Blood+ SRT Subtitles

[TwileDragon]

Being unsatisfied with the vobsubs for Blood+ I extracted the English subs and did OCR on them. I then used SubRip and MS Word to find and fix OCR errors.

Now I have these nice SRT files which can be muxed in with the episodes for prettier, customizable subtitles. All to myself. And that feels wrong. I wish to share them. I was told that I should bring it up here and tell people about my process so they can, like, make sure I did stuff right. So here I am.

I used MKVExtract to pull out the files, SubResync to do OCR, SubRip to do automated corrections ("l" to "I", for example), and Word to find more complex OCR errors which created spelling problems. I did this for each of the 50 episodes. I kept copies of the SRT files at this stage.

Then I did something a bit more controversial. I made new copies and removed linebreaks in the subtitles where I deemed them unnecessary--when a single person talking stretches onto more than one line, for example. I retained linebreaks with things like song lyrics, where the subsequent lines were capitalized, but other than that a single person talking gets a single line. I did this for several reasons--first, the vobsubs were huge, so in some cases you'd have a mere 6 or 7 words split into 2 lines when they can easily fit on one if rendered at a reasonable size. I prefer for there to be fewer, longer lines because then the subtitles stay at the bottom of the image and creep into the frame less. For example:

- Hey! Where do you
think you're going?
- Er, nowhere!

goes to

- Hey! Where do you think you're going?
- Er, nowhere!

and

Row row row your boat
Gently down the stream

is unchanged.

As I said this is a bit controversial because some people like having shorter lines, or exactly what the DVD had, for whatever reason. Some people use programs which aren't that intelligent about dealing with long subtitle lines. I believe that players like Popcorn Hour will just let the subtitles run off the edge of the image rather than wrapping them over onto the next line. So this might not work for everyone.

Still, I kept the un-de-linebreaked SRTs for people who want them. Because I didn't pour over them line by line as I did my personal, fewer-but-longer-delined version, they might still have the occasional OCR error or two, spread out over the 50 episodes. I'm willing to offer them up if people want them, both versions.

Thoughts? Moving this to the right forum section? Questions?

[Takeshi]

Well, I think this should've been posted in the Uploader's Forum but....

I'm just curious why you didn't use SHS's subs and did a OCR on them to offer them with Arigatou video. The main issue with Arigatou's release is that the DVD release of Blood contained a dubtitle based on the dub instead of a subtitle translated from the Japanese track. I don't know if there could be problems with removed scenes from the DVD release. But I would definitely like to see ORC'ed subs from SHS's release that can be used on Arigatou's video.

I recently did a OCR on my Animatrix offer, and I started to remove the two-line subs, too, and put it on one big line. Though I stopped it. But I can definitely see the advantage of having them in two lines. It eases the eye so it doesn't have to travel from left to right all the time. By this you only have to look down to read it, and look up again.

[TwileDragon]

I didn't put it there because this is just like a perk for one of the torrents, not a torrent itself. I dunno, I'm new here ~_~

I used the Arigatou subs because, well, those are the ones I had access to, and because I know they match up with the other tracks. I assumed that if there were a better version of Blood+, that's what would be on BakaBT, because isn't quality the whole point of this site--torrents which are directly inferior are tossed in favor of those which are better quality? If the Arigatou version is what's here, I assume that the streams are better quality, and I'm not trading something vague like subtitle accuracy for something quatifiably and qualitatively better: codec efficiency and bitrate.

If by SHS you mean Shinsen-Subs (as I said, I'm new to this--not super familiar with group acronyms), then I didn't get subtitles from them because a) it'd waste 8.4 gigs of my download bandwidth just so I can get "better" subtitles which, for all I know, won't match up at all, and b) OCR'ing is a betch when you've got the words in their own stream, I don't want to imagine what fun it'd be to pull subtitles from a hardsubbed file.

I won't try to argue the advantage of a subtitle or dubtitle over the other, except to say that dubtitles at least preserve enough of the intent to make the experience understandable. Clearly, whoever dubbed Blood+ thought that the story, character interactions, etc made sense with the new dialog they cooked up, regardless of its authenticity.

I really prefer to remove the superfluous linebreaks even though it's crazy time-consuming (I mapped middle click to [left click, space bar, delete] so I just have to click at the end of a line) because I think the final product looks better (and I'm all about quality) and it forces me to check over the subtitles line-by-line which catches more errors.

[Takeshi]

Anime Discussions is used to discuss the anime itself, and we're not really doing that so this'll probably be moved when the mods starts to punch in...

It's just that a lot of people have pointed out that it's just a dubtitle. Hey, you choose to make the dubtitle more attractive and that's awesome, you seem to have put a lot of work into this.

I haven't seen it so I can't say whether or not the dubtitle is good enough or not. In my Animatrix offer it seemed that most of the stuff got through the dubtitles, but I wouldn't be surprised to hear translators and old timers bitch about them. The company who released it has really been lazy.

When you're done you can put all the subs in an archive and ask the uploader to add it as an attachment. I would suggest that you refrain from styling them unless you know what you're doing. SRT subs are good enough.

But I'd still like to see someone OCR the subs of SHS's Blood release.

[TwileDragon]

I welcome the mods to move it to the correct place

Yep, it is a dubtitle. But it's better than the one which was included in the Arigatou version. I guess I could've maybe made it better. Hell, I could've taken up Japanese and translated it myself. But my laziness has more bounds than my selflessness ;p

I didn't realize that the Animatrix had dubtitles. I thought it was originally made in English, and the Japanese was the dub?

Way I see it, a dubtitle can work one of two ways. They can translate it well for the subs and then use those as a script for the dub, or they can use lip movement to dictate less accurate translations, which they then use for subtitles. The former is double plus good, and the latter is double minus bad.

I did put them in an archive and ask the uploader to add it as an attachment. He suggested that I discuss it with people here first D: I intended to do flat SRTs with no styling, as I only like ASS subs when they make good use of custom fonts, display effects, placement, rotation, etc. I haven't even seen Blood+, I'm not going through the trouble to style it appropriately ;p

[Zalis116]

Since you were looking for controversy, here's a little case for 2-line subs  

Then I did something a bit more controversial. I made new copies and removed linebreaks in the subtitles where I deemed them unnecessary--when a single person talking stretches onto more than one line, for example. I retained linebreaks with things like song lyrics, where the subsequent lines were capitalized, but other than that a single person talking gets a single line. I did this for several reasons--first, the vobsubs were huge, so in some cases you'd have a mere 6 or 7 words split into 2 lines when they can easily fit on one if rendered at a reasonable size. I prefer for there to be fewer, longer lines because then the subtitles stay at the bottom of the image and creep into the frame less.

Two-line subs are used in a lot of DVDs and more intelligent fansubs for a reason -- it's simply easier to read short 2-line subs than long 1-line subs. Basically, it's not a matter of "vobsubs are too big to fit on 1 line"; the 2-line subs are intentional. As explained here:

first, our subtitles are often two lines (two short lines are easier to read than 1 long line because they can fit into your foveal vision and be read with little or no side to side scanning by your eyes),

Having 1 long line means that viewers have to spend more mental energy scanning across the screen reading the subs, thereby increasing the chance of missing out on other areas of the image. Same with having subs way down at the very bottom of the screen -- viewers have to constantly "bounce" their vision downward to read the subs, and then back up to see what's going on. So in a way, 1-line subs at the very bottom of the screen spanning the entire width of the image (like typical fansubs) are more "obtrusive" than short 2-line subs slightly higher up (like most DVD subs), because they demand more active energy and attention to read. Also, subs that stay away from the very bottom and extreme sides of the screen won't get hit by overscan on older displays, like if people are outputting to a CRT TV.

A given amount of text will always cover the same amount of space, regardless of how high it's placed or how many lines it's broken into. So if it's going to be covering screen space either way, it's better to make it easier to read. Besides, the human eye can interpolate what lies behind the letters. And if people absolutely must see something, they can always switch the subs off for a moment if the subs are soft.

That's why I intentionally put hard line breaks in the subs I work on. Fortunately, the same effect can be accomplished at the viewer level by increasing horizontal margins, which is what I do with most softsubbed fansubs and rips. Still, forced line breaks at logical points can make lines more coherent. For instance,

Go to the store and buy \N
a head of cabbage

is more coherent than

Go to the store and buy a
head of cabbage.

Now, I do agree with tweaking things to avoid 3-line subs, and I have seen some DVDs take it too far, like "How do \N you do?" in AnimEigo's Otaku no Video. Also, Aeigsub (the program most fansubbers use) has a spell-check, and you can also use Find+Replace to remove line breaks if you want to. Just search for \N and replace with a space or nothing.

[TwileDragon]

I still maintain that linebreaks in SRTs are usually crappy, just like linebreaks on websites can be crappy. You don't make a website to be read with a fixed font size and window size, nor should you for anime. I might change the font style and size to better suit my preferences, distance from the screen, resolution, etc. With my approach, and any decent player, you can limit the number of characters on one line and it'll automatically add the linebreaks during playback. Flexible. With the hard-coded linebreak approach, all you do is lose the distinction between a necessary linebreak (song lyrics, new person talking, etc) and those which seemed good at the arbitrary font size the subtitle maker was using.

I mean, isn't that half the point of SRT (the other half being better resolution scaling)? The ability to pick font, size, characters per line, position on screen, and more? Which also nullifies the point about overscan, because you can adjust the subtitles to be higher in the screen.

I mean I understand the need to have subtitles the way they are on DVDs that are shipping, and even on Blu-rays. They're intended to be viewed on TVs, TVs which might have the overscan issues you mentioned, with devices that can't always be counted on to do fancy stuff like rendering in an arbitrary font at an arbitrary resolution. Things that most people probably don't care about very much. But that's just not the case with SRTs.

I can't blindly remove linebreaks though, because they're needed for cases where multiple characters are talking, you have song lyrics, and other special cases that don't come to mind immediately.

So my overall point, although some people might say it's more comfortable to have shorter lines, that can be achieved in software per the user's preferences for SRT files.

[Havoc10K]

long lines are terrible no matter how convenient they seem to be, in the case of long speaking yo ucan short it out as well with

l went to the parl to play a violin for my friends
    because l love them and l love violin too

in a monolog

instead of

l went to the parl to play a violin for my friends because l love them and l love violin too

wich you imply from what l'm reading, this is simply bad, because when you do that, viewer needs to read the whole thing before he can look at the screen, while in two liners it's a lot shorter, because our eyes already catch the second line as we read the first one and it's faster to read the second line.

adding a third line would also be a bad thing as it would take away too much space.

l went to the parl to play a violin
for my friends because l love
 them and l love violin too

it takes actually more time to read this way because the second line keeps interrupting the reading of the third line, l often found myself pausing in such moments and even then my eyes couldn't focus on the screen and went to the second line. this is the reason why most fansubber groups stopped using multiliners and stick to 2 liners, even if that means faster reading pace, or pausing for a second because the text might switch too fast, even if the sentence is long but stil lmenages to be in 2 lines before it flashes into next sentence, keeping a 2 lines ratio is the most efficient method.

when you do those fixes you should also be careful with timing, some subs are like you listed

For example:

- Hey! Where do you
think you're going?
- Er, nowhere!

goes to

- Hey! Where do you think you're going?
- Er, nowhere!

but in some cases the timing in subs can be even brought to one line per 3 seconds so it can easily become :

- Hey! Where do you think you're going?
[3 second period between voicing while the character looks around or something]
new line
- Er, nowhere!

this is also used more frequently whenever avaible, so you should keep an eye out fir such scenes.

good luck with your work.

[TwileDragon]

They're the opposite of convenient. It takes at least 2-3 minutes for me to edit the linebreaks out of a 22 minute anime episode. Multiply by 50 episodes and that's easily 2 hours of extremely monotonous work. But I do it because I prefer how it looks. And anyone else who does is certainly welcome to my subtitles. People who don't can have the older version of the subtitles, before I removed the linebreaks. Although at this point, nobody has expressed interest in them, period.

In cases where there's a lot of dialog, the preferrable thing to do, I think, is to use ... to break it into several different sets of subtitles.

I went to the parl to play a violin for my friends...

...because I love them and I love violin too

with the second line appearing after the first has vanished.

Because I don't watch the subtitles with the anime as I edit them, timing issues might in theory occur, although I don't really understand how. The time it takes for your eyes to scan from the first line to the second is absolutely minimal and shouldn't impact what line you're reading when certain characters start talking. Even if it does, there's so much variation in a person's reading speed that there's no guarantee at all they'll be at the right point at the right time, anyway. If there's a long enough temporal gap that reading speed is an important consideration, or if dramatic effect is so important that what you read has to match what's being said just right, then they should really break the subtitles up so they don't all appear on the screen at once. Which they often do.

If anyone actually wants my archive of the before and after SRT files, I can provide them, and you can use or edit or butcher them to your heart's content.

[Havoc10K]

l really don't have anything against your work, maybe you could apply to some fansubbing group or something, you certainly picked up a long series to fix subs for, if l were you l'd try with 11-12 eps series first, there are some that could use a fix certainly, l personally don't really give a damn about 2-3 liners, l just read, it's like a book.

l know that editing this stuff takes a huge load of monotonious time, l was running Ragnarok Online server in the past and that was also a long work, in many cases l was simply editing stuff and uploading patched versions, this can be really monotonious, also editing pictures is also a dull work sometimes, if you really want to do this don't complain, no one asks you to do it, youre doing it mainly for yourself and possibly for others, l think there will be ppl that apreciate you work, that in it'sefl should be enough of a reward, in that regard, eighter continue or drop if you don't really want to do this.

the time when l was watching this series was like 3-4 years ago and l marathoned this and it wasn't all that great to me back then, maybe l should rewatch it, l don't even remember whose release l have but l have the 2 lined one.

[Zalis116]

Well, I didn't want to imply that you shouldn't have removed the linebreaks -- I just wanted to present the case for why they were originally in there. I guess I was thinking more about hardsubs and .ass softsubs, both of which are less flexible than .srt. (That is, you have to change styles on .ass subs every episode if you want shorter lines, as opposed to .srt which will always look the same.)

As for "blindly replacing," I didn't mean "Find \N and Replace All" -- rather, you can go through by pressing "Find Next" and then "Replace Next" if you deem a line break unnecessary.

And at any rate, I'm sure people will appreciate your work, linebreaks or not. Although Blood+ is not a show I'm interested in, I always support people's efforts to change/convert DVD vobsubs rather than just complaining about them.

But I'd still like to see someone OCR the subs of SHS's Blood release.

If by SHS you mean Shinsen-Subs (as I said, I'm new to this--not super familiar with group acronyms), then I didn't get subtitles from them because a) it'd waste 8.4 gigs of my download bandwidth just so I can get "better" subtitles which, for all I know, won't match up at all, and b) OCR'ing is a betch when you've got the words in their own stream, I don't want to imagine what fun it'd be to pull subtitles from a hardsubbed file.

You're right, OCRing hardsubs from a video file is far more time-consuming than OCRing .sub files. Maybe I was doing it wrong, but I estimate it took me 45-60 minutes per episode for some that I did. Plus you would have to do some retiming and manually redo all subs that aren't in the primary dialogue area, like karaoke and onscreen text translations.

The easiest way to produce accurate softsubs for Blood+ would be to take the .srts that TwileDragon has so kindly produced, watch the Shinsen-Subs releases, and adjust lines where the R1 dubtitles clearly deviate from the actual translations. That would be faster than OCRing Shinsen's subs, but it would take some translator knowledge. And a lot of time, of course.

[TwileDragon]

l really don't have anything against your work, maybe you could apply to some fansubbing group or something, you certainly picked up a long series to fix subs for, if l were you l'd try with 11-12 eps series first, there are some that could use a fix certainly, l personally don't really give a damn about 2-3 liners, l just read, it's like a book.

l know that editing this stuff takes a huge load of monotonious time, l was running Ragnarok Online server in the past and that was also a long work, in many cases l was simply editing stuff and uploading patched versions, this can be really monotonious, also editing pictures is also a dull work sometimes, if you really want to do this don't complain, no one asks you to do it, youre doing it mainly for yourself and possibly for others, l think there will be ppl that apreciate you work, that in it'sefl should be enough of a reward, in that regard, eighter continue or drop if you don't really want to do this.

the time when l was watching this series was like 3-4 years ago and l marathoned this and it wasn't all that great to me back then, maybe l should rewatch it, l don't even remember whose release l have but l have the 2 lined one.

:p I already did it, is the thing. That was my point in posting the thread, to say "Hey, I did this, if you want to use it I can give it to you..." No idea how good it is myself as I've never seen it. I tried to not pay much attention to the content of the subtitles, I picked up something about vampires, pregnant, concert broadcast >_>