dovac linked this and we had a nice lol about it ’cause its like a wall of tl;dr that doesn’t really come off as “newbie friendly.”
Basic Fansubbing Tools
- AegiSub: Currently the best program for subtitling assistance. If you’re looking to get into fansubbing, I recommend that you download AegiSub and read over some of the Manual.
- Encoding Faggotry Tools: VirtualDub, AviSynth, meGUI
Fansubbing Related Termo
- .ass (Advanced SubStation): File format used for subtitle files.
- {} brackets: In .ass, things written in-between {}’s don’t show up on-screen.
Fansubbing Jobs
- Translation (TL): Translating the anime’s dialogue from Japanese to English. Also responsible for any important signs (such as episode titles) and songs that are in the episode. Hardest/rarest job to come by. These days, thanks to transport stream capturing, translators are sometimes able to get assistance from closed captions for shows that have them. Translators have stylistic preferences, such as liberal translation versus literal translation.
- Translation Checker (TLC): Another translator who checks the other translators’ work for errors, and/or gives a 2nd opinion. Sometimes they will watch the whole episode through and give remarks, sometimes they will only fill in lines the main translator wasn’t able to figure out.
- Editing (Edit): Responsible for fixing up the translated script. Like how there are liberal versus literal translators, there are some liberal and literal editors too. Universally, editors handle things like spelling, grammar, rewording, and sense-making.
- Timing (Time): An entry level grunt job. Requires some common sense (a.k.a. ability to differentiate between different voices) and a computer that can handle loaded audio. Timers load the audio from a video and a text script into a subtitle program. After, they time the lines of the script to the audio and match up voices where they start/end with some precision. Afterwards comes fine timing. Fine timing (or “Postprocessing”, so named after the AegiSub feature) refers to a theory of subtitle readability that doesn’t really belong in a basic FAQ. :D An example of this kind of timing is adjusting line times to fix lines where the line is on-screen past a scene change even though no voice is heard.
- Karaoke Timing (k Timing): Timing the syllables in a song. Includes timing the Japanese characters. Need some knowledge of or a little guidance on to know how to split Japanese syllables and how to match up the syllables with the characters.
- Encoding: Responsible for filtering and producing a video file. Also responsible for creating a lower quality “workraw” for other staff members to work with. Preferably, the final video encode should not look like shit. Things to avoid as an encoder are an abundant use of WarpSharp and Deen, unless you want to be laughed at terribly. (Oh, and as ToraDora has proven, people are more prone to notice grain than they are prone to notice blocking and banding. :p)
- Typesetting (TS): The magical art of placing translated sign text near the original text so that people know what it says. Fonts, colors, etc, are picked. There are 2 types of typesetting: .ass typesetting and AFX typesetting. .ass typesetting uses what’s available with the .ass format. AFX refers to Adobe AfterEffects. AFX typesetting = More e-penis. Sometimes AFX ts involves repainting over signs, or recreating effects for logos, or recreating effects for complex signs.
- Karaoke Effects (kFX): Animated typesetting for songs, timed to the song karaoke timing. Some people write their own programs for this. It can be done in AegiSub using the Automation feature. Like Typesetting, it can also be done in AFX. These days, it’s extremely pointless and is hardly a required element for a release. Truthfully, it can look good, but a lot of the time it’s absolutely intrusive bullshit. If you’re going to make hardsubbed karaoke effects, please don’t make them look like absolute shit. Thanks! :D
- Styling: Picking font/colors for speech, and applying them to character speech throughout the episode. In example, if you have an Italics style used for character thought, the styling person should be responsible for setting them every episode. Sometimes this job is lumped together with Typesetting, which is why I bunched it under here.
- Quality Checking (QC): A Quality Checker should preferably know a bit about timing and editing. They watch the episode through and find errors, such as typos or an audible line not being timed, and make suggestions on things to improve.
- Raw Providing: Downloading a raw from a source such as a P2P program like Winny or Perfect Dark for the staff to work with. Transport stream capping in Japan has become preferred over P2P video sources. This isn’t really worth being called its own profession.
- Distro: Refers to IRC bots and BitTorrent seeding. Distro people get files in advance to help seed and serve for releases. IRC bots run off of scripts such as dinoex.
Sample workflow
- Airing: Tales of the Abyss airs on MBS. ilifin and koda watch it. ilifin tries to make speech markers while watching so he doesn’t miss any lines later. Afterwards, we get the first raw available, which would be Horrible Raws, and begins translating from that.
- Workraw: `Grain gets the transport stream cap after TotA finishes airing on CBC. (This is an hour after it finishes airing on MBS.) After downloading it to his computer, he makes a workraw for timing, typesetting, and to have better audio than Horrible Raws.
- Premux: Since we don’t do any hardsubs, `Grain can immediately make the final encode after he finishes off the workraw for us.
- Editing and Timing: ilifin uploads the translation. If the workraw is uploaded, then el will time first. If the workraw isn’t uploaded, then koda will edit the text file using the Horrible Raw as a raw source.
- Typesetting and Song Shifting: After the workraw is uploaded, koda typesets the signs and loads the audio to shift the subtitles for the opening and ending songs.
- Quality Check: SDX and twilight watch script + video and point out things that are not okay.
- Release: The script gets muxed with the uploaded final video. After, the file gets put through a BitTorrent seeding chain, and is distributed to multiple bots before being submitted to TokyoTosho.

thanks for the hard work gg