« HOME | SHOWS | FAQ | ABOUT | NYAA | XDCC

Some Vivid Thoughts on Various Aspects of Fansubbing


Why you no typeset signs?

2015-01-25 by unanimated
There have been some concerns about Vivid not typesetting some signs.
I will address these concerns by mostly not giving a shit and just writing whatever I feel like.


TL;DR:

Well, it is, of course, because Vivid is a bunch of incompetent lazy fucks, especially their typesetter, that yatsu unanimated, i.e. me. I don't even know why you watch Vivid when there's HorribleSubs. It should be vividly clear that horrible subs are always better. So yeah, if you spot a missing sign, please immediately drop Vivid and go watch properly-typeset HS.


Not TL;R:

Here comes a long rant about some typesetting-related stuff because I feel like writing one.
After all, having a blog with only one post would look like we're laz— oh wait...

Well, anyways. Let's make some things clear.
1. Fansubbing is something we do in our free time. (Wow, how original. Yes, I know you've heard that 1000 times before, but some of you keep forgetting it every few days.) There are absolutely no rules we are obliged to follow, we can do whatever we want, and we owe nothing to nobody. That's at least my view. Xythar and Devana have nicer views because they're nicer people, but I think that still doesn't invalidate my point. I don't "have to" do anything, nor "should" I do things that people expect from me. That should be the starting premise.
2. I do NOT aim for typesetting every sign. That is NOT my goal. Some morons out there seem to think that that is the goal of typesetting. It only is if you set it as your goal. I haven't. And honestly, I think it's a rather stupid goal, but that's just an opinion. The point is that I don't aim for that. So if a sign is "missing", it's usually not because I was too lazy to do it, but rather because it wasn't my intention to typeset it in the first place, for a number of possible reasons.
3. I'm not writing this to justify anything, because I couldn't care less about what you think. I'm just gonna clarify why some things are the way they are. Unlike some others in Vivid, it doesn't bother me one bit when people complain about my TS. I know what I'm doing and why I'm doing it that way, and I really don't care about what others think about it. You're always free to not watch it if you think it's that bad. I'm just writing because every now and then I feel like writing.

Now, let's look at some mechanics of typesetting.
As at least some of you know, typesetting can be the most demanding task in fansubbing. Can be. Usually isn't. Certainly isn't the most important. But the fact is that typesetting is the only part that can reasonably take over, let's say, 15 hours. (I've seen/done up to about 160 signs for one episode.) Such a case gives the typesetter two options. Either he can just do it as long as it takes and delay the release, or he can cut corners somewhere, for a number of reasons. One of those could be that the guy just doesn't have that much time. Another could be that it would only take that long if all signs were typeset, and the typesetter doesn't have to deem that necessary, again for various reasons.
Cutting corners can also be done in many ways. Some are reasonable; some can be pretty terrible. You may have to cut corners out of necessity when doing everything properly would cause too much lag.
So that's already several limits on what you can do: time constraints, lag, necessity of the sign. I'll get back to the necessity in a bit.

Another limit, one that most non-typesetters don't seem to understand at all, is placement. For some signs, there just isn't enough space anywhere to do them nicely. So you can either do something that looks lame and out of place, or you can skip the sign. If the sign isn't really important, I may choose to skip it not out of laziness, but because it just couldn't look good no matter what I'd do with it, and I think no sign (when not important) is better than a bad sign. That's, again, just an opinion, but that's what I prefer. Often these signs would be really easy to do, but if they make things look awkward, it's just not worth it.

Are all signs necessary?

Hell no. If you look at "professional" subtitles, you'll see thousands of "signs" that aren't even translated, and of course professional subtitle typesetting practically doesn't exist. There is absolutely no reason why every single thing that has letters should be typeset. If you think it actually is necessary, you are so fucking retarded I don't even have words.
This "must TS everything" view is mainly held by Internet idiots who have nothing better to do than make comments that are supposed to show their superiority in some way. It makes them feel good when they can point out that you didn't do something that they "know" you should have. (Not that they'd have a clue how.)
It may come as a surprise to some, but people who download and watch anime are not my boss.

So what's an important sign and what isn't?
That's totally up to the group to decide. There are levels of importance. There are very few signs, in general, without which the movie/episode wouldn't make sense. I would say those are the only really important ones.
There are plenty of signs that are absolutely unimportant. Random signs on buildings or shops that the characters never enter or talk about, a sign in a bus/metro showing the next station when that station has no relevance to the plot, signs on a blackboard behind a teacher that are just part of the classroom setting and have no relevance to the plot, and so on.
What should be the policy on those? In my opinion, completely up to the typesetter/group.
What's my policy? Nothing rigid. If the translator translates them and they don't look like much effort, sure, why not. If it looks like too much hassle for no benefit, why do them? I'm not at school or doing my homework. (Thank fucking god.) I just do some stuff in my free time, and I do it any way I like.

A specific kind are signs that say the same thing as the dialogue.
This is an interesting topic because I see a number of factors that play a role here.
Sometimes I'll do these signs; other times I won't.
Basically I take into consideration all the reasons to do them and all the reasons not to, and I weigh them against each other. Here are some factors:
1. How much effort would it be? (As this can give you anywhere between a minute and a few hours of work, this is pretty important. Especially when there are a lot of these in the episode and you want to be consistent.)
2. How much space is there on the screen for the typesets? (Big issue, as little space means the TS will probably look like shit.)
3. How many other signs do I have to do? (Obviously with 100 signs total, I'm probably not doing this.)
4. Will it make the screen look better or worse? (That's a reasonable concern. If it's easy and I can make it look nice, then it can add something extra, but in other cases, it may just make the screen overcrowded with unnecessary crap.)
5. How easy will they be to read? (If they disappear quickly or there's fast dialogue, there may not be enough time to read them, and this will either make you pause the video unnecessarily or get you distracted and miss the dialogue.)

Now, let's take a real example from a recently-released episode, Durarara!!x2 Shou - 02.
I recommend that you download the episode and check the segment around 3-4 minutes into the episode so that you really know what we're talking about here.





Before we ask the questions above, let me remind you that all these signs only match what's in the dialogue, so you're not missing anything if they're not typeset. Let's look at the picture above.

1. How much effort would it be?
It may not seem like much from the screenshot, but if you watch the video, you can see it would be a fuckton of effort. The signs are not static. They move around and appear with several frames of animation from smaller size to larger, so you'd have to make several frames for each sign.

2. How much space is there on the screen for the typesets?
Now, those who claim that all these signs should be typeset, tell me, how would you typeset the grey one?
There's not much space in the grey area. There's not much space outside of it either. And English text is almost always longer than Japanese.
You could use very small letters and place the text under the Japanese. That poses several problems, though.
- It may not even be readable.
- You can either be consistent and use the same small text for all of them, which will look stupid for the short ones, or you can use different size for each sign, which will look stupid too. Either way, any fucktard capable of complaining that these signs weren't typeset will complain about what you've done because complaining about your work is the primary objective of their existence.
- In this case here, you'll also have overlaps with dialogue, whether you place the dialogue at the top or bottom. The signs are everywhere. So some signs would still be unreadable.
The other option (besides small letters) is to mask the sign and do only English.
Good fucking luck with that. The background is not two colours. It's two gradients. Which means you'll have possibly hundreds of lines for each frame, that you'll be moving around or recreating all the time. Maybe the viewer won't have to pause the video to read everything, though, because it will lag so badly that there will be enough time to read it all.
You may say that the masks could be done with only two colours. Well, sure. It would still be a lot of effort, and as I've said, fucktards who complain about not typesetting this will complain that you suck and you're lazy because you didn't do the gradients.

3. How many other signs do I have to do?
I'm not gonna count, but maybe a hundred here? The segment is about a minute and a half long. The dialogue is very fast, so the new signs appear pretty quickly, sometimes 3 new signs per second. (With each new one, the other ones move lower, and often the screen zooms in at the same time, changing the size of the signs.)
Remember how I told you that you need several frames for the animation of each sign? Yeah, so add and multiply.

4. Will it make the screen look better or worse?
Well, if you try to squeeze the English text somewhere around the Japanese, it will make the screen look worse for sure.
If you go for the masking, it may look good, but with the gradients it's probably not realistically doable at all, and without them it will at the very least be a lot more effort than just text.
And you'll still have overlaps of typesetting with dialogue, which certainly doesn't make things better.

5. How easy will they be to read?
On its own (ignoring dialogue), very hard. Including dialogue, impossible. This tells you what the result of your long hours of very tedious effort will be: nobody reading it.

So, maybe now you can see what kind of effort this would actually take.
Then again, maybe you don't mind if this gets released in April.

And just to make a point, you can have a look at another example.





I hope you like perspective. I hope you like that this is slowly moving for 6 seconds, changing the perspective as it goes. (In other words, it's not just a pan that lets you use linear move.)
I hope you like that it covers most of the screen, making whatever will have a tendency to lag MUCH worse.
I hope you know how to do moving gradients with perspective, because I don't.

Well, I wanted to challenge people to typeset the whole segment, but you know what? I challenge anyone to typeset this one 6-second sign in a way that people won't laugh at.
Please find me one damn typesetter who will do it. Because with so many people convinced that all these signs should have been typeset, there should be plenty of typesetters who can do it, right?
I could do it with small text under the Japanese, but even for that I'm confident that most typesetters out there would fail.
With a gradiented mask, I don't think this is even possible. But go ahead, try it any way you want.

So. Most of the people criticising the lack of typesetting have absolutely no idea whatsoever what it takes to do this.
And I'm using about 50 scripts that I had to write myself to make all this work easier. Without those, you can multiply the time spent typesetting by 3 or 4.

To sum it up, my reasons for not typesetting this were:
1. It's absolutely not necessary because it just matches the dialogue.
2. It would be incredible amounts of effort to do it so-so, and impossible to do it well (at least some parts).
3. In many cases, it would just look worse than not typeset at all.
4. It would be for little to no benefit because you couldn't read it without pausing anyway.
There is one more reason, though in the light of the above, it's fairly irrelevant. For the last 3 months or so, my hands have been hurting like hell when working on the PC. I can do most other things reasonably well, and luckily even typing is manageable (otherwise you wouldn't be reading this), but once I put my hands on the mouse and keyboard in a fixed, stiff position, the pain comes on within 2 seconds, and several hours of typesetting are pure suffering. Heck, I typeset wearing gloves to at least have some soft padding between my hands and the mouse. So it would probably literally destroy my hands completely if I tried this, but like I've said, that's not even relevant because the reasons above are more than enough for any sane person to not typeset these signs.

Am I lazy? Well, lately I have been. It wasn't like that a few years ago. After all, I'm the guy who spent 5 hours doing this:





even though I absolutely didn't have to. I just enjoyed typesetting and liked the challenge.
At this point though, I'm just not interested in fansubbing or anime at all. I haven't really watched anime for a year or so. I haven't even watched through most of the shows I've typeset lately.
And if it wasn't for the people in Vivid, I would have quit fansubbing a few seasons ago.
But I haven't dropped any signs just out of laziness. I don't drop signs that the translator/editor consider important. Laziness can only be an extra factor when deciding if I wanna do an unimportant sign just because I can. And in most cases, I do that, if my hands can take it.

I could get into a number of other things here. For example, knowing what to typeset and what not to, knowing where to draw the line, is the reason why I have probably the best track record of finishing what I start and not delaying things. I have never dropped a show, and I've only ever delayed a few episodes, maybe not counting Acchi Kocchi, which always took me two days. (Let me remind you that it took about 4 typesetters in Tsumiki several weeks, or even months for the last episodes.)

Hdr is a brilliant typesetter who can do amazing things. But unlike me, he tries to do everything and do the best possible version. This is the reason why 1. he's known as the killer of toasters or "Antichrist of Softsubbing", and 2. JoJo isn't finished and a lot of other shows often got stalled.
I'm not saying that's wrong. I admire his work. I'm just saying that everything has its downsides and upsides. It's because I know where to draw the line that there are no unfinished or significantly delayed Vivid shows.

Another thing that comes to mind here is the irony of it all. Typesetting today is incomparable with what it was in 2011 when I started typesetting. The quality has drastically changed, and so have expectations.
The irony is that I have largely myself to blame for that. For a long time, I was probably the primary factor driving the standards forward and up.
Not only did I often typeset 4-6 shows a week back then, but I also wrote the typesetting guide that probably every new typesetter today worth anything has used, as well as those ~50 automation scripts without which most typesetters today (save maybe for those who write their own scripts) would probably give up pretty quickly.
The standards in 2011 were pathetic. Typesetting all signs wouldn't even have come to anyone's mind, and if it had, it would mostly have been just \an8.
But then I had to go and say "WTF is this shit? That won't do." I started hammering blur into people's heads and doing signs in layers, bothered with details that few others did, finished Maria Holic Alive, and so on... and people got used to it.
And now everyone is a fucking expert on typesetting and tells you what you must do. (Of course those who tell you the most have no damn clue how any of it works.)
And if you don't spend 20 hours typesetting something insane that's in the dialogue anyway (because you're not a fucking moron), they tell you that you're lazy and don't care about the quality.
I wonder why it is that no other group is doing Durarara. I wish somebody did. It's always good to have something to put things into perspective, and HorribleSubs doesn't exactly count.

Anyway, I consider myself mostly retired, but not completely yet. I can probably maintain the 2 shows per season that I've been doing lately.
But like I said, it is not and it never was my goal to typeset all signs. If you think that all signs should be typeset no matter what, you're a fucking retard, you have no idea what you're talking about, and you should probably get a life.



Chicago, Webster, and Ellipses—Oh my!

2014-10-21 by Devana
Okay, so this may very well be the death of my much-loved anonymity (or at least inconspicuousness) in the fansubbing world, but... Well, unanimated suggested that I write some posts on my convoluted yet informative thoughts regarding editing, and I suddenly warmed up to the idea.

So I'll mainly be discussing editing-oriented subjects, particularly of episodes of anime I edit or QC (which is not much, lately). Like many others, I'm vaguely aware of fansubbing drama, but just as you might do with a crazy person on the street, I try not to look it in the eye lest it direct its insanity towards me as well. So I'll just stick to writing words about words. :)

There are also some things you, the reader, should know since I'll reference them often.

I was/am an aspiring novel copyeditor. I therefore gravitate towards the style guide used by novel copyeditors: the Chicago Manual of Style (16th Edition). (Hey, if it's good enough for those who edit classic literature, shouldn't it be good enough for subtitles of similarly plot-driven content?) That being said, there are times to deviate from any style guide, and it's not uncommon to handle the same editing issue in multiple ways. There's almost never an absolutely "correct" way of saying something. Language, after all, is constantly evolving, and dialogue (the dominant part of subtitles) often demonstrates how dramatic that evolution can be. Furthermore, a style guide aimed at the printed word will likely fall short in relation to the on-screen word. So the CMoS certainly isn't the be-all and end-all of every matter of grammar, syntax, punctuation, or style; it's just a better option than 'going with your gut' most of the time. The way I figure it, there may not be an exactly right way of saying things, but there sure are plenty of wrong ways.

Now, the CMoS uses Merriam-Webster's Unabridged Dictionary, so I use it too, however grudgingly. I'm much more partial to the American Heritage Dictionary. (Yes, I'm American, although the Oxford English Dictionary would be my second choice.) Personally, I somewhat dislike Webster's (and you can go look up "literally" to see one of my reasons), but I concede that it's one of the largest, most inclusive dictionaries in the world. So I get it, Chicago. I'm still grudging, though.

Aside from the CMoS and Webster's, there are some in-house subtitling rules I use as well. I'll address them as they come up, but there are two I can think of right off the bat. One is that semicolons, while extremely effective and arguably necessary in many technical or literary works, are generally eyesores in subtitles. I avoid them when I can—not quite like I avoid the Plague, but almost as much. Besides, aside from complex series, I don't think there's ever an instance in which replacing a semicolon with a period is totally incorrect.

I feel similarly about the colon's being an eyesore, but it's sometimes far more necessary. And a vague ellipsis doesn't always cut it in those cases, you know? So the colon's a necessary evil on occasion.

The second in-house rule is about ellipses and is so important that I'll just finish the rest of the post with it. So let's talk about ellipses, shall we?

Relevant-to-subtitle ellipses are discussed in CMoS §13.48–56, the most notable sections being 13.53 and 13.54. Interestingly, omissions that occur after a complete thought take both a period and a three-point ellipsis, which means that you should expect four ellipsis points in such a case. Also interestingly, in manuscripts, ellipses are formatted sort of like this: "x . . . ? X" or "x. . . . X," each of which have a few nonbreaking spaces. Authors can use the single-glyph three-dot ellipsis character (i.e., "…"), but in the end, editors replace them with spaced periods.

Now, can you imagine seeing that shit while you're watching anime? On-screen spaced periods look heinous, and the single-glyph crap isn't even existent in some dialogue fonts. Oh, and in some, it just looks like the Charlie-Brown-smile version of the underscore. So yeah, not happening.

Look, editing the written word and editing the on-screen word both seek to achieve the same thing: to minimize the reader's distraction without compromising his or her grasp of the content. So imo, those CMoS ellipsis rules go right out the window when it comes to subtitles.

Other punctuation books (e.g., The Elements of Typographic Style by Robert Bringhurst, Punctuate It Right! by Harry Shaw) have plenty of other things to say about ellipses and their spacing. Really, every possible combination of spacing has been suggested in some or other reputable source.

So Xythar and I talked about it a while back, when I was new to fansubbing, and we agreed that the most uninvasive way to punctuate the ellipsis in subtitles is "x... X/x," regardless of whether the preceding sentence ends in a full thought or not. Personally, I tend towards making what comes after that lowercase (a continuation) if possible. But really, that tendency is a matter of personal preference. Whether you want to capitalize "come" in a line like "Please... come back," is entirely up to you, even if you decide to split it into two lines.

Now, what about punctuation that comes after the ellipsis? Like, what do you do about question marks? I think the answer to that one comes kind of organically. Let's say your character asks a full question but sort of trails off. In that case, your ellipsis is only decorative (the trailing off), and the question mark is pretty important (the full question). But really, I'd say to stick with one or the other. Half-finished questions are generally okay with just an ellipsis, and full questions look fine with just question marks even when they trail off. There are rare exceptions, like when the innocent Ai from Kamisama no Inai Nichiyoubi trails off and seems to be asking not-quite-said-aloud questions to thin air. And really, I probably should've chosen one or the other even when I didn't. But we live and we learn, right?

As long as my sense of grammar and language is constantly evolving, there will always be newfound errors with my old edits. And we all make mistakes of all kinds. I'm just as infallible as the next person, no matter how much I do my due diligence and study these topics. But I said I'd discuss editing and not philosophy, right? :P

Anyway, when it comes to combining punctuation marks (not just with ellipses), I'd say it's generally best to pick the more necessary one. If you read the line with the audio/scene over and over again, and it just doesn't seem right without both punctuation marks, then go ahead and choose both. In the end, whatever is least noticeable to the viewer is what's right...

except when we get to grammatically unacceptable things, like comma splices.

Man, I hate comma splices.

More on that at some other time. :)

~Devana.