The Scanco. Scanlation Guide - Vol. 1 Ch. 1.1 - 102: Why MTL is Never the Answer

[MangaDex]

 

[xyzzy]

THANK YOU

seriously, I’d take one officially translated chapter over a dozen bulk-uploaded MTL chapters

 

[Machmeathead]

This is very valid.

 

[CrimsonSix]

This is outdated. Even at the time it was released, OCR was a known and well-tested technology, so what is with "cant process images"? And you can always paste text into multiple translators to get alternative translations. Even in a single translator, many will give alternative translations if you trim down to a single kanji or phrase.

And of course, I'm commenting in 2026, when MTL has improved quite a lot compared to 2023.

 

[essensue]

This is outdated. ...

Everything you're saying is true, but even perfect OCR can't extract the context provided by the images that surround the text, like who's speaking, or the tone that's implied by their expression. Maybe some of that gap can be closed by an LLM if you do a second MTL pass where you both ask it to TL and provide the missing context by telling it the rest of the story. But most of the people who need to be told not to MTL aren't putting in that level of effort, much less what you describe of seeking other TL models or isolating specific words.

All translation is adaptation. Being a good translator and a good author are different skills. And faithfulness to the original work isn't necessarily its own virtue as long as the reader understands that's what they're signing up for. There's some really well done high-effort MTL stuff out there that I've really enjoyed. And if there's an older, completed series that hasn't been translated, I'd rather see it get an "adequate"-but-loving MTL than nothing at all.

BUT. If I had to pick "for" or "against" as blanket advice for everyone who's enthusiastically looking to get started scanlating, I'd absolutely say "don't MTL." MTL as a tool in most people's hands results in a pretty crap end product that literally makes the world a worse place by displacing and preventing a better effort by others. MTL is also kind of anathema to scanlation as a culture, if preserving traditional roles and encouraging the group aspect of the hobby matters to you at all. The outliers who would do a good job MTLing won't listen anyway.

 

[CrimsonSix]

Everything you're saying is true, but even perfect OCR can't extract the context provided by the images that surround the text, like who's speaking, or the tone that's implied by their expression. Maybe some of that gap can be closed by an LLM if you do a second MTL pass where you both ask it to TL and provide the missing context by telling it the rest of the story. But most of the people who need to be told not to MTL aren't putting in that level of effort, much less what you describe of seeking other TL models or isolating specific words.

All translation is adaptation. Being a good translator and a good author are different skills. And faithfulness to the original work isn't necessarily its own virtue as long as the reader understands that's what they're signing up for. There's some really well done high-effort MTL stuff out there that I've really enjoyed. And if there's an older, completed series that hasn't been translated, I'd rather see it get an "adequate"-but-loving MTL than nothing at all.

BUT. If I had to pick "for" or "against" as blanket advice for everyone who's enthusiastically looking to get started scanlating, I'd absolutely say "don't MTL." MTL as a tool in most people's hands results in a pretty crap end product that literally makes the world a worse place by displacing and preventing a better effort by others. MTL is also kind of anathema to scanlation as a culture, if preserving traditional roles and encouraging the group aspect of the hobby matters to you at all. The outliers who would do a good job MTLing won't listen anyway.

Thanks for taking the time to reply.

For people dedicated enough to use MTL properly, if they'd ignore an explicit instruction to not use MTL in scanlating, they won't turned off by the existence of low-quality MTL of manga they like. I say that with confidence because I was relying on MTL for years* before I started learning Japanese. And low-quality MTL isn't a bad thing necessarily. There are far more works that need translation than there are people with the skills and willingness to translate decently.

*: for translating games, not manga