While I love the weekly “what is TLC?” jokes, I figured I’d make a little summary of each thing we have over in the sidebar to the right for future reference. Note that this is for original translations; simuledits are slightly different.
Encoding: Creating the actual video file. First a workraw is made for the group to use. The workraw has a smaller filesize and minimal processing and just exists to let the other work start faster. After that the premux is made, which is done with proper processing and is attached to the final release. If you’ve ever done any video processing, you’ll understand that this is cpu intensive and takes time (read: hours).
Timing: Deciding how to split all of the dialogue lines. I do this before translating for convenience. First you “rough time”, which is cutting off each line exactly where the audio starts and stops while keeping in mind the keyframes and scene changes/cuts. If a character speaks a line that runs over two different cuts, I’ll usually also cut the spoken line there too for aesthetic purposes. Then you do “fine timing”, which is adding lead-in and lead-out (the subs are visible before and after every spoken line) and snapping between lines and keyframes. Timing is easy to learn, but takes a fair bit of experience to be good at.
Translation: You know, making the entire script into English (or whatever target language). For most shows you only have the audio to go by, but for LWA and some other shows there are closed captions in Japanese to refer to.
Editing: Going over the translated script and, well, editing it to be nicer. This can include small changes like line spacing or comma placement, or big changes like reworking an entire conversation to flow better. Translators and editors have a unique relationship where you can really feel each other’s base writing style and how your upbringing and surroundings are different. Back in the old days, TL used to be making a Japanese-sounding script and editing was making it English-sounding, but nowadays the original TL is designed to be read from the start, and editing makes it better.
TLC: Translation check. Traditionally, a second TL in the group goes over the script to make sure there aren’t any errors, but there’s kind of a dearth of us these days. We use it to mean the TL going over the edited script and making sure there weren’t any accidental errors added, and also to double check himself.
Typesetting: Making all of the Japanese signs into English. This can be very easy (if there are no signs) or very hard (if there are a ton and they move all over the place). In LWA they’re mostly in English already (thanks Tattun), so our job is easy. TS may also include styling the songs, but doing so is so rare we usually don’t even bother including it in the percentage.
QC: Quality check. You combine the script, songs, and signs into a single file and watch through it as a whole to make sure it all works. During this point we check everything “in motion”, as the viewer would see it. It’s the last chance to catch stupid mistakes like typos and incorrect layers and timing before someone yells at you for it. All that’s left after this is muxing and releasing, which are so fast/minor that they don’t need to be put up here.
Love: Without love it cannot be seen, but it’s always at 100% or more.
I’ve seen various write ups different groups have done on the timing, and editing processes they use, but I’m curious what the background is for a typical encoding process. I know it’s CPU/time intensive, but I’m interested in how the encoders go about their work, and what they place value on. Do they spend hours agonizing over which encoding settings to use, and what is considered to be “good enough” or unacceptable.
I had meant to mention that it’s archaic black magic. If you’re lazy you make a script for one episode and then use it on the rest because they likely don’t differ much in quality, if you’re crazy you scenefilter frame by frame or something.
The encoding process is basically this:
1. preprocessing and filtering the source using Avisynth (includes stuff like debanding, antialiasing, denoising, etc.). For most shows, you’ll have a base filter chain that you apply to every episode. This is because each episode will tend to have similar problems in the source. The time consuming part of this process is analysing the source and deciding what filters are needed–and in what order are they needed–to satisfy *most* of the scenes.
2. There may be scenes where the base filter chain doesn’t suffice. So you do filtering of individual scenes (scenefiltering) on top of it. You obviously don’t want to apply heavy filtering that is required in one scene to every single other scene.
3. Then you guesstimate your x264 settings (crf, deblocking, and so on) and serve the filtered source to x264
4. Review the encoded result and see if x264 settings need tweaking (potentially time consuming because you’ll be waiting hours for it to encode)
asenshi’s love-filled fansubs, yum
Good article, though I’ve been present in a few groups that do some of the steps out of order. Nonetheless, more transparency about how fansubbers do what they do is going to slow down the death of the scene. 🙂
Much appreciative of the fact that you have release status listed and broken down by category, though not as appreciative that you probably all are that you don’t get *as much* annoying inquiries as to how the next episode is coming along…
Is this where we insert the weekly “what is TLC” joke? Or perhaps where we pretend that karaoke is still a thing in 2017 and ask about it? 🙂
>not singing along
Back to /b/
Question regarding encoding..
I have been encoding for a while so i know a bit abt it like debanding, antialiasing, denoising, later choosing crf and stuff. But one thing that bothers me is that after encode in a lot of cases the OP ED or some signs blink and in some cases dont appear at all. How do i Fix this?
Abt Editing Sub:
When I open subs in aegisubs to edit them The sub blocks & some effects dont appear, also since sometimes the fonts are not installed they are replaced by existing fonts in my aegisub. Is there a simple solution to these problems?
But how is karaoke even necessary for singing along? The lyrics, sure, but animating them, not so much.
This was supposed to be a reply to Cezar, but it seems I dun goofed.
Former typesetter here. Let me tell you how much of a nightmare typesetting can be:
Acchi Kocchi effectively destroyed any joy I had being part of fansubbing.
DAE I MISS YOU ;_;
<3 I'm back on IRC in #evetaku again.
Boy it’s been a while since I’ve seen a group as dedicated as you guys. Feels like the old days.
love of cours!
Thanks! This was really informative! However, I am still curious about what it means to have one of the progress bars go to 200%, as it has happened before. Was that a mistake? Some kind of joke? 0.0 Very mysterious…
Some people do that as a joke. Most of the time, it’s indicating that someone had to redo certain work, or we found another sign we had to typeset after thinking we were done with it.
Please just keep doing this. I feel so much passion but it seems useless to comment at the same time. So why do I even post this? No one cares just keep doing episodes. I need it! NEED!
But why there’s no love in release status right there?
Love surrounds us.
I always thought it meant “tender loving care” 🙁