Tool RPGM SLR Translator - Offline JP to EN Translation for RPG Maker VX, VX Ace, MV, and MZ

[taodentumienque]

If you don't really care about optimization you can use any of the official Sugoi Toolkit versions (except 9-9.4) and run Sugoi in offline mode. (The port will already be the correct one, you do not need to change anything.)

Or you can read this post:
https://f95zone.to/threads/slr-tran...aker-vx-vx-ace-mv-and-mz.208393/post-13751939

I do not want to give any specific tutorial or recommendation, because it's highly dependent on your hardware and personal preference.

I use laptop with 12900h and 3070ti, 16gb ram,which version should i use for high performance

 

[Entai2965]

I use laptop with 12900h and 3070ti, 16gb ram,which version should i use for high performance

For more detailed notes on the Sugoi-Repackage, see the

You must be registered to see the links

, expand the small summary below, and also read this.

 

[taodentumienque]

hi,i try sugoi repackage- CUDA but instead of using nvidia gpu,it use igpu,how to fix it?

 

[Entai2965]

hi,i try sugoi repackage- CUDA but instead of using nvidia gpu,it use igpu,how to fix it?

Edit2: This thread is to discuss SLR Translator. To discuss the repackage, please use the repackage

You must be registered to see the links

or the Sugoi thread.

 

[Shisaye]

hi,i try sugoi repackage- CUDA but instead of using nvidia gpu,it use igpu,how to fix it? View attachment 4050473

I'm sorry, but I can't provide support for Sugoi itself, or Sugoi repacks, because I did not make them and do not know how they function.

Please ask in the dedicated Sugoi thread on F95, the thread of the repacks on fuwa, or in the offical Sugoi discord instead.
(But if you ask in the official discord, please do not mention SLR to avoid potential drama.)

 

[Shisaye]

I've released v1.100.

Improved word wrapping by making it consider punctuation more often when splitting.
Meaning if a text is

A long text about a big ol' futa cock, that needs to be wrapped.

and you were to wrap to 50 characters, it will no longer wrap it like this:

A long text about a big ol' futa cock, that needs
to be wrapped.

It will instead wrap it to

A long text about a big ol' futa cock,
that needs to be wrapped.

That is still a bit experimental and could cause it to make more lines than necessary.
I would need some feedback on the current implementation in case there are issues.
I'm debating whether or not it should use "," for splitting or if it should only consider "." "!" "?".

It will also now preserve indents of the original dialogue if it's the same across all lines or the same across all lines except the first.
Meaning it will preserve starting spaces like this:

Text
  Text Text Text
  Text Text Text
  Text Text Text

even when they are not encased in a speech bubble.

Edit: I forgot to apply a minimum distance from the start based on the wrap amount specified by the user. That "could" cause some less ideal splits wasting a bunch of space. I will fix that in v1.101.

 

[Shisaye]

I've released 1.101
Only notable change is that wrapping now has a minimum character count according to user input. Meaning for example, if you use "size", which wraps to ~70 characters (40 monospace characters), it will not accept a break on punctuation if it would make the original line shorter than ~45 characters. (And it keeps that ratio regardless what you put in.)

 

[Shisaye]

I've released 1.102.
I realized that just adding a fixed ratio based on the wrap goal input is not actually "logical" placement. Because if someone were to manually translate, they wouldn't just break after a certain character count if it means the next line only has 1 word.
Instead they would be much more likely to split it evenly according to an existing pause like a dot, even if it makes the first line shorter than the second.

It will now instead be a variable ratio based on the difference of the wrapping goal set by the user and the original line.
In the sense that it will now always try to consider punctuation if it does not cause the next line to be longer than the wrap goal.
(Because then it would have to break again, creating an unnecessary line.)

 

[taodentumienque]

I've released 1.102.
I realized that just adding a fixed ratio based on the wrap goal input is not actually "logical" placement. Because if someone were to manually translate, they wouldn't just break after a certain character count if it means the next line only has 1 word.
Instead they would be much more likely to split it evenly according to an existing pause like a dot, even if it makes the first line shorter than the second.

It will now instead be a variable ratio based on the difference of the wrapping goal set by the user and the original line.
In the sense that it will now always try to consider punctuation if it does not cause the next line to be longer than the wrap goal.
(Because then it would have to break again, creating an unnecessary line.)

Do you know any tool support translating Visual novel ?
There's Textractor but it seem stop updating
Other option is OCR but not good enough

 

[linranni]

after i use the translation and open some menu, it appears like this and the item in that menu doest appear. any idea?

 

[linranni]

after i use the translation and open some menu, it appears like this and the item in that menu doest appear. any idea?

ah nvm i found it, i think it translates something that shouldn't be translated in the js file but I don't know which one I just delete something that doesn't fit and there are some things that it doesn't translate but it changes dots and stuff

 

[Shisaye]

Can you be a bit more specific?
What's the DLsite code of the game?

 

[Entai2965]

Do you know any tool support translating Visual novel ?
There's Textractor but it seem stop updating
Other option is OCR but not good enough

This is not really the place to talk about VNs considering the current scope of SLR, but consider

You must be registered to see the links

, SugoiToolkit, and

You must be registered to see the links

. Garbro is for extracting VN assets and determining the game engine you are working with. Once you know the game engine, search for tools on Github, Codeberg, and other software repositories associated with that engine.

Also try searching for the engine on

You must be registered to see the links

, like

You must be registered to see the links

. If the engine exists, it will be found in a release

You must be registered to see the links

. If there are no English releases for that game engine, then there might not be any tools to work with it. Most game engines have at least one release, but if there are no tools to work with it, then Luna Hook/OCR software would be the only option to translate such games.

 

[linranni]

Can you be a bit more specific?
What's the DLsite code of the game?

You must be registered to see the links

this game. slr translated some text in the js file but didn't really translate it and just deleted some dots and made the game error, sorry i can't give some examples because i already deleted the old traslation save file

 

[Xand]

Do you know any tool support translating Visual novel ?
There's Textractor but it seem stop updating
Other option is OCR but not good enough

You can try MTool, and then do a translation using this method using Sugoi. Fair warning - it is a bit roundabout way of translating non RPGM stuff.
It's not as fast as SLR and wont work on Unity, but can yield passable results with this method.

For RPGM Games - in my limited knowledge - SLR is by far the most efficient tool out there for machine translations.

 

[Shisaye]

I've released v1.103.
There was some unusual "if" statement constellations, that weren't properly escaped and could have caused crashes for games with quirky devs.

 

[Shisaye]

I've been working and testing stuff with the re-translation using the Levi model and the prospect of not always using the dictionary to get more varied results, but I had to come to the conclusion that my algorithm for detecting bad translations just is not good enough.

It works well enough to be a good indicator, but when doing runs without the dictionary it kept missing bad translations that would have been fixed with the dictionary and the end result was not actually better.
So basically that idea is shelved for now, until I find a much more reliable way of detecting bad translations.

I've also noticed that while it can miss bad translations there's also still some false alerts. Sure you can just use regex to get rid of those errors, but it's pretty annoying.


One Idea to address that would be to instead of having it create an actual annoying "UNUSUALLYSHORT" error, that it tags those cells in a fancy new color.
That would have the benefit that you could just tell the batch translator to re-translate that color with a different translator and then run the fix cells button again to see if that fixed the issues.
Basically instead of there being automatic re-translations with Levi, just make Levi another option in the translator list, and maybe even add support for more options.

Some feedback would be nice.

 

[Xand]

Would like to test that out, the unusually short error really is annoying. And is a pain to fix manually if there are hundreds of the same.

Do I need to have a copy of Levi installed locally to run the new version of SLR?

 

[Shisaye]

Do I need to have a copy of Levi installed locally to run the new version of SLR?

I haven't activated that in the current release, yet, I was trying to get opinions on that first.

If you would want to use Levi then you would need to have a server running for it on port 14370, while V4 would still be 14366, but after the current idea that would not be required at all because there would be no automatic re-translation.

It would just tag those cells Purple and you could use the strict whitelist in the batch translator to use whatever on them. (You could also try Red Google instead for example.)

So would you say that's an improvement over the UNUSUALLYSHORT error stuff?


Another question would be how to handle the POTENTIALFILENAME error because that one is also pretty annoying and constantly has false alerts, but missing an actual filename would crash stuff...

 

[Xand]

honestly, as a one-off, it would be a better solution, however, I would think in the long run having a multiple color coded tags for multiple error codes like UNUSUALLYSHORT/ POTENTIALFILENAME or LOSTBACKSLASH would be as infuriating as having the number of errors.

But, in the short term - yes. It is a better idea.

edit -
For POTENTIALFILENAME or LOSTBACKSLASH, leave it be. It needs to be checked anyway.