Creating Fansubs - Questions and Answers

[jfs]

Aegisub on Mac is not pretty :P It's way too hard (IMO) to make a good build of it and several things don't work too well.
I'm still hoping to make a useful build of it... some day.

[Skyward]

Yes, you can extract the ass files out of mkv files, as long as they are softsubbed. You will need mkvtoolnix and (optionally) mkvextractgui if you don't like using the command line. Assuming you're using mkvextractgui, just open the file in it and select which tracks you want to extract. The font attachments listed are fonts contained in the mkv and are usually used by the ass file.

Do note that the ass file will already be timed though.

[Sylf]

Originally Posted by adrianmak

there are quit a many under windows.
how about on osx ?

There's also Miyu. It doesn't handle ass, as far as I can tell.

[Sableye]

Need a bit of assistance here.

I have the video and audio of the episode that I am fansubbing. The episode has been fansubbed and I hardsubbed my subtitles into the video. Now I also have the video and audio of the opening and endings. Karaoke is done. My question is: how do I put all of this together? Do I combine the audio and video files together and then mux them with mkvmerge? What program would I use to merge the audio and video files? Or is there a different way to do this?

[jfs]

Usually you'll join videos together (eg. op+episode+ed) during encoding, using Avisynth, such that you get one long video stream with everything in it.

For combining things together, remember that if you can avoid touching the audio at all that's the best, as soon as you start encoding the audio you're transcoding it from one lossy format to another which means an inevitable quality loss. Use the audio directly from your raw when possible. (The exception is DVD rips, you'll usually want to transcode the audio there.)

For muxing video and audio together you can use mkvtoolnix or AviMuxGui depending on what your target is. Don't ask me about MP4 though.
(Muxing simply is taking one or more data streams and putting them together into some sort of container. You don't first "merge video and audio", that is the actual muxing.)

[Ray*]

Could you please tell me what program should I use if I want to hardsub the karaoke in a .mkv file, with the rest of the translation being softsubbed?
For example, like in Formula-Subs' release of "Shugo Chara!" ?
Thanks in advance.

[martino]

http://aegisub.cellosoft.com/docs/At...itles_to_video

^ That has it written out very well.

So basically you keep the karaoke/handsubs in a different script than the dialogue/softsubs and just do the magic with both of them. That being, TextSub the hardsub script(s), and mux in the softsub script(s).

[Ikki-Kun]

Looking for a nice encoding tutorial, if anyone can help me out I'd appreciate it!!

[Sylf]

Oldie, but goodie.

[martino]

here's something called "beta 3", so I suppose it's something updated. Don't know to what extent though since I've never really used them myself, just remember having them bookmarked.

Some other stuff that might be of some use;
http://www.fansubbers.org/uploads/KB/doom9xvid.pdf - Xvid configuration explained
http://aflux.deltaanime.net/Zero1/MP4/x264.html - x264 configuration explained

[Ikki-Kun]

Thanks guys, I owe you one!

Now I can start encoding my own stuff to see the finished product 8D

[raziel666]

I would like to ask if there's some way to enable deblocking through ffmpegsource, similar to "deblock=false" that AVCSource has. I've been reading the documentation of ffmpegsource and a couple of topics in doom9, but I couldn't find any good example of this. Has anyone tried such a thing? Thanks a lot in advance.

[martino]

Originally Posted by raziel666

I would like to ask if there's some way to enable deblocking through ffmpegsource, similar to "deblock=false" that AVCSource has. I've been reading the documentation of ffmpegsource and a couple of topics in doom9, but I couldn't find any good example of this. Has anyone tried such a thing? Thanks a lot in advance.

None that I'm personally aware of. The deblock parameter is quite specific to AVC/H.264 decoders, just like you can switch it off in CoreAVC and ffdshow. However the deblocking is most likely enabled by default. Then you can always try AviSynth filters to do some cleaning up that you might require, like DeBlock.

[Nicholi]

The ppquality parameter. Unless you were referring to H.264's in-loop deblocking which of course is always on.

[TheFluff]

Originally Posted by raziel666

I would like to ask if there's some way to enable deblocking through ffmpegsource, similar to "deblock=false" that AVCSource has. I've been reading the documentation of ffmpegsource and a couple of topics in doom9, but I couldn't find any good example of this. Has anyone tried such a thing? Thanks a lot in advance.

As Nicholi says, if you're talking about the H.264 inloop deblocking, that's always on in ffmpegsource (not that you should ever want to turn it off). If you REALLY want to turn it off for testing, use directshowsource and ffdshow or something.

If you're talking about the usual postprocessing kind of deblocking, then yes, that is available in ffmpegsource (but turned off by default) through the ppstring and ppquality parameters. I don't see why you should use them though, there are much better deblocking filter plugins for Avisynth.

[raziel666]

@martino, Nicholi & TheFluff: Thank you all for your answers, you sure did enlighten me!

[Heibi]

Just purchased one of those new LCD TVs. Watched our works and other groups' works on the new set and saw digital fragments in nearly all the dark scenes(i.e. Space scenes, dugout scenes...). Basically the old DVD problem that was since fixed long ago when little digital boxes could be seen in the black areas.

Was wondering if anyone has encode tips for this? I'm sure there's a thread somewhere but I must not be searching very well or with the right search wording. A best settings thread or such for encodes?

Thanks for any help.

[TheFluff]

Originally Posted by Heibi

Just purchased one of those new LCD TVs. Watched our works and other groups' works on the new set and saw digital fragments in nearly all the dark scenes(i.e. Space scenes, dugout scenes...). Basically the old DVD problem that was since fixed long ago when little digital boxes could be seen in the black areas.

Was wondering if anyone has encode tips for this? I'm sure there's a thread somewhere but I must not be searching very well or with the right search wording. A best settings thread or such for encodes?

Thanks for any help.

Well, since this is a filtering question maybe it should go in the Avisynth help thread but having it here doesn't really hurt anyone so here goes:

Your problem is pretty common. It's a whole lot easier to see blocking in dark scenes on LCD/TFT monitors than it is on CRT ones, mostly because dark areas are, well, lighter on LCD's. Unfortunately these blocking artifacts are part of how the DCT-based compression algorithms in the MPEG2/4 codecs work so it can be really hard and really annoying to get rid of them.

First off you need to apply filtering before encoding. There are several different strategies here; perhaps the most common one is to try to dither the image (the dark areas in particular) using various smoothing filters, another is to add gaussian noise/grain to try to "hide" the blocking (I'd not recommend using this except in very extreme cases).
After that you need to tweak the encoder settings to preserve these filter choices; with x264 you usually want to use some kind of adaptive quantization scheme (usually Haali's) to force it to use more bits on the dark areas where it'd normally block. Experimenting with the inloop deblocking filter parameters can also have interesting results. Do not use trellis=2, it's frequently worse than 1 but is much slower.
With XviD things are a lot harder; you pretty much need to empirically determine what the encoder does to different kinds of input and adapt your filtering to fit that (this is also true with x264, but since H.264 is a much better compression scheme and x264 has a lot more parameters to tweak, it's easier there). I never got really good at this since I concentrated on x264 instead; someone like Mentar or Nicholi who encoded tons of high quality stuff with DivX5 and the like could probably tell you more.

[Dark Shikari]

Originally Posted by TheFluff

After that you need to tweak the encoder settings to preserve these filter choices; with x264 you usually want to use some kind of adaptive quantization scheme (usually Haali's) to force it to use more bits on the dark areas where it'd normally block. Experimenting with the inloop deblocking filter parameters can also have interesting results. Do not use trellis=2, it's frequently worse than 1 but is much slower.

Various things I've found:

1) AQ can work well (even my VAQ, which was designed for live action, not anime), but one of the main problems seems to be that at lower quantizers, the video can look more blocky due to the inloop deblocker being weaker. I would suggest trying high deblock settings.

I would strongly suggest experimenting with VAQ, especially since all the places I've seen it fail in anime were at absurdly, unnecessarily low bitrates.

2) Trellis-2 shouldn't be a problem except on particularly detailed videos; I would think anime is the last place it would be a problem. Its a lot slower though.

[Heibi]

Thanks, I'll continue to experiment with the settings using your suggestions. It was all so simple before the new LCD screens.

Heibi
Central Anime