Insert Trumpets Here

Monday, April 25, 2011

"Scanbooru"

My previous statement about how I was going to approach Oriko Magica was kinda vague. I'll expand on it in this post.

Problems:
  • Oriko Magica is a 150-page spoiler bomb. Excuse me, it's a series of 150-page spoiler bombs, one per month. Since the Madoka manga stuck to the anime, it wasn't very interesting (at least to me) besides the explicit detail in the Charlotte scene.

    Due to its explosive potential, this is a manga that needs to be translated yesterday, and I don't usually say that. I'm sure that a lot of other people are dying to work on it as well, considering that a few anons (including DKThias) churned out Madoka vol 2 in a few days.

  • Madoka vol 2 was an example of a scanlation effort on /a/ done right. Nevertheless, a few days ago, I saw it go wrong. A couple of anons were working on the Lotte's Toy manga, much to the chagrin of Facepalm Scans. However, someone else pointed out that the editor was putting in the text in the wrong bubbles. The anons stopped, presumably embarrassed.

    I can't really blame them, since chan-style image boards aren't really made for scanlation. During the little experience I had of scanlating on /a/ (while using someone else's translation), I found that it's easy to lose track, even with scripts like backwash.
Proposed Solution:
  • As I've said, someone suggested by e-mail that I should develop some sort of soft translation software. I probably misconstrued what he meant, but in any case, I dismissed the idea at the time.

  • I occasionally visited the Madoka general to see what's up, and maybe help translate a few things, like those few magazine scans that talked about film theory in Madoka a few weeks ago. I would frequently see the anon who is of purportedly French origin (named "Some french anon") editing Madoka pics and doujins. One time, I noticed that he was asking for sound effects translations, so I pointed him to the motherlode of sfx translations. He said something to the effect of "well, that's not useful for me, since I can't read it". I then realized that he had been editing all of these pictures with translations from Danbooru and some anons in the general thread. He can't even read Hiragana or Katakana, and yet he produces fairly good edits.

    I haven't been keep up with it (there's probably an image pack on the wiki), but he must have edited hundreds of images without even so much as speaking with the translators who put up the translations on Danbooru. That's the kind of efficiency that Oriko Magica needs.

  • My proposition is that we also use a Danbooru-style image board for the scanlation of Oriko Magica. I don't know if anyone has ever attempted a full-blown scanlation effort like this. Nevertheless, I want to give this a try.

    The basic procedure is like this:

    Note that the editing is still done offline with Photoshop (or whatever). The image board is just a mode of collaboration, and is not meant to be a full scanlation suite.

    The diagram collapses many jobs into a few (and forgets QC, just like me). It's presumed that the image gets downloaded, worked on, and re-uploaded by many people doing different jobs. The point of this solution is to try to make the scanlation as parallel as possible; you can think of this as an assembly line of sorts.

    One feature of Danbooru-style image boards that will be abused is tags, because it is important to keep track of which pages are at which stage.

  • Obviously, we can't use THE Danbooru for this. Danbooru's code is open source, but it requires Ruby on Rails and all of that jazz. I tried gelbooru 0.1.* last time, but it was buggy for me. The last choice left is...Shimmie, which I didn't even know about until I read the Wikipedia article on image boards. Back then, I thought that there wasn't a photo note extension for Shimmie, so I implemented one. However, right after that, I noticed that someone was indeed developing one. Nevertheless, I can't get it to work, so I'm using my implementation in the meanwhile.

    It works fine so far, though there might be a few injection attack vulnerabilities here and there, or something like that. Shimmie is surprisingly extensible, and it even comes with a Danbooru skin.

Advantages:
  • Parallel scanlation: Ideally, instead of having to wait for other team members for everything, the translating, cleaning, typesetting and editing will all happen at the same time, so the required time is shorter than their combined time.

  • No group structure required: Members of an orchestra aren't required to know each other. They're just required to come to the rehearsals and the performances and perform their parts together. Just like how the musicians can do their parts from their scores, I trust that given a scene like the screenshot above, editors/cleaners/typesetters would know what to do.

  • No messing up the order of the bubbles: The translated text is right on the bubble, man!

  • Doubles as an online reader: A bonus of this idea is that once it's finished, it can be put under a tag or a pool, and readers can be pointed to that pool if they prefer to read online (hey, it's better than having them read it online elsewhere). Moreover, it doesn't even have to be finished; readers can see the manga being scanlated in real time. Scanlating on chan-style image boards also has this benefit, but with photo notes, even if translation has finished but not editing, readers can easily read the translations, just like on Danbooru.
Disadvantages:
  • Inconsistency: There exists a saying: "too many chefs ruin the broth". Given the translation wars that I've seen happen in the general threads, I have a bad omen about this. For the record, I use "magical girl" purely because the incorrect Latin bugs me, but if the majority overwhelmingly votes for "puella magi", I'll go with that. If we have an even split, everyone will just do what they want. I would rather have an inconsistent translation than no translation at all.

    Furthermore, font family, styling and sizes will have to be agreed on beforehand. The standard dialog font will probably be AnimeAce BB 2.0, because that's what all of cool kids are using.

  • Requires Many People: I'm afraid that I'm just talking to myself and that no one will participate in this. In any case, I'll go ahead with this plan even if no one comes.
Anyways, after I translate 7o3x ch 6 and the next Ryuushika chapter, I'm going to test it with Jiya in private, fix anything that pops up, and probably do it in public with Cat Diary.

6 comments:

  1. Puella Magi is likely not incorrect Latin, I imagine "girl of the mage" was actually intentional. However, I don't approve of using "Puella Magi" due to the fact that as a Latin student I think "Puella Magi" is supposed to always be in conjunction with "Magica" (magical girl of the mage), and referring to everything as "Puella Magi _________ Magica" is just stupid.

    Keep up the good work! I hope things work out~

    ReplyDelete
  2. I hope things work out, seems like a good idea just that like you said it might be hard getting people, that being said i think it could have a fair amount of success if we can get the word out there since ideally people would jump in and contribute to it since it would be a project they like and don't if they don't like it, it doesn't feel as restrictive so i feel that it would get more people to work on the projects.

    ReplyDelete
  3. If this is something that anyone, at any point, can chime in to help, I'll be glad to participate in translation (I don't have any major picture editing programs).

    I've always been interested in seeing how my knowledge of linguistics works in entertainment, considering the limits of space in frames/on the pages and the range of variety in the general anime/manga fanbase's knowledge of Japanese language.

    I hope this works out for the best and doesn't end up causing more hassle than benefit!

    ReplyDelete
  4. @Aster: Well, it is correct, but I mean that it's just not really a correct translation of mahou shoujo/magical girl, which it's supposed to replace.

    @Jordan: Yea, that will be the main problem.

    @Juno: Any help will be appreciated.

    ReplyDelete
  5. This sound like a good idea, as you've mentioned in the OP, the threads and translations, often tend to get messy.

    I fully support this idea, and I surely will be helping out as Cleaner/Typesetter once in a while, when I have time.

    Hope this is project will be good
    ~DKThias

    ReplyDelete