Fansub - History - Modern Fansub Techniques

Modern Fansub Techniques

Modern fansubs are produced entirely on computers. A raw is still required, but unlike the fansubbers who relied on laser discs, most raw sources comes directly from recordings off Japanese TV, which are widely available via Japanese peer-to-peer programs such as Winny, Share, or Perfect Dark. Some larger fansubbing groups have cappers in Japan that supply them with an MPEG transport stream. While TV recordings are now the primary type of raw used today, rips of region 2 DVDs are also used. For older shows not available on DVD, some modern fansubbers use computers equipped with video capture hardware to get digital copies of older analog media (laserdisc or tape) to work with.

Once the video is in the computer it can be edited and subtitles applied with minimal or no loss of quality, compared to the playback-recording cycle required in traditional fansubbing. However, a majority of the encoding formats used generally cause some loss of quality versus the original broadcast or DVD. A relatively inexpensive PC can perform all of the manipulation necessary, without the need for expensive and complex devices such as editing decks and a genlock.

Translation is usually done solely by listening to the recording. Mostly, translators are not experienced with fansub technology and only provide a translation. While commercial releases will often have access to the scripts, fansubbers have to translate by ear. This can sometimes lead to mistakes or unclear spellings of names. The latter is most common with shows that use Western names. Because of ambiguities resulting from Japanese pronunciation and transcription of English names, names like Alice can sound or be spelled like "Arisu" - which can be misheard as any number of Alice alternatives. This can lead to different fansubbing groups using different spellings. A famous example is Winry Rockbell from Fullmetal Alchemist, who was variously spelled as Winry, Winly and Rinry by different groups due to the equivalence of the alveolar approximant and alveolar lateral approximant in Japanese. Many groups have translation checkers to reduce the chances of letting translation errors slip through, and/or to give an alternative wording/meaning of a certain line to aid in editing an ambiguous translation. Translations for most shows are between 200 and 300 lines, though some dialogue-heavy shows may reach over 500 lines.

One alternative to using the raw Japanese file for audio translation is the use of video that has been subtitled in Chinese. China, Hong Kong and Taiwan have their own fansub groups that also release to the Internet. Several fansubbers are known to translate into English from the Chinese translations of the original Japanese, although this inherently reduces the accuracy of the translation because of the fact it has gone through two translations. To account for this, fansub groups using Chinese subs often have one or more Japanese translation checkers to minimize the loss of original meaning. A recent example of a show that was fansubbed entirely using Chinese subs is My-Otome; Doremi, one of the groups that worked on the show, used two native Chinese speakers for the project, although several translation checkers were on hand to verify against the original Japanese. In a similar way, English-subbed series can be retranslated into other languages, notably Russian.

Another, more recent, alternative with the growing availability and usage of .ts raws is translation from Japanese closed captions. The closed captions can be exported from the .ts raw into various formats, and most fansub groups use a program called C-Cats to accomplish it. This method often results in a fast, yet still fairly accurate translation due to greater ease of translating text to text, rather than audio to text. This method, however, is not as widespread, as it is still not commonplace to have a .ts raw for a show. In addition, not all .ts raws have the closed captions in them, as some raw providers remove the captions, and some Japanese broadcasting stations do not broadcast with closed captions. Groups that use closed captions from a .ts raw use the audio to verify the closed caption translation, as it cannot be guaranteed that the closed captions are flawless.

Timing can take place before or after translation, and currently Aegisub is the most popular program for this process. Many groups will "pre-time" before the translation is done, then upon completion of the translation, apply the translation to the timed lines, while at the same time doing what is called "fine timing." Fine timing often involves applying "scene timing," which is a process whereby a line's start or end point is made to correspond with a nearby scene change. This prevents "scene bleeds," which occur when every line has the same lead-in or lead-out time, resulting in some lines starting before or after a scene change.

The next process is to typeset both the text or other parts of the video which have been translated (signs, cellphone screens, etc.). Many groups make viewing easier and more organized by utilizing different colors and/or styles for different conditions that the current line is under. In this way, viewers can differentiate between, for example, speech by an on-screen character, speech by an off-screen character, thoughts, announcements (e.g. train boarding notices), or any other conditions which may require differentiation. Many groups use AFX, which is the process of typesetting signs or other on-screen text onto the video such that they blend in seamlessly with or on top of the original Japanese ones. Due to the limitations of softsubs, AFX is usually encoded directly into the video. Many groups who either do not have skilled typesetters or are attempting to release as fast as possible will often just put up another subtitle line (usually at the top of the screen) with the translation of the on-screen text (e.g. "Sign: John's Pub").

Editing takes place any time after the translation has been completed. Most translators are more proficient in Japanese than they are in English, and as such their translations are often ambiguous or grammatically incorrect. It is the editor's job to make the subtitles as easily understandable to a native English speaker as the Japanese audio would be to a native Japanese speaker, while still retaining as much of the original meaning as possible. Different groups have different guidelines for editing. Some insist upon keeping as literal subtitles as possible, thus the editor would merely fix spelling and grammar mistakes, while other groups are more liberal with their editing, in which case the editor often rewrites/rewords lines in their entirety. Many groups have the translator or translation checker view the episode with the edited subtitles to ensure that the editor has not accidentally changed the meaning of a line. Fansub editors on the whole do not require high-level English education, as the dialogue lines are of course not extremely complex.

Quality control, or QC, is one of the final stages of fansubbing. Many groups do what is called a "soft QC", then encode the episode, then do what is called a "hard QC." The goal of quality checking an episode is to catch any typesetting, timing, editing, and, in the case of hard QC, encoding errors. Most groups have multiple QCers, each of whom compiles a report of errors in the episode and submits it, and any errors are then fixed. Quality checkers often are capable of doing other fansub jobs, or have some overall knowledge of the fansubbing process, as well as an eye for spotting various errors.

The subtitles are then encoded using VirtualDub or a similar program. There are several methods of subbing currently used. "Hard" subtitles, or hard subs, are encoded into the footage, and thus become hard to remove from the video without losing video quality (this can be done with a VirtualDub Filter). "Soft" subtitles, or soft subs, are subtitles applied at playback time from a subtitle datafile, either muxed directly into the video file (.mkv, .ogm, etc.), or in a separate file (.ssa, .srt, etc.). With the correct media player or an auxiliary program, softsubs are superimposed on the footage and appear indistinguishable from hardsubs. Soft subs can also be rendered at higher resolutions, which can make for easier reading if the viewer is upscaling the file. Hard subs have traditionally been more popular than softsubs, due to a lack of player support and worries over plagiarism, but most fansub groups now release a softsub version of their releases. Since modern video media can contain multiple softsubs, some groups release fansubs with several translations into different languages, or differently styled subtitles to fit different preferences. Some groups have begun to release the opening and ending animations as separate files in order to reduce the size of each individual episode, though this introduces conflicts with player support, thus this method is not yet widespread.

In the case of hard subtitles a video editor (commonly VirtualDub) uses an AVISynth script to load the raw video file and the subtitle file (created by the translators) then the video software applies the subtitles on the video and captures video with the subtitles "burned" in.

The resulting fansub is a computer video file. In the case of soft subs, the companion sub data can be supplied as a separate file; however the complete package often now comes in a suitable media container such as Matroska. It can be copied to CD or DVD media for physical distribution, but is most often distributed using online file-sharing protocols such as viral video, DDL, BitTorrent and by file-sharing bots on IRC. This distribution is usually handled by a distribution team, or "distro" team, composed of one or more individuals with a server or very high upload speed. This allows modern anime fans to download the finished product at little or no cost to themselves or to distributors, as the distro team usually uses servers that are not dedicated to fansub releases, or that are paid for through donations to their respective fansub group.

The internet allows for highly collaborative fansubbing, and each member of a fansub team may only complete one task. Online fansubbing communities are able to release a fully subtitled episode (including elaborate karaoke with translation, kana, and kanji for songs, as well as additional remarks and translations of signs) within 24 hours of an episode's debut in Japan. While this kind of speed is possible, the groups that favor speed in determent of quality are known as "speedsub" groups and tend to release low-quality fansubs (in terms of subtitle accuracy, video quality, and other aspects). "Quality" groups often take several days, weeks, or even months to release each episode after its initial airing. However, with the advent of new techniques and technology, such as softsubs and modern hardware capable of encoding high quality video quickly, combined with larger fansub groups tending to have a large staff capable of performing tasks in parallel, the line between speedsubs and quality subs is gradually becoming blurred.