reipatcher for unity)
\n indicates a line break.
It's finicky about spaces/line breaks so you have to keep the formatting the same.
Also, how do you get around the content policies for GPT3.5? I've tried to use it and I always get interrupted as soon as it detects anything sexual in the translation. Or is that only specific to the web interface?
Would you be able to adapt the tool for this use-case?
If you do decide to do this, here is the formatting that the text files normally use (reipatcher for unity)
Code:
Yes、調査開始=I've arrived at the location.
彼女はウユハート・ヴァーグマイン。\n汚染された地域の調査や生物兵器（怪物）を駆除する組織に属している。=She is Uyu Heart Vagmine.\nShe belongs to an organization that specializes in investigating contaminated areas and exterminating biological weapons (monsters).
彼女は現在、単独での任務を行っており\n新兵器の実地試験と潜伏する怪物の討伐。\nそして、現地の簡易調査を行っている。=Currently, she's undertaking a solo mission,\ninvolving field testing a new weapon and subduing lurking monsters.\nAdditionally, she's conducting a brief on-site investigation.It's finicky about spaces/line breaks so you have to keep the formatting the same.
Also, how do you get around the content policies for GPT3.5? I've tried to use it and I always get interrupted as soon as it detects anything sexual in the translation. Or is that only specific to the web interface?
Last edited:
- Jul 24, 2017
- 2,536
- 28,611
Yup I could I just haven't done a Unity game yet since it usually requires a complete text file first. That means a requester would have to provide me with such a file or I would have to go through every single piece of text I could find in the game first. If you have a game in mind feel free to let me know and I can pick it up and at the same time create a script for Unity.
Web Interface is way stricter than API for content policy.
Last edited: