- Aug 28, 2018
- 402
- 556
- Oct 21, 2018
- 222
- 513
we were talking about types of MTL.
And "worst case" is such a low bar to clear, i can gather from the screenshots that it's at least readable.
- Jun 16, 2018
- 305
- 810
Making grammatical sense doesn't make it an accurate translation. But I thought DeepL already has near perfect grammar, so what exactly is MAT? Did the translator "rewrite" it or is MAT simply fixing the bits of grammar?
- May 25, 2017
- 207
- 255
Sadly, voodoo only works on believers.please tag the person who MTLed this, we need to give them credit.
So I could voodoo him
- Jun 23, 2020
- 92
- 60
- Jan 13, 2020
- 1,328
- 9,073
- Aug 10, 2020
- 416
- 1,274
- Dec 4, 2018
- 316
- 190
Okay, am I insane, or is this from the same guy who keeps making NTR RPGs under different groups? まぜ, ワンドワンド, 三つ雨, ネイビス, and now S2, right? Are there others I'm missing? Does this artist have a twitter account or something, or is he seriously pretending to be different people?
Fair point, then it's basically a MTL without grammar errors. Unfortunately that doesn't make it much better than MTL, so that's a bummer. But hey, thanks still. DeepL omits too much information for me to like it, but I'll have to make do
- Jun 8, 2017
- 268
- 653
From what I've seen, all of JustXuX's translations fix the basic bugs like overflow and broken sfx like moans and whatnot, then they differ in that "low quality" tier is mostly just fixing DeepL's grammar, while the mid and high tiers spend more time ensuring the translation itself is correct. So, low is a much quicker translation that has a few parts where the translation is wrong, but it's still easy to follow along with the story and dialogue, while medium and high take much longer, but ensure that most or all of the translation is spot-on, respectively. He's stopped doing high translations since, for the most part, an H game doesn't need absolutely perfect dialogue to be properly enjoyed, and isn't worth the extra time and effort - it's like making sure you're getting 4k resolution and 144 fps while playing Stardew Valley.
- Jan 13, 2020
- 1,328
- 9,073
No, it's basically me sitting and feeding DeepL and Sugoi MTLs, one dialogue at a time, then checking if the translation is right (because I know basic Japanese and still learning). And if both of them suck I edit the whole dialogue.
If some parts need edit I edit them to shorten the dialogue, whenever it's possible.
Translating as good as I can core dialogues by following the context and story.
Sometimes making it sound more like native English.
You see, even feeding MTLs the right way can improve their output translation. From experience, I already know how and what to feed to MTLs. Japanese differs a lot from English, sometimes the writer can use punctuations, sometimes not. Because it's optional in Japanese, although welcomed. So I know when the next line is a continuation of the previous sentence or just another sentence. If the *heart symbol* is used to end the sentence or just to emphasize a part of text.
I may mistranslate some stuff, but for important dialogues I go extra mile to ensure they are max accurate.
There could be grammatical errors, because I still manually edit text.
I have many little tricks and hints that I use while translating, that I didn't mention in the Translation Types Google Doc.
Most of those tricks I use automatically, without thinking. Anyway...
My goal is to translate as fast as possible while retaining the game's playability. It's a compromise translation.
More games available and decent enough translation to play and enjoy them. Plus I fix bugs too, if I or someone encounters them.
Also, this game is my last LQMAT project, because I found a way to translate to MQMAT standard with minimal time penalty, when compared to LQMAT. But it's worth it.
Her name reminds me of shiri-tori.. a game where the last sound of a word is used to decide the beginning make a new word.
If a new word an an old word are too similar, the player looses.
I have some questions as someone who doesn't know Japanese, but still makes edited MTLs using Sugoi and DeepL. I can't really truly gauge my quality, but i usually atleast aim for coherence and they're usually slow to complete.
1. How often to these two tools fail on such level that if you only see English output you can't tell that something went wrong. (I mainly mean a completely grammatically correct sentence that's still pretty seriously wrong)
2. What methods to feed text to MTL the right way do you employ? I only know that DeepL absolutely can't handle line breaks so feeding DeepL as single line usually drastically improves results it spits out. Other times if original japanese employed commas I would paste both parts of sentence separately. Are there any other methods? And how implementable are they for someone with no real knowledge of Japanese.
3. Have you ever tried TL++s RedSugoi? Although it requires for Vanilla Sugoi to be on, when it's told to batch translate it often enough produces different results from regular Sugoi. It might have some level of memory cache when translating, but I'm not sure.
4. Onomatopoeia in Japanese is always a pain to handle. Does stuff like
You must be registered to see the links
generally get things right when used for reference? Are there some other good sources for that?
- Jan 13, 2020
- 1,328
- 9,073
Sorry I have no time and willingness to answer all your questions here, by writing.
Join my public Discord and ping me, we'll schedule a time we can both talk via voice.
I do so many things that I have to make several tutorials to include most of it.
- Mar 20, 2020
- 2,194
- 1,987
I know the pain, I have my own translation projects on the go.
Keep up the good work!
Side note, any tentacle pregnancy in this one?