Tool RPGM DazedMTLTool - A tool that provides quality MTL translations using ChatGPT

[cpammy]

Rpg maker VX is not supported, right?

 

[Ashino8155]

Does anyone know how to get a structured output from LM studio?
I've been getting "Error code: 400 - {'error': "'response_format.type' must be 'json_schema' or 'text'"}"

 

[revyfan]

which makes me believe this MIGHT be the local model OpenAI were teasing a few months ago.

Nope, I was completely wrong. OpenAI released the two local models today (Called GPT OSS 120b/20b) and both are complete and utter jokes of LLMs. Its writing capabilities is abysmal, the translations are dog shit and worse of all, both are very much censored. So even if its capabilities were amazing, it would still be absolute horse shit.

 

[jaden_yuki]

Nope, I was completely wrong. OpenAI released the two local models today (Called GPT OSS 120b/20b) and both are complete and utter jokes of LLMs. Its writing capabilities is abysmal, the translations are dog shit and worse of all, both are very much censored. So even if its capabilities were amazing, it would still be absolute horse shit.

The superior quality of the Horizon model goes beyond translation, it's very more likely to be GPT-5

 

[Hallokun]

Does anyone know how to get a structured output from LM studio?
I've been getting "Error code: 400 - {'error': "'response_format.type' must be 'json_schema' or 'text'"}"

Change your translation.py in DazedMTl/util with this one, worked for me

 

[jaden_yuki]

The superior quality of the Horizon model goes beyond translation, it's very more likely to be GPT-5

seems like I was right. the horizon beta ended right before the release of GPT-5, so we don't have free API access to GPT-5 anymore . but the GPT-5-mini is insanely cheap, looking forward to see translation tests with it

 

[jaden_yuki]

g4f isn't working here anymore, at least with gpt 5 and 4.1

 

[timanuk]

dazedanon Wow, you’ve translated so many hentai games. Does the GPT API not censor sexual content?

 

[jaden_yuki]

DeepSeek v3.1 is out!

 

[blackhedgehog]

Any guide how to use this with deepseek?

 

[jaden_yuki]

Any guide how to use this with deepseek?

depends on the provider you're going to use

 

[kaossss]

Any guide how to use this with deepseek?

How to use deepseek api?

in the ".env" file set:

api="https://api.deepseek.com"
key="{your deepseek API key}"
model="deepseek-chat"

 

[flashi99]

I currently try to translate a game that has Quest notes stored in the items.json file under "note".
Translating this file does not translate these notes.

Is there currently a way to translate those automatically or do I have to manually translate those?

I attached the items.json, that I have translated using DazedMTLTool, where the notes are not translated, see line 8521 for the first quest information. (Added .txt-extension so upload is possible)

 

[dazedanon]

I currently try to translate a game that has Quest notes stored in the items.json file under "note".
Translating this file does not translate these notes.

Is there currently a way to translate those automatically or do I have to manually translate those?

I attached the items.json, that I have translated using DazedMTLTool, where the notes are not translated, see line 8521 for the first quest information. (Added .txt-extension so upload is possible)

You have to modify the code in the mzmz module and add a regex that can cath the text in those notes so it can translate them.

 

[flashi99]

You have to modify the code in the mzmz module and add a regex that can cath the text in those notes so it can translate them.

I'll look into it but it will probably to high of a skill for me to do.
Thank you.
EDIT: Found the issue!

I had an out of date mvmz module, there was a mistake in the following if-clause:

if "Client:" in match_text or "Client :"

which was fixed !1! day after (28.08.) I pulled changes :

if "Client:" in match_text or "Client :" in match_text:

Which once again should reinforce in me: Pull before asking.
Sorry for the trouble.

Also I wanted to ask:
In the section containing CODE356, you replace spaces with underscores (lines 2115 - 2116).
Would it not be better to use a Non-Breaking space:

translatedText = translatedText.replace(" ", "\u00a0")
translatedText = translatedText.replace("__", "\u00a0")

Or is there a specific reason that this would cause issues?

 

[dazedanon]

I'll look into it but it will probably to high of a skill for me to do.
Thank you.
EDIT: Found the issue!



Also I wanted to ask:
In the section containing CODE356, you replace spaces with underscores (lines 2115 - 2116).
Would it not be better to use a Non-Breaking space:

translatedText = translatedText.replace(" ", "\u00a0")
translatedText = translatedText.replace("__", "\u00a0")

Or is there a specific reason that this would cause issues?

Non breaking spaces can be inconsistent between games either by not working at all or having larger spacing than normal. I like underscores because the plugins can be easily edited to replace them with normal spaces.


Also they tend to be easier to edit if needed since I don't need to hunt down the non breaking space which isn't on a keyboard.

 

[flashi99]

Non breaking spaces can be inconsistent between games either by not working at all or having larger spacing than normal. I like underscores because the plugins can be easily edited to replace them with normal spaces.

View attachment 5207100
Also they tend to be easier to edit if needed since I don't need to hunt down the non breaking space which isn't on a keyboard.

Thanks for the explanation.
Will try it next time.

 

[blackhedgehog]

How I fix this part?

 

[dazedanon]

How I fix this part?

View attachment 5208014

You must be registered to see the links

 

[HikariSomnium]

Tencent recently released new translation models: Hunyuan-MT-7B and Hunyuan-MT-Chimera. According to this information -

You must be registered to see the links

- they compete with powerful LLMs and even outperform some.

I haven't looked into it deeply yet, but it seems they are "compatible" with the OpenAI API.

I wanted to test them, but text-generation-webui refuses to load the models using any loaders. I don't have time to deal with it right now, so I might just wait for a GGUF version to check them out.

If the models run fast enough, high-quality machine translation might soon be available to everyone, and we won't have to spend $20-100 on translating each game.