- Dec 29, 2017
- 3,392
- 6,060
- Dec 29, 2017
- 3,392
- 6,060
It's one of those cases in which I'm honestly not sure if it's a pathetic attempt at DRM, a dev just "being quirky doe", or gross incompetence.
You can load the files into SLR, but it's pretty annoying and I'm not even sure if it works with the current version.
(I used the SLR v2.0 prototype)
So instead of trying to make that work on your end, here's a rough SLRMTL.
Replace the executable you get in the DEFAULT FOLDER with this one and it should be mostly translated.
You must be registered to see the links
Edit: Feel free to share it around if someone else wants it, but executables are scary so expect weird posts.
If you actually want to make the translation process work on your end tell me, it's not actually hard, but a bit annoying to setup.
Last edited:
This is great thanks! Was just suprised as normally you can tell if SLR will work by the engine, like SLRPG games don't but as this was an RPGM MZ it was just more wanting to know why. Thanks again
- Dec 29, 2017
- 3,392
- 6,060
It's actually MV. The www folder was packed into the executable, that was packed into the original executable with enigma.
Why did the dev do that? I don't fucking know.
Hi Shisaye,
for this game when I do the translation it gives the syntax error 'Invalid or unexpected token'. The file in issue is the plugins, but not sure what needs to be fixed?
Game is:
for this game when I do the translation it gives the syntax error 'Invalid or unexpected token'. The file in issue is the plugins, but not sure what needs to be fixed?
Game is:
You must be registered to see the links
LinkR34
Member
- Dec 9, 2017
- 295
- 261
I have 64GB ram and a nvidia 16gb gpu and plenty of nvme and ssd storage left, whats the module?
I have translated a few game using slr, both using SugoiV4 , and local LLMs
Last edited:
- Dec 29, 2017
- 3,392
- 6,060
To load picture based text into SLR Translator as if it's normal text and draw new text on the picture during export.
Basically you feed it PNG, it will use PaddleOCR to determine boxes around text and do an initial OCR, it then crops those boxes and sends them to a vision enabled AI for secondary OCR (if the confidence level was below 0.95). Then it sends the entire pictures to the LLM and asks it to evaluate the collected OCR in context of the picture and to pick the best ones.
It then loads those results into a normal SLR Translator project, and you can run whatever translation you want on them.
Once you tell it to export it will use the box data to use blur to remove the original text and then it will use pillow to draw new text on top of it, adjusting the font size so the new text never leaves its box boundaries, while making an effort to have boxes with the same height also use the same font size.
Last edited:
Do you manually add the pngs? as in decrypt all the image files, find the ones with relevant text and feed them to this module.
then re encrypt them and overwrite the originals?
I was just doing the lazy thing and using the google translate ocr image translator to get in the ballpark.
- Dec 29, 2017
- 3,392
- 6,060
Well... Letting the AI scan every single picture of a game would be incredibly slow.
Someone only using CPU could need ~2 minutes per picture and usually of like 400 pictures only 30 actually need translation.
And if the user is roughly picking them out anyway automating the decryption process also doesn't seem like a high priority.
In the current version you just select a folder and it will grab all unencrypted PNG files from that folder, and it will also output unencrypted files.
Well the main point of my system would be to be fully offline, and to be able to manually check/fix bad translations before drawing them on.
Also most free online services don't support multiple files at the same time.
Well when its ready soon i have a AI slop game i was going to translate, 23 images need to be translated.
---edit---
also has anyone tried
Sugoi LLM 14B Ultra its meant to be much better
Last edited:
- Dec 29, 2017
- 3,392
- 6,060
If it would be ready I would just release it and wouldn't need testers.
I'll pm you a prototype I guess.
- Dec 29, 2017
- 3,392
- 6,060
You need to be a bit more specific. When does it give a syntax error?
Using game v24.08.27.
I clicked through a bunch of menus and fucked the green eyes knight girl and I did not see an error with a default translation.
I did see some preview images failing to appear in the recruit menu, but no actual crashes.
When i copied the data and js files over and started the game, it gave the syntax error. After moving them one at a time it was when i moved the js folder that the error came, hence i assumed it was that. The game can load if i only put the translated data folder, but some parts are not translated
- Dec 29, 2017
- 3,392
- 6,060
That is not the case on my end. I can use the translated plugins.js just fine.
Remember to only export actually changed files.
Will try the translation again, if possible could u send your translation file to see the diff please?
Tried again twice, used the chinese version as well, both giving syntax error. Followed guide excatly, only thing different is I don't change the spelling mistakes if flagged, but most games normally work with that
Last edited:
- Dec 29, 2017
- 3,392
- 6,060
I've released SLR Translator v2.0.
Besides a bunch of other changes, Introducing the SEP offline picture translation system.
It will load pictures into a SLR project as if they are text files, letting you translate the text however you want. When exporting it will automatically mask the original text and draw the new one in its place making absolutely sure the text never covers more general area than the original text.
It will also attempt to logically group the text so that font size stays consistent.
You can also override the automated system if you want to specify a different font, font size, or font color.
During tests using Qwen2.5-VL-32B-Instruct-abliterated as assistant LLM and translating with DeepSeek DSLR, SEP consistently outperformed the free picture translation services of both Google and Yandex. (For horizontal Japanese.)
Current Limitations:
Only works on PNG without encryption.
Only works on horizontal text. (Slightly tilted text technically works, but it will draw the new text horizontal and probably in a very small font size.)
Besides a bunch of other changes, Introducing the SEP offline picture translation system.
It will load pictures into a SLR project as if they are text files, letting you translate the text however you want. When exporting it will automatically mask the original text and draw the new one in its place making absolutely sure the text never covers more general area than the original text.
It will also attempt to logically group the text so that font size stays consistent.
You can also override the automated system if you want to specify a different font, font size, or font color.
During tests using Qwen2.5-VL-32B-Instruct-abliterated as assistant LLM and translating with DeepSeek DSLR, SEP consistently outperformed the free picture translation services of both Google and Yandex. (For horizontal Japanese.)
Current Limitations:
Only works on PNG without encryption.
Only works on horizontal text. (Slightly tilted text technically works, but it will draw the new text horizontal and probably in a very small font size.)
LinkR34
Member
- Dec 9, 2017
- 295
- 261
- Dec 29, 2017
- 3,392
- 6,060
If you want to include yellow cells in the batch translation you can just click on the yellow button under blacklist, to remove it from the blacklist.
But beware that there "might" also be cells in there that can't be translated, so if you run into errors you will have to manually review them.
The most common reason something is tagged yellow is because it's a generic "string literal" entry. So essentially you have no clue which plugin is actually using it and so it could be completely fine, or make everything crashy mc sploodey.