Timing Guide

[clem-kun]

Okay, it wasn't it. I think Ac3Dec only works with .vob files, which I do not have. I have an .avi file...

Is there some filter for VirtualDub that can decompress audio in .avi files?

[curlyconnor]

The reason why is because I didn't know you could do that on SSA (I'm still fairly new to this). Now that I do, I feel like banging my head into my keyboard, typing in every one of those values. I've gotcha now, but what's that thing about scenechanges?

[getfresh]

http://www.evabeast.com/flf/wbboard/...d=10&styleid=1

read section 3 of this guide for an overview on scene change overlaps ^^

[StormD]

Originally Posted by getfresh

The fastest and most precise way of going about it is still by timing using waveforms as ref. Timing using a linear method is obsolete and has been for some time. Also having to stop to type in values is a waste of time when you can just use the capture button in the time from wave mode within ssa. Not tring to flame you or anything, but why in hell are you using that method? If you can explain to me in a convincing manor why that method is useful then more power to you. I say this because I've been timing for quite awhile and never seen a benefit to linear sub time capture...

I like the results I get from using Subtitle Workshop to capture timings. I can see the benefit of having a visual waveform queue, but honestly, you can get just as good by using your ears and the capture button while watching the video (especially since SW lets you play at reduced speed), and tweaking times up and down slightly to correct irregularities, and get sub in/outs exactly on key-frame changes when necessary. You get the best of both worlds, in that you can click to capture/adjust timings, while still getting to watch the video and use on-screen queues effectively. Once I'm done with SW, All I have to do manually is define my styles (SW doesn't support multiple-styles in output).

Of course, this method is no good for karaoke if you're into that (which I'm not), but for basic dialogue timing, and especially for things like on-screen text that is totally reliant on what's in the video display, I think it's currently the best solution out there.

[ichido reichan]

actually timing for me become even more fun, I hook up my genlock and I do the timing "on the fly" that is watching the movie in real time and turn on-off the subtitles depending on what character is talking.

after I got a "beta timing" I go back and check those missing lines and adjusting the actual subtitle until is satusfactory.

when I finish the script, the last thing I do is put the credits on it.

I can understand those fancy karaokes that everybody likes nowadays, I think those are far distracting with all the funky colors and effects, it gets mixed up with the opening and finally you never cant catch the words to sing at the same time.

[SirCanealot]

Capturing the timings "on the fly" like that, is in no way going to be as fast or as acurate as using the wave file. Human reaction time hangs around 25-30ms (microseconds) on average. So, you play the file back at half speed, you half the problem with human reaction time down to around 10-15ms, and you double the ammount of time it takes to time it - therefore completely getting rid of any advantage in doing it this way. Add to that - I know I'd notice if there was a 10ms delay before a subtitle came up whenever someone spoke. Anything other than grabing your start times with a wave is going to be worse imo. The wave allows you to(usually) grab the start time within a few ms, not 10-30.
And going back and having to tweak every other subtitle's start time is either going to be even more work than fine timing usually is, or look horrible (as the timing all of ichido reichan's fansubs does imo).

[exedore]

Originally Posted by SirCanealot

Capturing the timings "on the fly" like that, is in no way going to be as fast or as acurate as using the wave file.

Quoth someone who's never seen a tape from Central Anime. Every one of their subs was done on an Amiga (no wave capabilities, 20MB hard drives) and timed by Todd Perkins sitting in front of the LD player with a joystick. Sure it took him 2-3 passes to get it done, but Central never was (is?) about speed subbing, just quality scripts. And damned if their scripts aren't quality.

[StormD]

Well, my standard procedure allows for the delay in my reaction time. As soon as someone starts talking, I click on the video screen to pause it, click 'set start time' and then click the down arrow on the actual time drop once or twice depending on how long it seemed I missed the timing by (each click adjusts .1 seconds), then click on the video to re-start it. Rinse repeat for closing the subs. I'd say it probably takes the same amount of time as clicking through a .wav file and setting the times that way. I'm not claiming to save time in the actual timing, per se, but I do think I'm cutting some corners in the cleanup process (because I can check as I go that I'm not overlapping key-frame transitions, as well as being able to time on-screen text accurately during the same pass. Plus, I don't have to futz with extracting the .wav file in the first place. On average it takes me about 2h to completely time (and by completely, I mean good enough to pass QC) a 30m episode.

[SirCanealot]

And how long did it take Mr Perkins? I'm not saying it can't be done, it's just that a wave file is the faster of the methods of rough timing anyway.

It takes me about 2-4 seconds to time a short line in SSA - identify and select start point, end point and grab times. The same on the longer lines, except I actually have to listen to the line. And that's a very acurate grabing of the start and whatever end times you want to give a sub. There's no way you're matching my speed doing it any other way. A rough time should be perfectly rough - it's not about having to adjust anything up or down when you're just rough timing it.

Of course, in a race you'd still beat me, since I take AGES in fine timing, and like to get things timed PERFECTLY (which is something I give great respect for when I see other groups do it) - which accounts for a lot of crap I don't have names for, not just checking for scene bleeds and subtitle blinks. But that's just me. I'm stupidly fast when I want to be when rough timing.

Edit: Please don't take this as me saying I'm better than you, I'm not.
I have to say timing is about finding a method that most suits you. I have done, and being the type of PC user I am, I couldn't see any other way being better, apart from taking short cuts. If you time sitting upside down, attached to the ceiling, then bloody well so be it!

[zalas]

Originally Posted by SirCanealot

And how long did it take Mr Perkins? I'm not saying it can't be done, it's just that a wave file is the faster of the methods of rough timing anyway.

It takes me about 2-4 seconds to time a short line in SSA - identify and select start point, end point and grab times. The same on the longer lines, except I actually have to listen to the line. And that's a very acurate grabing of the start and whatever end times you want to give a sub. There's no way you're matching my speed doing it any other way. A rough time should be perfectly rough - it's not about having to adjust anything up or down when you're just rough timing it.

If someone is timing using a genlock in real time, he'll definitely beat you in rough timing speed for shows with a moderate to heavy amount of dialogue. This is because the shorter lines he'll get through real time, and for the longer lines you'd spend at least the amount of time he's spending, since you have to listen to everything anyways.

[SirCanealot]

He can't time it in real time though - he has to hit the button when they start talking, hit it when they stop, then at somepoint they need to ajust the timing backwards to account for human reaction time. Even seasoned gamers I know don't go under 15ms when tested :P

[Mr_Paper]

Just to play the devil's advocate for a moment...

You're assuming he's timing it in real time using only the audio for the timing ques. It would be safe to assume that he might also be using video ques to aid in the timing. Particularly with older series, as the mouth tends to open about 13ms before the coresponding audio is played. If he's fast enough to pick up on that, the rough timing can still be fairly accurate. Seasoned gamers might only have reaction times of 15ms, but I'm sure their ability to predict timings is far more accurate. =P

[ichido reichan]

Originally Posted by Mr_Paper

Just to play the devil's advocate for a moment...

You're assuming he's timing it in real time using only the audio for the timing ques. It would be safe to assume that he might also be using video ques to aid in the timing. Particularly with older series, as the mouth tends to open about 13ms before the coresponding audio is played. If he's fast enough to pick up on that, the rough timing can still be fairly accurate. Seasoned gamers might only have reaction times of 15ms, but I'm sure their ability to predict timings is far more accurate. =P

actually timing in real time is preety cool, not because you are following the story, but because you have to listen to the breath and the animation to know when they will stop and when the next person start talking.

actually on SSA I use 2 buttons instead of one, the start button (space) and the tab button (end) but after finishing timing, when you do a mistake you, you have to go back and time it again with the wav file, I usually throw the script backwards
50 ms (the natural delay that has been talked about) and after that I adjust the lines that were out of the context.

One of my rules is:

-you must leave some time to people to read the subs

-the subtitle cant be as short as 90 ms or as long as 5-6 seconds

-lines must be mandatory 2 per page, 3 lines on a screen is distracting.

-subtitle must have at least 10-10-30 (wide and height) or 10-10-35 on digisub

-Script must be corrected at least 3 times.

[ff7799]

Mods This topic needs to be made into a sticky.

[LytHka]

Sakaki-'s way is still the best way of timing anime, no matter how you look at it. o.o

[zalas]

Originally Posted by SirCanealot

He can't time it in real time though - he has to hit the button when they start talking, hit it when they stop, then at somepoint they need to ajust the timing backwards to account for human reaction time. Even seasoned gamers I know don't go under 15ms when tested :P

Reaction lag is usually around the same amount, especially if it's a repetitive task, so he/she can just brush through once timing real time and then tell SSA to shift everything earlier by say 200ms, depending on his/her reflexes.

[ZeppelinJ0]

"(Note: If you get a warning that the video stream uses vbr audio redo this step selecting "Nandub VBR MP3 compatibility mode”)"

Where do i find this???

[Secret Squirrel]

Originally Posted by runpsicat

....their viewing experience is then compromised by the fact that they start noticing all these "glitches" they were not aware of before (the fate of all QCers, apparently).

..tell me about it! I used to enjoy most groups fansubs without complaint. Now I can't help but automatically QC every damn ep I watch

..and as a "trainee timer" teaching myself, this thread is one of the best reference guides out there! especially with people posting their "pet hates" to watch for things to avoid.

[SirCanealot]

Originally Posted by Secret Squirrel

..tell me about it! I used to enjoy most groups fansubs without complaint. Now I can't help but automatically QC every damn ep I watch

Heh, that allows me to really appreciate a quality product though - whenever they do come along.

[Imaginer]

Originally Posted by StormD

I like the results I get from using Subtitle Workshop to capture timings. I can see the benefit of having a visual waveform queue, but honestly, you can get just as good by using your ears and the capture button while watching the video (especially since SW lets you play at reduced speed), and tweaking times up and down slightly to correct irregularities, and get sub in/outs exactly on key-frame changes when necessary. You get the best of both worlds, in that you can click to capture/adjust timings, while still getting to watch the video and use on-screen queues effectively. Once I'm done with SW, All I have to do manually is define my styles (SW doesn't support multiple-styles in output).

Of course, this method is no good for karaoke if you're into that (which I'm not), but for basic dialogue timing, and especially for things like on-screen text that is totally reliant on what's in the video display, I think it's currently the best solution out there.

I have just started to play around in the subbing and timing field and yes I do agree. SW is better in getting down the times of the subs and alot better interface for that as well as you can not only listen, but view the video as well. (though lacks the styles and karaoke stuff of SSA)

SW also allows an ADJUSTABLE WINDOW SIZE! SSA however, it likes to stay at full screen. Grrrrrr....